diff --git a/llvm/lib/Target/X86/X86TargetTransformInfo.cpp b/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
index ca54e1b86b5b9..c5440c8b8c64f 100644
--- a/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
+++ b/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
@@ -2698,20 +2698,28 @@ InstructionCost X86TTIImpl::getCmpSelInstrCost(unsigned Opcode, Type *ValTy,
 
     { ISD::SELECT,  MVT::v8i64,   1 },
     { ISD::SELECT,  MVT::v4i64,   1 },
+    { ISD::SELECT,  MVT::v2i64,   1 },
     { ISD::SELECT,  MVT::v16i32,  1 },
     { ISD::SELECT,  MVT::v8i32,   1 },
+    { ISD::SELECT,  MVT::v4i32,   1 },
     { ISD::SELECT,  MVT::v8f64,   1 },
     { ISD::SELECT,  MVT::v4f64,   1 },
+    { ISD::SELECT,  MVT::v2f64,   1 },
+    { ISD::SELECT,  MVT::f64,     1 },
     { ISD::SELECT,  MVT::v16f32,  1 },
     { ISD::SELECT,  MVT::v8f32 ,  1 },
+    { ISD::SELECT,  MVT::v4f32,   1 },
+    { ISD::SELECT,  MVT::f32  ,   1 },
 
     { ISD::SETCC,   MVT::v32i16,  2 }, // FIXME: should probably be 4
     { ISD::SETCC,   MVT::v64i8,   2 }, // FIXME: should probably be 4
 
     { ISD::SELECT,  MVT::v32i16,  2 },
     { ISD::SELECT,  MVT::v16i16,  1 },
+    { ISD::SELECT,  MVT::v8i16,   1 },
     { ISD::SELECT,  MVT::v64i8,   2 },
     { ISD::SELECT,  MVT::v32i8,   1 },
+    { ISD::SELECT,  MVT::v16i8,   1 },
   };
 
   static const CostTblEntry AVX2CostTbl[] = {
@@ -2752,14 +2760,14 @@ InstructionCost X86TTIImpl::getCmpSelInstrCost(unsigned Opcode, Type *ValTy,
   };
 
   static const CostTblEntry SSE41CostTbl[] = {
-    { ISD::SELECT,  MVT::v2f64,   1 }, // blendvpd
-    { ISD::SELECT,  MVT::f64,     1 }, // blendvpd
-    { ISD::SELECT,  MVT::v4f32,   1 }, // blendvps
-    { ISD::SELECT,  MVT::f32  ,   1 }, // blendvps
-    { ISD::SELECT,  MVT::v2i64,   1 }, // pblendvb
-    { ISD::SELECT,  MVT::v4i32,   1 }, // pblendvb
-    { ISD::SELECT,  MVT::v8i16,   1 }, // pblendvb
-    { ISD::SELECT,  MVT::v16i8,   1 }, // pblendvb
+    { ISD::SELECT,  MVT::v2f64,   2 }, // blendvpd
+    { ISD::SELECT,  MVT::f64,     2 }, // blendvpd
+    { ISD::SELECT,  MVT::v4f32,   2 }, // blendvps
+    { ISD::SELECT,  MVT::f32  ,   2 }, // blendvps
+    { ISD::SELECT,  MVT::v2i64,   2 }, // pblendvb
+    { ISD::SELECT,  MVT::v4i32,   2 }, // pblendvb
+    { ISD::SELECT,  MVT::v8i16,   2 }, // pblendvb
+    { ISD::SELECT,  MVT::v16i8,   2 }, // pblendvb
   };
 
   static const CostTblEntry SSE2CostTbl[] = {
diff --git a/llvm/test/Analysis/CostModel/X86/arith-sminmax.ll b/llvm/test/Analysis/CostModel/X86/arith-sminmax.ll
index f98256bf24eb6..4322ea029d43f 100644
--- a/llvm/test/Analysis/CostModel/X86/arith-sminmax.ll
+++ b/llvm/test/Analysis/CostModel/X86/arith-sminmax.ll
@@ -69,9 +69,9 @@ define i32 @smax(i32 %arg) {
 ;
 ; SSE42-LABEL: 'smax'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.smax.i64(i64 undef, i64 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.smax.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I64 = call <4 x i64> @llvm.smax.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I64 = call <8 x i64> @llvm.smax.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I64 = call <2 x i64> @llvm.smax.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.smax.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I64 = call <8 x i64> @llvm.smax.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.smax.i32(i32 undef, i32 undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.smax.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.smax.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -88,7 +88,7 @@ define i32 @smax(i32 %arg) {
 ;
 ; AVX1-LABEL: 'smax'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.smax.i64(i64 undef, i64 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.smax.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I64 = call <2 x i64> @llvm.smax.v2i64(<2 x i64> undef, <2 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I64 = call <4 x i64> @llvm.smax.v4i64(<4 x i64> undef, <4 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8I64 = call <8 x i64> @llvm.smax.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.smax.i32(i32 undef, i32 undef)
@@ -107,7 +107,7 @@ define i32 @smax(i32 %arg) {
 ;
 ; AVX2-LABEL: 'smax'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.smax.i64(i64 undef, i64 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.smax.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I64 = call <2 x i64> @llvm.smax.v2i64(<2 x i64> undef, <2 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4I64 = call <4 x i64> @llvm.smax.v4i64(<4 x i64> undef, <4 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I64 = call <8 x i64> @llvm.smax.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.smax.i32(i32 undef, i32 undef)
@@ -265,9 +265,9 @@ define i32 @smin(i32 %arg) {
 ;
 ; SSE42-LABEL: 'smin'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.smin.i64(i64 undef, i64 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.smin.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I64 = call <4 x i64> @llvm.smin.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I64 = call <8 x i64> @llvm.smin.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I64 = call <2 x i64> @llvm.smin.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.smin.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I64 = call <8 x i64> @llvm.smin.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.smin.i32(i32 undef, i32 undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.smin.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.smin.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -284,7 +284,7 @@ define i32 @smin(i32 %arg) {
 ;
 ; AVX1-LABEL: 'smin'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.smin.i64(i64 undef, i64 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.smin.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I64 = call <2 x i64> @llvm.smin.v2i64(<2 x i64> undef, <2 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I64 = call <4 x i64> @llvm.smin.v4i64(<4 x i64> undef, <4 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8I64 = call <8 x i64> @llvm.smin.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.smin.i32(i32 undef, i32 undef)
@@ -303,7 +303,7 @@ define i32 @smin(i32 %arg) {
 ;
 ; AVX2-LABEL: 'smin'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.smin.i64(i64 undef, i64 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = call <2 x i64> @llvm.smin.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V2I64 = call <2 x i64> @llvm.smin.v2i64(<2 x i64> undef, <2 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4I64 = call <4 x i64> @llvm.smin.v4i64(<4 x i64> undef, <4 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I64 = call <8 x i64> @llvm.smin.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.smin.i32(i32 undef, i32 undef)
diff --git a/llvm/test/Analysis/CostModel/X86/arith-ssat.ll b/llvm/test/Analysis/CostModel/X86/arith-ssat.ll
index b3c12fd6e0490..d7b5c32f4a9b5 100644
--- a/llvm/test/Analysis/CostModel/X86/arith-ssat.ll
+++ b/llvm/test/Analysis/CostModel/X86/arith-ssat.ll
@@ -56,13 +56,13 @@ define i32 @add(i32 %arg) {
 ;
 ; SSE42-LABEL: 'add'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.sadd.sat.i64(i64 undef, i64 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.sadd.sat.i32(i32 undef, i32 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.sadd.sat.i16(i16 undef, i16 undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -75,11 +75,11 @@ define i32 @add(i32 %arg) {
 ;
 ; AVX1-LABEL: 'add'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.sadd.sat.i64(i64 undef, i64 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.sadd.sat.i32(i32 undef, i32 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.sadd.sat.i16(i16 undef, i16 undef)
@@ -94,11 +94,11 @@ define i32 @add(i32 %arg) {
 ;
 ; AVX2-LABEL: 'add'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.sadd.sat.i64(i64 undef, i64 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.sadd.sat.i32(i32 undef, i32 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.sadd.sat.i16(i16 undef, i16 undef)
@@ -170,13 +170,13 @@ define i32 @add(i32 %arg) {
 ;
 ; SLM-LABEL: 'add'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.sadd.sat.i64(i64 undef, i64 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.sadd.sat.i32(i32 undef, i32 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.sadd.sat.i16(i16 undef, i16 undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -189,13 +189,13 @@ define i32 @add(i32 %arg) {
 ;
 ; GLM-LABEL: 'add'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.sadd.sat.i64(i64 undef, i64 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.sadd.sat.i32(i32 undef, i32 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.sadd.sat.i16(i16 undef, i16 undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -208,11 +208,11 @@ define i32 @add(i32 %arg) {
 ;
 ; BTVER2-LABEL: 'add'
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.sadd.sat.i64(i64 undef, i64 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.sadd.sat.i32(i32 undef, i32 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.sadd.sat.i16(i16 undef, i16 undef)
@@ -290,13 +290,13 @@ define i32 @sub(i32 %arg) {
 ;
 ; SSE42-LABEL: 'sub'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.ssub.sat.i64(i64 undef, i64 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.ssub.sat.i32(i32 undef, i32 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.ssub.sat.i16(i16 undef, i16 undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -309,11 +309,11 @@ define i32 @sub(i32 %arg) {
 ;
 ; AVX1-LABEL: 'sub'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.ssub.sat.i64(i64 undef, i64 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.ssub.sat.i32(i32 undef, i32 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.ssub.sat.i16(i16 undef, i16 undef)
@@ -328,11 +328,11 @@ define i32 @sub(i32 %arg) {
 ;
 ; AVX2-LABEL: 'sub'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.ssub.sat.i64(i64 undef, i64 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.ssub.sat.i32(i32 undef, i32 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.ssub.sat.i16(i16 undef, i16 undef)
@@ -404,13 +404,13 @@ define i32 @sub(i32 %arg) {
 ;
 ; SLM-LABEL: 'sub'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.ssub.sat.i64(i64 undef, i64 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 49 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.ssub.sat.i32(i32 undef, i32 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.ssub.sat.i16(i16 undef, i16 undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -423,13 +423,13 @@ define i32 @sub(i32 %arg) {
 ;
 ; GLM-LABEL: 'sub'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.ssub.sat.i64(i64 undef, i64 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.ssub.sat.i32(i32 undef, i32 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.ssub.sat.i16(i16 undef, i16 undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = call <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
@@ -442,11 +442,11 @@ define i32 @sub(i32 %arg) {
 ;
 ; BTVER2-LABEL: 'sub'
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.ssub.sat.i64(i64 undef, i64 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.ssub.sat.i32(i32 undef, i32 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.ssub.sat.i16(i16 undef, i16 undef)
diff --git a/llvm/test/Analysis/CostModel/X86/arith-uminmax.ll b/llvm/test/Analysis/CostModel/X86/arith-uminmax.ll
index fe5e7a38b446d..4a12c3b892dd8 100644
--- a/llvm/test/Analysis/CostModel/X86/arith-uminmax.ll
+++ b/llvm/test/Analysis/CostModel/X86/arith-uminmax.ll
@@ -69,9 +69,9 @@ define i32 @umax(i32 %arg) {
 ;
 ; SSE42-LABEL: 'umax'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.umax.i64(i64 undef, i64 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.umax.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.umax.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.umax.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.umax.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.umax.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call <8 x i64> @llvm.umax.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.umax.i32(i32 undef, i32 undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.umax.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.umax.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -88,7 +88,7 @@ define i32 @umax(i32 %arg) {
 ;
 ; AVX1-LABEL: 'umax'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.umax.i64(i64 undef, i64 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.umax.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.umax.v2i64(<2 x i64> undef, <2 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.umax.v4i64(<4 x i64> undef, <4 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I64 = call <8 x i64> @llvm.umax.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.umax.i32(i32 undef, i32 undef)
@@ -107,7 +107,7 @@ define i32 @umax(i32 %arg) {
 ;
 ; AVX2-LABEL: 'umax'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.umax.i64(i64 undef, i64 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.umax.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.umax.v2i64(<2 x i64> undef, <2 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I64 = call <4 x i64> @llvm.umax.v4i64(<4 x i64> undef, <4 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I64 = call <8 x i64> @llvm.umax.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.umax.i32(i32 undef, i32 undef)
@@ -265,9 +265,9 @@ define i32 @umin(i32 %arg) {
 ;
 ; SSE42-LABEL: 'umin'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.umin.i64(i64 undef, i64 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.umin.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.umin.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.umin.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.umin.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.umin.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call <8 x i64> @llvm.umin.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.umin.i32(i32 undef, i32 undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = call <4 x i32> @llvm.umin.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = call <8 x i32> @llvm.umin.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -284,7 +284,7 @@ define i32 @umin(i32 %arg) {
 ;
 ; AVX1-LABEL: 'umin'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.umin.i64(i64 undef, i64 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.umin.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.umin.v2i64(<2 x i64> undef, <2 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I64 = call <4 x i64> @llvm.umin.v4i64(<4 x i64> undef, <4 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I64 = call <8 x i64> @llvm.umin.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.umin.i32(i32 undef, i32 undef)
@@ -303,7 +303,7 @@ define i32 @umin(i32 %arg) {
 ;
 ; AVX2-LABEL: 'umin'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.umin.i64(i64 undef, i64 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V2I64 = call <2 x i64> @llvm.umin.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.umin.v2i64(<2 x i64> undef, <2 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V4I64 = call <4 x i64> @llvm.umin.v4i64(<4 x i64> undef, <4 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I64 = call <8 x i64> @llvm.umin.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.umin.i32(i32 undef, i32 undef)
diff --git a/llvm/test/Analysis/CostModel/X86/arith-usat.ll b/llvm/test/Analysis/CostModel/X86/arith-usat.ll
index 8fe4e24cdf69b..0966082a6575c 100644
--- a/llvm/test/Analysis/CostModel/X86/arith-usat.ll
+++ b/llvm/test/Analysis/CostModel/X86/arith-usat.ll
@@ -57,9 +57,9 @@ define i32 @add(i32 %arg) {
 ;
 ; SSE42-LABEL: 'add'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.uadd.sat.i64(i64 undef, i64 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.uadd.sat.i32(i32 undef, i32 undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4I32 = call <4 x i32> @llvm.uadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I32 = call <8 x i32> @llvm.uadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -76,7 +76,7 @@ define i32 @add(i32 %arg) {
 ;
 ; AVX1-LABEL: 'add'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.uadd.sat.i64(i64 undef, i64 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.uadd.sat.i32(i32 undef, i32 undef)
@@ -95,7 +95,7 @@ define i32 @add(i32 %arg) {
 ;
 ; AVX2-LABEL: 'add'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.uadd.sat.i64(i64 undef, i64 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.uadd.sat.i32(i32 undef, i32 undef)
@@ -171,9 +171,9 @@ define i32 @add(i32 %arg) {
 ;
 ; SLM-LABEL: 'add'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.uadd.sat.i64(i64 undef, i64 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.uadd.sat.i32(i32 undef, i32 undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4I32 = call <4 x i32> @llvm.uadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I32 = call <8 x i32> @llvm.uadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -190,9 +190,9 @@ define i32 @add(i32 %arg) {
 ;
 ; GLM-LABEL: 'add'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.uadd.sat.i64(i64 undef, i64 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.uadd.sat.i32(i32 undef, i32 undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4I32 = call <4 x i32> @llvm.uadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I32 = call <8 x i32> @llvm.uadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -209,7 +209,7 @@ define i32 @add(i32 %arg) {
 ;
 ; BTVER2-LABEL: 'add'
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.uadd.sat.i64(i64 undef, i64 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.uadd.sat.i32(i32 undef, i32 undef)
@@ -291,9 +291,9 @@ define i32 @sub(i32 %arg) {
 ;
 ; SSE42-LABEL: 'sub'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -310,7 +310,7 @@ define i32 @sub(i32 %arg) {
 ;
 ; AVX1-LABEL: 'sub'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef)
@@ -329,7 +329,7 @@ define i32 @sub(i32 %arg) {
 ;
 ; AVX2-LABEL: 'sub'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef)
@@ -405,9 +405,9 @@ define i32 @sub(i32 %arg) {
 ;
 ; SLM-LABEL: 'sub'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -424,9 +424,9 @@ define i32 @sub(i32 %arg) {
 ;
 ; GLM-LABEL: 'sub'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
@@ -443,7 +443,7 @@ define i32 @sub(i32 %arg) {
 ;
 ; BTVER2-LABEL: 'sub'
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef)
-; BTVER2-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; BTVER2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
 ; BTVER2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef)
diff --git a/llvm/test/Analysis/CostModel/X86/fptoi_sat.ll b/llvm/test/Analysis/CostModel/X86/fptoi_sat.ll
index d530bdd4a4b43..2ed34d07f11f2 100644
--- a/llvm/test/Analysis/CostModel/X86/fptoi_sat.ll
+++ b/llvm/test/Analysis/CostModel/X86/fptoi_sat.ll
@@ -135,85 +135,85 @@ define void @casts() {
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %f64u32 = call i32 @llvm.fptoui.sat.i32.f64(double undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %f64s64 = call i64 @llvm.fptosi.sat.i64.f64(double undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %f64u64 = call i64 @llvm.fptoui.sat.i64.f64(double undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v2f32s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f32(<2 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %v2f32s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f32(<2 x float> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v2f32u1 = call <2 x i1> @llvm.fptoui.sat.v2i1.v2f32(<2 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v2f32s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f32(<2 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %v2f32s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f32(<2 x float> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v2f32u8 = call <2 x i8> @llvm.fptoui.sat.v2i8.v2f32(<2 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v2f32s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f32(<2 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v2f32s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f32(<2 x float> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v2f32u16 = call <2 x i16> @llvm.fptoui.sat.v2i16.v2f32(<2 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v2f32s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f32(<2 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v2f32s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f32(<2 x float> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v2f32u32 = call <2 x i32> @llvm.fptoui.sat.v2i32.v2f32(<2 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v2f32s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f32(<2 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %v2f32s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f32(<2 x float> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v2f32u64 = call <2 x i64> @llvm.fptoui.sat.v2i64.v2f32(<2 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v2f64s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f64(<2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %v2f64s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f64(<2 x double> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v2f64u1 = call <2 x i1> @llvm.fptoui.sat.v2i1.v2f64(<2 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v2f64s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f64(<2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %v2f64s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f64(<2 x double> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v2f64u8 = call <2 x i8> @llvm.fptoui.sat.v2i8.v2f64(<2 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v2f64s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f64(<2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v2f64s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f64(<2 x double> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v2f64u16 = call <2 x i16> @llvm.fptoui.sat.v2i16.v2f64(<2 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v2f64s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f64(<2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v2f64s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f64(<2 x double> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v2f64u32 = call <2 x i32> @llvm.fptoui.sat.v2i32.v2f64(<2 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v2f64s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f64(<2 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %v2f64s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f64(<2 x double> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v2f64u64 = call <2 x i64> @llvm.fptoui.sat.v2i64.v2f64(<2 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v4f32s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f32(<4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v4f32s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f32(<4 x float> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v4f32u1 = call <4 x i1> @llvm.fptoui.sat.v4i1.v4f32(<4 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v4f32s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f32(<4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %v4f32s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f32(<4 x float> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v4f32u8 = call <4 x i8> @llvm.fptoui.sat.v4i8.v4f32(<4 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v4f32s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f32(<4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v4f32s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f32(<4 x float> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v4f32u16 = call <4 x i16> @llvm.fptoui.sat.v4i16.v4f32(<4 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v4f32s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f32(<4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v4f32s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f32(<4 x float> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v4f32u32 = call <4 x i32> @llvm.fptoui.sat.v4i32.v4f32(<4 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v4f32s64 = call <4 x i64> @llvm.fptosi.sat.v4i64.v4f32(<4 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %v4f32s64 = call <4 x i64> @llvm.fptosi.sat.v4i64.v4f32(<4 x float> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %v4f32u64 = call <4 x i64> @llvm.fptoui.sat.v4i64.v4f32(<4 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %v4f64s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f64(<4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %v4f64s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f64(<4 x double> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v4f64u1 = call <4 x i1> @llvm.fptoui.sat.v4i1.v4f64(<4 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %v4f64s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f64(<4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v4f64s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f64(<4 x double> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v4f64u8 = call <4 x i8> @llvm.fptoui.sat.v4i8.v4f64(<4 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %v4f64s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f64(<4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %v4f64s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f64(<4 x double> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %v4f64u16 = call <4 x i16> @llvm.fptoui.sat.v4i16.v4f64(<4 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %v4f64s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f64(<4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %v4f64s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f64(<4 x double> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v4f64u32 = call <4 x i32> @llvm.fptoui.sat.v4i32.v4f64(<4 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %v4f64s64 = call <4 x i64> @llvm.fptosi.sat.v4i64.v4f64(<4 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %v4f64s64 = call <4 x i64> @llvm.fptosi.sat.v4i64.v4f64(<4 x double> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %v4f64u64 = call <4 x i64> @llvm.fptoui.sat.v4i64.v4f64(<4 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %v8f32s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f32(<8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %v8f32s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f32(<8 x float> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %v8f32u1 = call <8 x i1> @llvm.fptoui.sat.v8i1.v8f32(<8 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %v8f32s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f32(<8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v8f32s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f32(<8 x float> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v8f32u8 = call <8 x i8> @llvm.fptoui.sat.v8i8.v8f32(<8 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %v8f32s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f32(<8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %v8f32s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f32(<8 x float> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %v8f32u16 = call <8 x i16> @llvm.fptoui.sat.v8i16.v8f32(<8 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %v8f32s32 = call <8 x i32> @llvm.fptosi.sat.v8i32.v8f32(<8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v8f32s32 = call <8 x i32> @llvm.fptosi.sat.v8i32.v8f32(<8 x float> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v8f32u32 = call <8 x i32> @llvm.fptoui.sat.v8i32.v8f32(<8 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %v8f32s64 = call <8 x i64> @llvm.fptosi.sat.v8i64.v8f32(<8 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %v8f32s64 = call <8 x i64> @llvm.fptosi.sat.v8i64.v8f32(<8 x float> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 66 for instruction: %v8f32u64 = call <8 x i64> @llvm.fptoui.sat.v8i64.v8f32(<8 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %v8f64s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f64(<8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %v8f64s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f64(<8 x double> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %v8f64u1 = call <8 x i1> @llvm.fptoui.sat.v8i1.v8f64(<8 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %v8f64s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f64(<8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %v8f64s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f64(<8 x double> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %v8f64u8 = call <8 x i8> @llvm.fptoui.sat.v8i8.v8f64(<8 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %v8f64s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f64(<8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %v8f64s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f64(<8 x double> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %v8f64u16 = call <8 x i16> @llvm.fptoui.sat.v8i16.v8f64(<8 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %v8f64s32 = call <8 x i32> @llvm.fptosi.sat.v8i32.v8f64(<8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %v8f64s32 = call <8 x i32> @llvm.fptosi.sat.v8i32.v8f64(<8 x double> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %v8f64u32 = call <8 x i32> @llvm.fptoui.sat.v8i32.v8f64(<8 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %v8f64s64 = call <8 x i64> @llvm.fptosi.sat.v8i64.v8f64(<8 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %v8f64s64 = call <8 x i64> @llvm.fptosi.sat.v8i64.v8f64(<8 x double> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %v8f64u64 = call <8 x i64> @llvm.fptoui.sat.v8i64.v8f64(<8 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %v16f32s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f32(<16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %v16f32s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f32(<16 x float> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %v16f32u1 = call <16 x i1> @llvm.fptoui.sat.v16i1.v16f32(<16 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %v16f32s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f32(<16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %v16f32s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f32(<16 x float> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %v16f32u8 = call <16 x i8> @llvm.fptoui.sat.v16i8.v16f32(<16 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %v16f32s16 = call <16 x i16> @llvm.fptosi.sat.v16i16.v16f32(<16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %v16f32s16 = call <16 x i16> @llvm.fptosi.sat.v16i16.v16f32(<16 x float> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %v16f32u16 = call <16 x i16> @llvm.fptoui.sat.v16i16.v16f32(<16 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %v16f32s32 = call <16 x i32> @llvm.fptosi.sat.v16i32.v16f32(<16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %v16f32s32 = call <16 x i32> @llvm.fptosi.sat.v16i32.v16f32(<16 x float> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %v16f32u32 = call <16 x i32> @llvm.fptoui.sat.v16i32.v16f32(<16 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %v16f32s64 = call <16 x i64> @llvm.fptosi.sat.v16i64.v16f32(<16 x float> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %v16f32s64 = call <16 x i64> @llvm.fptosi.sat.v16i64.v16f32(<16 x float> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %v16f32u64 = call <16 x i64> @llvm.fptoui.sat.v16i64.v16f32(<16 x float> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 89 for instruction: %v16f64s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f64(<16 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 90 for instruction: %v16f64s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f64(<16 x double> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %v16f64u1 = call <16 x i1> @llvm.fptoui.sat.v16i1.v16f64(<16 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 89 for instruction: %v16f64s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f64(<16 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 90 for instruction: %v16f64s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f64(<16 x double> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %v16f64u8 = call <16 x i8> @llvm.fptoui.sat.v16i8.v16f64(<16 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 82 for instruction: %v16f64s16 = call <16 x i16> @llvm.fptosi.sat.v16i16.v16f64(<16 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %v16f64s16 = call <16 x i16> @llvm.fptosi.sat.v16i16.v16f64(<16 x double> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %v16f64u16 = call <16 x i16> @llvm.fptoui.sat.v16i16.v16f64(<16 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %v16f64s32 = call <16 x i32> @llvm.fptosi.sat.v16i32.v16f64(<16 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %v16f64s32 = call <16 x i32> @llvm.fptosi.sat.v16i32.v16f64(<16 x double> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %v16f64u32 = call <16 x i32> @llvm.fptoui.sat.v16i32.v16f64(<16 x double> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %v16f64s64 = call <16 x i64> @llvm.fptosi.sat.v16i64.v16f64(<16 x double> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 136 for instruction: %v16f64s64 = call <16 x i64> @llvm.fptosi.sat.v16i64.v16f64(<16 x double> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %v16f64u64 = call <16 x i64> @llvm.fptoui.sat.v16i64.v16f64(<16 x double> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
@@ -238,69 +238,69 @@ define void @casts() {
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %f64u32 = call i32 @llvm.fptoui.sat.i32.f64(double undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %f64s64 = call i64 @llvm.fptosi.sat.i64.f64(double undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %f64u64 = call i64 @llvm.fptoui.sat.i64.f64(double undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v2f32s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f32(<2 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v2f32s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f32(<2 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v2f32u1 = call <2 x i1> @llvm.fptoui.sat.v2i1.v2f32(<2 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v2f32s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f32(<2 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v2f32s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f32(<2 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v2f32u8 = call <2 x i8> @llvm.fptoui.sat.v2i8.v2f32(<2 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v2f32s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f32(<2 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v2f32s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f32(<2 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %v2f32u16 = call <2 x i16> @llvm.fptoui.sat.v2i16.v2f32(<2 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v2f32s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f32(<2 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v2f32s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f32(<2 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v2f32u32 = call <2 x i32> @llvm.fptoui.sat.v2i32.v2f32(<2 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v2f32s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f32(<2 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v2f32s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f32(<2 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %v2f32u64 = call <2 x i64> @llvm.fptoui.sat.v2i64.v2f32(<2 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v2f64s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f64(<2 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v2f64s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f64(<2 x double> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v2f64u1 = call <2 x i1> @llvm.fptoui.sat.v2i1.v2f64(<2 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v2f64s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f64(<2 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v2f64s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f64(<2 x double> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v2f64u8 = call <2 x i8> @llvm.fptoui.sat.v2i8.v2f64(<2 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v2f64s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f64(<2 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v2f64s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f64(<2 x double> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %v2f64u16 = call <2 x i16> @llvm.fptoui.sat.v2i16.v2f64(<2 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v2f64s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f64(<2 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v2f64s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f64(<2 x double> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v2f64u32 = call <2 x i32> @llvm.fptoui.sat.v2i32.v2f64(<2 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v2f64s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f64(<2 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v2f64s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f64(<2 x double> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %v2f64u64 = call <2 x i64> @llvm.fptoui.sat.v2i64.v2f64(<2 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v4f32s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f32(<4 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v4f32s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f32(<4 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v4f32u1 = call <4 x i1> @llvm.fptoui.sat.v4i1.v4f32(<4 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v4f32s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f32(<4 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v4f32s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f32(<4 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v4f32u8 = call <4 x i8> @llvm.fptoui.sat.v4i8.v4f32(<4 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v4f32s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f32(<4 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v4f32s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f32(<4 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %v4f32u16 = call <4 x i16> @llvm.fptoui.sat.v4i16.v4f32(<4 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v4f32s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f32(<4 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v4f32s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f32(<4 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v4f32u32 = call <4 x i32> @llvm.fptoui.sat.v4i32.v4f32(<4 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %v4f32s64 = call <4 x i64> @llvm.fptosi.sat.v4i64.v4f32(<4 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %v4f32u64 = call <4 x i64> @llvm.fptoui.sat.v4i64.v4f32(<4 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v4f64s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f64(<4 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v4f64s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f64(<4 x double> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v4f64u1 = call <4 x i1> @llvm.fptoui.sat.v4i1.v4f64(<4 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v4f64s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f64(<4 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v4f64s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f64(<4 x double> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v4f64u8 = call <4 x i8> @llvm.fptoui.sat.v4i8.v4f64(<4 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v4f64s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f64(<4 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v4f64s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f64(<4 x double> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v4f64u16 = call <4 x i16> @llvm.fptoui.sat.v4i16.v4f64(<4 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v4f64s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f64(<4 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v4f64s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f64(<4 x double> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v4f64u32 = call <4 x i32> @llvm.fptoui.sat.v4i32.v4f64(<4 x double> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %v4f64s64 = call <4 x i64> @llvm.fptosi.sat.v4i64.v4f64(<4 x double> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %v4f64u64 = call <4 x i64> @llvm.fptoui.sat.v4i64.v4f64(<4 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v8f32s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f32(<8 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v8f32s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f32(<8 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v8f32u1 = call <8 x i1> @llvm.fptoui.sat.v8i1.v8f32(<8 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v8f32s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f32(<8 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v8f32s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f32(<8 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v8f32u8 = call <8 x i8> @llvm.fptoui.sat.v8i8.v8f32(<8 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v8f32s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f32(<8 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v8f32s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f32(<8 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v8f32u16 = call <8 x i16> @llvm.fptoui.sat.v8i16.v8f32(<8 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v8f32s32 = call <8 x i32> @llvm.fptosi.sat.v8i32.v8f32(<8 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %v8f32u32 = call <8 x i32> @llvm.fptoui.sat.v8i32.v8f32(<8 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %v8f32s64 = call <8 x i64> @llvm.fptosi.sat.v8i64.v8f32(<8 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %v8f32u64 = call <8 x i64> @llvm.fptoui.sat.v8i64.v8f32(<8 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %v8f64s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f64(<8 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %v8f64s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f64(<8 x double> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v8f64u1 = call <8 x i1> @llvm.fptoui.sat.v8i1.v8f64(<8 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %v8f64s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f64(<8 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %v8f64s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f64(<8 x double> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v8f64u8 = call <8 x i8> @llvm.fptoui.sat.v8i8.v8f64(<8 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %v8f64s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f64(<8 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %v8f64s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f64(<8 x double> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v8f64u16 = call <8 x i16> @llvm.fptoui.sat.v8i16.v8f64(<8 x double> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %v8f64s32 = call <8 x i32> @llvm.fptosi.sat.v8i32.v8f64(<8 x double> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %v8f64u32 = call <8 x i32> @llvm.fptoui.sat.v8i32.v8f64(<8 x double> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %v8f64s64 = call <8 x i64> @llvm.fptosi.sat.v8i64.v8f64(<8 x double> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %v8f64u64 = call <8 x i64> @llvm.fptoui.sat.v8i64.v8f64(<8 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %v16f32s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f32(<16 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %v16f32s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f32(<16 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v16f32u1 = call <16 x i1> @llvm.fptoui.sat.v16i1.v16f32(<16 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %v16f32s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f32(<16 x float> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %v16f32s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f32(<16 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v16f32u8 = call <16 x i8> @llvm.fptoui.sat.v16i8.v16f32(<16 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %v16f32s16 = call <16 x i16> @llvm.fptosi.sat.v16i16.v16f32(<16 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v16f32u16 = call <16 x i16> @llvm.fptoui.sat.v16i16.v16f32(<16 x float> undef)
@@ -308,9 +308,9 @@ define void @casts() {
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %v16f32u32 = call <16 x i32> @llvm.fptoui.sat.v16i32.v16f32(<16 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %v16f32s64 = call <16 x i64> @llvm.fptosi.sat.v16i64.v16f32(<16 x float> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 134 for instruction: %v16f32u64 = call <16 x i64> @llvm.fptoui.sat.v16i64.v16f32(<16 x float> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %v16f64s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f64(<16 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %v16f64s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f64(<16 x double> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %v16f64u1 = call <16 x i1> @llvm.fptoui.sat.v16i1.v16f64(<16 x double> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 53 for instruction: %v16f64s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f64(<16 x double> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %v16f64s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f64(<16 x double> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %v16f64u8 = call <16 x i8> @llvm.fptoui.sat.v16i8.v16f64(<16 x double> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %v16f64s16 = call <16 x i16> @llvm.fptosi.sat.v16i16.v16f64(<16 x double> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %v16f64u16 = call <16 x i16> @llvm.fptoui.sat.v16i16.v16f64(<16 x double> undef)
@@ -341,69 +341,69 @@ define void @casts() {
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %f64u32 = call i32 @llvm.fptoui.sat.i32.f64(double undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %f64s64 = call i64 @llvm.fptosi.sat.i64.f64(double undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %f64u64 = call i64 @llvm.fptoui.sat.i64.f64(double undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v2f32s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f32(<2 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v2f32s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f32(<2 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v2f32u1 = call <2 x i1> @llvm.fptoui.sat.v2i1.v2f32(<2 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v2f32s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f32(<2 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v2f32s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f32(<2 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v2f32u8 = call <2 x i8> @llvm.fptoui.sat.v2i8.v2f32(<2 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v2f32s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f32(<2 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v2f32s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f32(<2 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %v2f32u16 = call <2 x i16> @llvm.fptoui.sat.v2i16.v2f32(<2 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v2f32s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f32(<2 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v2f32s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f32(<2 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v2f32u32 = call <2 x i32> @llvm.fptoui.sat.v2i32.v2f32(<2 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v2f32s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f32(<2 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v2f32s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f32(<2 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v2f32u64 = call <2 x i64> @llvm.fptoui.sat.v2i64.v2f32(<2 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v2f64s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f64(<2 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v2f64s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f64(<2 x double> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v2f64u1 = call <2 x i1> @llvm.fptoui.sat.v2i1.v2f64(<2 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v2f64s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f64(<2 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v2f64s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f64(<2 x double> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v2f64u8 = call <2 x i8> @llvm.fptoui.sat.v2i8.v2f64(<2 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v2f64s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f64(<2 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v2f64s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f64(<2 x double> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %v2f64u16 = call <2 x i16> @llvm.fptoui.sat.v2i16.v2f64(<2 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v2f64s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f64(<2 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v2f64s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f64(<2 x double> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v2f64u32 = call <2 x i32> @llvm.fptoui.sat.v2i32.v2f64(<2 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v2f64s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f64(<2 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v2f64s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f64(<2 x double> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v2f64u64 = call <2 x i64> @llvm.fptoui.sat.v2i64.v2f64(<2 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v4f32s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f32(<4 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v4f32s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f32(<4 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v4f32u1 = call <4 x i1> @llvm.fptoui.sat.v4i1.v4f32(<4 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v4f32s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f32(<4 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v4f32s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f32(<4 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v4f32u8 = call <4 x i8> @llvm.fptoui.sat.v4i8.v4f32(<4 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v4f32s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f32(<4 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v4f32s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f32(<4 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %v4f32u16 = call <4 x i16> @llvm.fptoui.sat.v4i16.v4f32(<4 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v4f32s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f32(<4 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v4f32s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f32(<4 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v4f32u32 = call <4 x i32> @llvm.fptoui.sat.v4i32.v4f32(<4 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %v4f32s64 = call <4 x i64> @llvm.fptosi.sat.v4i64.v4f32(<4 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %v4f32u64 = call <4 x i64> @llvm.fptoui.sat.v4i64.v4f32(<4 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v4f64s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f64(<4 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v4f64s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f64(<4 x double> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v4f64u1 = call <4 x i1> @llvm.fptoui.sat.v4i1.v4f64(<4 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v4f64s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f64(<4 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v4f64s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f64(<4 x double> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v4f64u8 = call <4 x i8> @llvm.fptoui.sat.v4i8.v4f64(<4 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v4f64s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f64(<4 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v4f64s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f64(<4 x double> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v4f64u16 = call <4 x i16> @llvm.fptoui.sat.v4i16.v4f64(<4 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v4f64s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f64(<4 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v4f64s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f64(<4 x double> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v4f64u32 = call <4 x i32> @llvm.fptoui.sat.v4i32.v4f64(<4 x double> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %v4f64s64 = call <4 x i64> @llvm.fptosi.sat.v4i64.v4f64(<4 x double> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %v4f64u64 = call <4 x i64> @llvm.fptoui.sat.v4i64.v4f64(<4 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v8f32s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f32(<8 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v8f32s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f32(<8 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v8f32u1 = call <8 x i1> @llvm.fptoui.sat.v8i1.v8f32(<8 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v8f32s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f32(<8 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v8f32s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f32(<8 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v8f32u8 = call <8 x i8> @llvm.fptoui.sat.v8i8.v8f32(<8 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v8f32s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f32(<8 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v8f32s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f32(<8 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v8f32u16 = call <8 x i16> @llvm.fptoui.sat.v8i16.v8f32(<8 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v8f32s32 = call <8 x i32> @llvm.fptosi.sat.v8i32.v8f32(<8 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v8f32u32 = call <8 x i32> @llvm.fptoui.sat.v8i32.v8f32(<8 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %v8f32s64 = call <8 x i64> @llvm.fptosi.sat.v8i64.v8f32(<8 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 55 for instruction: %v8f32u64 = call <8 x i64> @llvm.fptoui.sat.v8i64.v8f32(<8 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %v8f64s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f64(<8 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v8f64s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f64(<8 x double> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v8f64u1 = call <8 x i1> @llvm.fptoui.sat.v8i1.v8f64(<8 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %v8f64s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f64(<8 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v8f64s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f64(<8 x double> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v8f64u8 = call <8 x i8> @llvm.fptoui.sat.v8i8.v8f64(<8 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %v8f64s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f64(<8 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v8f64s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f64(<8 x double> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v8f64u16 = call <8 x i16> @llvm.fptoui.sat.v8i16.v8f64(<8 x double> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %v8f64s32 = call <8 x i32> @llvm.fptosi.sat.v8i32.v8f64(<8 x double> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v8f64u32 = call <8 x i32> @llvm.fptoui.sat.v8i32.v8f64(<8 x double> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %v8f64s64 = call <8 x i64> @llvm.fptosi.sat.v8i64.v8f64(<8 x double> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %v8f64u64 = call <8 x i64> @llvm.fptoui.sat.v8i64.v8f64(<8 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %v16f32s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f32(<16 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v16f32s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f32(<16 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v16f32u1 = call <16 x i1> @llvm.fptoui.sat.v16i1.v16f32(<16 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %v16f32s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f32(<16 x float> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v16f32s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f32(<16 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v16f32u8 = call <16 x i8> @llvm.fptoui.sat.v16i8.v16f32(<16 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %v16f32s16 = call <16 x i16> @llvm.fptosi.sat.v16i16.v16f32(<16 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16f32u16 = call <16 x i16> @llvm.fptoui.sat.v16i16.v16f32(<16 x float> undef)
@@ -411,9 +411,9 @@ define void @casts() {
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %v16f32u32 = call <16 x i32> @llvm.fptoui.sat.v16i32.v16f32(<16 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %v16f32s64 = call <16 x i64> @llvm.fptosi.sat.v16i64.v16f32(<16 x float> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 110 for instruction: %v16f32u64 = call <16 x i64> @llvm.fptoui.sat.v16i64.v16f32(<16 x float> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %v16f64s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f64(<16 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %v16f64s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f64(<16 x double> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %v16f64u1 = call <16 x i1> @llvm.fptoui.sat.v16i1.v16f64(<16 x double> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %v16f64s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f64(<16 x double> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %v16f64s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f64(<16 x double> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %v16f64u8 = call <16 x i8> @llvm.fptoui.sat.v16i8.v16f64(<16 x double> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %v16f64s16 = call <16 x i16> @llvm.fptosi.sat.v16i16.v16f64(<16 x double> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %v16f64u16 = call <16 x i16> @llvm.fptoui.sat.v16i16.v16f64(<16 x double> undef)
@@ -650,85 +650,85 @@ define void @casts() {
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %f64u32 = call i32 @llvm.fptoui.sat.i32.f64(double undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %f64s64 = call i64 @llvm.fptosi.sat.i64.f64(double undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %f64u64 = call i64 @llvm.fptoui.sat.i64.f64(double undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v2f32s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f32(<2 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %v2f32s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f32(<2 x float> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v2f32u1 = call <2 x i1> @llvm.fptoui.sat.v2i1.v2f32(<2 x float> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v2f32s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f32(<2 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %v2f32s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f32(<2 x float> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v2f32u8 = call <2 x i8> @llvm.fptoui.sat.v2i8.v2f32(<2 x float> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v2f32s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f32(<2 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v2f32s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f32(<2 x float> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v2f32u16 = call <2 x i16> @llvm.fptoui.sat.v2i16.v2f32(<2 x float> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v2f32s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f32(<2 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v2f32s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f32(<2 x float> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v2f32u32 = call <2 x i32> @llvm.fptoui.sat.v2i32.v2f32(<2 x float> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %v2f32s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f32(<2 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %v2f32s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f32(<2 x float> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %v2f32u64 = call <2 x i64> @llvm.fptoui.sat.v2i64.v2f32(<2 x float> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v2f64s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f64(<2 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %v2f64s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f64(<2 x double> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v2f64u1 = call <2 x i1> @llvm.fptoui.sat.v2i1.v2f64(<2 x double> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v2f64s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f64(<2 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %v2f64s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f64(<2 x double> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v2f64u8 = call <2 x i8> @llvm.fptoui.sat.v2i8.v2f64(<2 x double> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v2f64s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f64(<2 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v2f64s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f64(<2 x double> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v2f64u16 = call <2 x i16> @llvm.fptoui.sat.v2i16.v2f64(<2 x double> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v2f64s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f64(<2 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v2f64s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f64(<2 x double> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v2f64u32 = call <2 x i32> @llvm.fptoui.sat.v2i32.v2f64(<2 x double> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %v2f64s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f64(<2 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %v2f64s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f64(<2 x double> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %v2f64u64 = call <2 x i64> @llvm.fptoui.sat.v2i64.v2f64(<2 x double> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v4f32s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f32(<4 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v4f32s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f32(<4 x float> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v4f32u1 = call <4 x i1> @llvm.fptoui.sat.v4i1.v4f32(<4 x float> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v4f32s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f32(<4 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %v4f32s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f32(<4 x float> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v4f32u8 = call <4 x i8> @llvm.fptoui.sat.v4i8.v4f32(<4 x float> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v4f32s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f32(<4 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v4f32s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f32(<4 x float> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v4f32u16 = call <4 x i16> @llvm.fptoui.sat.v4i16.v4f32(<4 x float> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %v4f32s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f32(<4 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v4f32s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f32(<4 x float> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v4f32u32 = call <4 x i32> @llvm.fptoui.sat.v4i32.v4f32(<4 x float> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %v4f32s64 = call <4 x i64> @llvm.fptosi.sat.v4i64.v4f32(<4 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %v4f32s64 = call <4 x i64> @llvm.fptosi.sat.v4i64.v4f32(<4 x float> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %v4f32u64 = call <4 x i64> @llvm.fptoui.sat.v4i64.v4f32(<4 x float> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %v4f64s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f64(<4 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %v4f64s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f64(<4 x double> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v4f64u1 = call <4 x i1> @llvm.fptoui.sat.v4i1.v4f64(<4 x double> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %v4f64s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f64(<4 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v4f64s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f64(<4 x double> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v4f64u8 = call <4 x i8> @llvm.fptoui.sat.v4i8.v4f64(<4 x double> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %v4f64s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f64(<4 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %v4f64s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f64(<4 x double> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %v4f64u16 = call <4 x i16> @llvm.fptoui.sat.v4i16.v4f64(<4 x double> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %v4f64s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f64(<4 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %v4f64s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f64(<4 x double> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v4f64u32 = call <4 x i32> @llvm.fptoui.sat.v4i32.v4f64(<4 x double> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %v4f64s64 = call <4 x i64> @llvm.fptosi.sat.v4i64.v4f64(<4 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %v4f64s64 = call <4 x i64> @llvm.fptosi.sat.v4i64.v4f64(<4 x double> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %v4f64u64 = call <4 x i64> @llvm.fptoui.sat.v4i64.v4f64(<4 x double> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %v8f32s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f32(<8 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %v8f32s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f32(<8 x float> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %v8f32u1 = call <8 x i1> @llvm.fptoui.sat.v8i1.v8f32(<8 x float> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %v8f32s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f32(<8 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v8f32s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f32(<8 x float> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v8f32u8 = call <8 x i8> @llvm.fptoui.sat.v8i8.v8f32(<8 x float> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %v8f32s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f32(<8 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %v8f32s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f32(<8 x float> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %v8f32u16 = call <8 x i16> @llvm.fptoui.sat.v8i16.v8f32(<8 x float> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %v8f32s32 = call <8 x i32> @llvm.fptosi.sat.v8i32.v8f32(<8 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v8f32s32 = call <8 x i32> @llvm.fptosi.sat.v8i32.v8f32(<8 x float> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v8f32u32 = call <8 x i32> @llvm.fptoui.sat.v8i32.v8f32(<8 x float> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %v8f32s64 = call <8 x i64> @llvm.fptosi.sat.v8i64.v8f32(<8 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %v8f32s64 = call <8 x i64> @llvm.fptosi.sat.v8i64.v8f32(<8 x float> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 90 for instruction: %v8f32u64 = call <8 x i64> @llvm.fptoui.sat.v8i64.v8f32(<8 x float> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %v8f64s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f64(<8 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %v8f64s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f64(<8 x double> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %v8f64u1 = call <8 x i1> @llvm.fptoui.sat.v8i1.v8f64(<8 x double> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %v8f64s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f64(<8 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %v8f64s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f64(<8 x double> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %v8f64u8 = call <8 x i8> @llvm.fptoui.sat.v8i8.v8f64(<8 x double> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %v8f64s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f64(<8 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %v8f64s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f64(<8 x double> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %v8f64u16 = call <8 x i16> @llvm.fptoui.sat.v8i16.v8f64(<8 x double> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %v8f64s32 = call <8 x i32> @llvm.fptosi.sat.v8i32.v8f64(<8 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %v8f64s32 = call <8 x i32> @llvm.fptosi.sat.v8i32.v8f64(<8 x double> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %v8f64u32 = call <8 x i32> @llvm.fptoui.sat.v8i32.v8f64(<8 x double> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %v8f64s64 = call <8 x i64> @llvm.fptosi.sat.v8i64.v8f64(<8 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %v8f64s64 = call <8 x i64> @llvm.fptosi.sat.v8i64.v8f64(<8 x double> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %v8f64u64 = call <8 x i64> @llvm.fptoui.sat.v8i64.v8f64(<8 x double> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %v16f32s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f32(<16 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %v16f32s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f32(<16 x float> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %v16f32u1 = call <16 x i1> @llvm.fptoui.sat.v16i1.v16f32(<16 x float> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %v16f32s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f32(<16 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %v16f32s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f32(<16 x float> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %v16f32u8 = call <16 x i8> @llvm.fptoui.sat.v16i8.v16f32(<16 x float> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %v16f32s16 = call <16 x i16> @llvm.fptosi.sat.v16i16.v16f32(<16 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %v16f32s16 = call <16 x i16> @llvm.fptosi.sat.v16i16.v16f32(<16 x float> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %v16f32u16 = call <16 x i16> @llvm.fptoui.sat.v16i16.v16f32(<16 x float> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %v16f32s32 = call <16 x i32> @llvm.fptosi.sat.v16i32.v16f32(<16 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %v16f32s32 = call <16 x i32> @llvm.fptosi.sat.v16i32.v16f32(<16 x float> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %v16f32u32 = call <16 x i32> @llvm.fptoui.sat.v16i32.v16f32(<16 x float> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 144 for instruction: %v16f32s64 = call <16 x i64> @llvm.fptosi.sat.v16i64.v16f32(<16 x float> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 152 for instruction: %v16f32s64 = call <16 x i64> @llvm.fptosi.sat.v16i64.v16f32(<16 x float> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 180 for instruction: %v16f32u64 = call <16 x i64> @llvm.fptoui.sat.v16i64.v16f32(<16 x float> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 89 for instruction: %v16f64s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f64(<16 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 90 for instruction: %v16f64s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f64(<16 x double> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %v16f64u1 = call <16 x i1> @llvm.fptoui.sat.v16i1.v16f64(<16 x double> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 89 for instruction: %v16f64s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f64(<16 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 90 for instruction: %v16f64s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f64(<16 x double> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %v16f64u8 = call <16 x i8> @llvm.fptoui.sat.v16i8.v16f64(<16 x double> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 82 for instruction: %v16f64s16 = call <16 x i16> @llvm.fptosi.sat.v16i16.v16f64(<16 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %v16f64s16 = call <16 x i16> @llvm.fptosi.sat.v16i16.v16f64(<16 x double> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %v16f64u16 = call <16 x i16> @llvm.fptoui.sat.v16i16.v16f64(<16 x double> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 84 for instruction: %v16f64s32 = call <16 x i32> @llvm.fptosi.sat.v16i32.v16f64(<16 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %v16f64s32 = call <16 x i32> @llvm.fptosi.sat.v16i32.v16f64(<16 x double> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %v16f64u32 = call <16 x i32> @llvm.fptoui.sat.v16i32.v16f64(<16 x double> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 176 for instruction: %v16f64s64 = call <16 x i64> @llvm.fptosi.sat.v16i64.v16f64(<16 x double> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 184 for instruction: %v16f64s64 = call <16 x i64> @llvm.fptosi.sat.v16i64.v16f64(<16 x double> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 208 for instruction: %v16f64u64 = call <16 x i64> @llvm.fptoui.sat.v16i64.v16f64(<16 x double> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
@@ -910,45 +910,45 @@ define void @fp16() {
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %f16u32 = call i32 @llvm.fptoui.sat.i32.f16(half undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %f16s64 = call i64 @llvm.fptosi.sat.i64.f16(half undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %f16u64 = call i64 @llvm.fptoui.sat.i64.f16(half undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %v2f16s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f16(<2 x half> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %v2f16s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f16(<2 x half> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %v2f16u1 = call <2 x i1> @llvm.fptoui.sat.v2i1.v2f16(<2 x half> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v2f16s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f16(<2 x half> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %v2f16s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f16(<2 x half> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %v2f16u8 = call <2 x i8> @llvm.fptoui.sat.v2i8.v2f16(<2 x half> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %v2f16s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f16(<2 x half> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v2f16s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f16(<2 x half> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v2f16u16 = call <2 x i16> @llvm.fptoui.sat.v2i16.v2f16(<2 x half> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %v2f16s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f16(<2 x half> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %v2f16s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f16(<2 x half> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %v2f16u32 = call <2 x i32> @llvm.fptoui.sat.v2i32.v2f16(<2 x half> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %v2f16s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f16(<2 x half> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %v2f16s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f16(<2 x half> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %v2f16u64 = call <2 x i64> @llvm.fptoui.sat.v2i64.v2f16(<2 x half> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %v4f16s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f16(<4 x half> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %v4f16s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f16(<4 x half> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %v4f16u1 = call <4 x i1> @llvm.fptoui.sat.v4i1.v4f16(<4 x half> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %v4f16s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f16(<4 x half> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %v4f16s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f16(<4 x half> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %v4f16u8 = call <4 x i8> @llvm.fptoui.sat.v4i8.v4f16(<4 x half> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %v4f16s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f16(<4 x half> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %v4f16s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f16(<4 x half> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %v4f16u16 = call <4 x i16> @llvm.fptoui.sat.v4i16.v4f16(<4 x half> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %v4f16s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f16(<4 x half> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %v4f16s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f16(<4 x half> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %v4f16u32 = call <4 x i32> @llvm.fptoui.sat.v4i32.v4f16(<4 x half> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %v4f16s64 = call <4 x i64> @llvm.fptosi.sat.v4i64.v4f16(<4 x half> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %v4f16s64 = call <4 x i64> @llvm.fptosi.sat.v4i64.v4f16(<4 x half> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %v4f16u64 = call <4 x i64> @llvm.fptoui.sat.v4i64.v4f16(<4 x half> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %v8f16s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f16(<8 x half> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 89 for instruction: %v8f16s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f16(<8 x half> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %v8f16u1 = call <8 x i1> @llvm.fptoui.sat.v8i1.v8f16(<8 x half> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %v8f16s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f16(<8 x half> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %v8f16s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f16(<8 x half> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 82 for instruction: %v8f16u8 = call <8 x i8> @llvm.fptoui.sat.v8i8.v8f16(<8 x half> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %v8f16s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f16(<8 x half> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 93 for instruction: %v8f16s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f16(<8 x half> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 83 for instruction: %v8f16u16 = call <8 x i16> @llvm.fptoui.sat.v8i16.v8f16(<8 x half> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %v8f16s32 = call <8 x i32> @llvm.fptosi.sat.v8i32.v8f16(<8 x half> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 90 for instruction: %v8f16s32 = call <8 x i32> @llvm.fptosi.sat.v8i32.v8f16(<8 x half> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %v8f16u32 = call <8 x i32> @llvm.fptoui.sat.v8i32.v8f16(<8 x half> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %v8f16s64 = call <8 x i64> @llvm.fptosi.sat.v8i64.v8f16(<8 x half> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %v8f16s64 = call <8 x i64> @llvm.fptosi.sat.v8i64.v8f16(<8 x half> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %v8f16u64 = call <8 x i64> @llvm.fptoui.sat.v8i64.v8f16(<8 x half> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 176 for instruction: %v16f16s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f16(<16 x half> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 177 for instruction: %v16f16s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f16(<16 x half> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 159 for instruction: %v16f16u1 = call <16 x i1> @llvm.fptoui.sat.v16i1.v16f16(<16 x half> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 180 for instruction: %v16f16s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f16(<16 x half> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 181 for instruction: %v16f16s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f16(<16 x half> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: %v16f16u8 = call <16 x i8> @llvm.fptoui.sat.v16i8.v16f16(<16 x half> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 184 for instruction: %v16f16s16 = call <16 x i16> @llvm.fptosi.sat.v16i16.v16f16(<16 x half> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 186 for instruction: %v16f16s16 = call <16 x i16> @llvm.fptosi.sat.v16i16.v16f16(<16 x half> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 166 for instruction: %v16f16u16 = call <16 x i16> @llvm.fptoui.sat.v16i16.v16f16(<16 x half> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 176 for instruction: %v16f16s32 = call <16 x i32> @llvm.fptosi.sat.v16i32.v16f16(<16 x half> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 180 for instruction: %v16f16s32 = call <16 x i32> @llvm.fptosi.sat.v16i32.v16f16(<16 x half> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 156 for instruction: %v16f16u32 = call <16 x i32> @llvm.fptoui.sat.v16i32.v16f16(<16 x half> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 176 for instruction: %v16f16s64 = call <16 x i64> @llvm.fptosi.sat.v16i64.v16f16(<16 x half> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 184 for instruction: %v16f16s64 = call <16 x i64> @llvm.fptosi.sat.v16i64.v16f16(<16 x half> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 200 for instruction: %v16f16u64 = call <16 x i64> @llvm.fptoui.sat.v16i64.v16f16(<16 x half> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
@@ -963,39 +963,39 @@ define void @fp16() {
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %f16u32 = call i32 @llvm.fptoui.sat.i32.f16(half undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %f16s64 = call i64 @llvm.fptosi.sat.i64.f16(half undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %f16u64 = call i64 @llvm.fptoui.sat.i64.f16(half undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %v2f16s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f16(<2 x half> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %v2f16s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f16(<2 x half> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v2f16u1 = call <2 x i1> @llvm.fptoui.sat.v2i1.v2f16(<2 x half> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v2f16s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f16(<2 x half> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %v2f16s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f16(<2 x half> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %v2f16u8 = call <2 x i8> @llvm.fptoui.sat.v2i8.v2f16(<2 x half> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %v2f16s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f16(<2 x half> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v2f16s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f16(<2 x half> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v2f16u16 = call <2 x i16> @llvm.fptoui.sat.v2i16.v2f16(<2 x half> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %v2f16s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f16(<2 x half> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %v2f16s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f16(<2 x half> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v2f16u32 = call <2 x i32> @llvm.fptoui.sat.v2i32.v2f16(<2 x half> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %v2f16s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f16(<2 x half> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %v2f16s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f16(<2 x half> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %v2f16u64 = call <2 x i64> @llvm.fptoui.sat.v2i64.v2f16(<2 x half> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %v4f16s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f16(<4 x half> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %v4f16s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f16(<4 x half> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %v4f16u1 = call <4 x i1> @llvm.fptoui.sat.v4i1.v4f16(<4 x half> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %v4f16s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f16(<4 x half> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %v4f16s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f16(<4 x half> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %v4f16u8 = call <4 x i8> @llvm.fptoui.sat.v4i8.v4f16(<4 x half> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %v4f16s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f16(<4 x half> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %v4f16s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f16(<4 x half> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %v4f16u16 = call <4 x i16> @llvm.fptoui.sat.v4i16.v4f16(<4 x half> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %v4f16s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f16(<4 x half> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %v4f16s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f16(<4 x half> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %v4f16u32 = call <4 x i32> @llvm.fptoui.sat.v4i32.v4f16(<4 x half> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %v4f16s64 = call <4 x i64> @llvm.fptosi.sat.v4i64.v4f16(<4 x half> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %v4f16u64 = call <4 x i64> @llvm.fptoui.sat.v4i64.v4f16(<4 x half> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %v8f16s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f16(<8 x half> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: %v8f16s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f16(<8 x half> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %v8f16u1 = call <8 x i1> @llvm.fptoui.sat.v8i1.v8f16(<8 x half> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: %v8f16s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f16(<8 x half> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %v8f16s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f16(<8 x half> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %v8f16u8 = call <8 x i8> @llvm.fptoui.sat.v8i8.v8f16(<8 x half> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: %v8f16s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f16(<8 x half> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %v8f16s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f16(<8 x half> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %v8f16u16 = call <8 x i16> @llvm.fptoui.sat.v8i16.v8f16(<8 x half> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %v8f16s32 = call <8 x i32> @llvm.fptosi.sat.v8i32.v8f16(<8 x half> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %v8f16u32 = call <8 x i32> @llvm.fptoui.sat.v8i32.v8f16(<8 x half> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %v8f16s64 = call <8 x i64> @llvm.fptosi.sat.v8i64.v8f16(<8 x half> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %v8f16u64 = call <8 x i64> @llvm.fptoui.sat.v8i64.v8f16(<8 x half> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 144 for instruction: %v16f16s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f16(<16 x half> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 145 for instruction: %v16f16s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f16(<16 x half> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 127 for instruction: %v16f16u1 = call <16 x i1> @llvm.fptoui.sat.v16i1.v16f16(<16 x half> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 149 for instruction: %v16f16s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f16(<16 x half> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 150 for instruction: %v16f16s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f16(<16 x half> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %v16f16u8 = call <16 x i8> @llvm.fptoui.sat.v16i8.v16f16(<16 x half> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 151 for instruction: %v16f16s16 = call <16 x i16> @llvm.fptosi.sat.v16i16.v16f16(<16 x half> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 132 for instruction: %v16f16u16 = call <16 x i16> @llvm.fptoui.sat.v16i16.v16f16(<16 x half> undef)
@@ -1016,39 +1016,39 @@ define void @fp16() {
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %f16u32 = call i32 @llvm.fptoui.sat.i32.f16(half undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %f16s64 = call i64 @llvm.fptosi.sat.i64.f16(half undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %f16u64 = call i64 @llvm.fptoui.sat.i64.f16(half undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %v2f16s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f16(<2 x half> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %v2f16s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f16(<2 x half> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v2f16u1 = call <2 x i1> @llvm.fptoui.sat.v2i1.v2f16(<2 x half> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %v2f16s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f16(<2 x half> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v2f16s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f16(<2 x half> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v2f16u8 = call <2 x i8> @llvm.fptoui.sat.v2i8.v2f16(<2 x half> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %v2f16s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f16(<2 x half> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v2f16s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f16(<2 x half> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v2f16u16 = call <2 x i16> @llvm.fptoui.sat.v2i16.v2f16(<2 x half> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %v2f16s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f16(<2 x half> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %v2f16s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f16(<2 x half> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v2f16u32 = call <2 x i32> @llvm.fptoui.sat.v2i32.v2f16(<2 x half> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %v2f16s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f16(<2 x half> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %v2f16s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f16(<2 x half> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %v2f16u64 = call <2 x i64> @llvm.fptoui.sat.v2i64.v2f16(<2 x half> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %v4f16s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f16(<4 x half> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %v4f16s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f16(<4 x half> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %v4f16u1 = call <4 x i1> @llvm.fptoui.sat.v4i1.v4f16(<4 x half> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %v4f16s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f16(<4 x half> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %v4f16s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f16(<4 x half> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %v4f16u8 = call <4 x i8> @llvm.fptoui.sat.v4i8.v4f16(<4 x half> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %v4f16s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f16(<4 x half> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %v4f16s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f16(<4 x half> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %v4f16u16 = call <4 x i16> @llvm.fptoui.sat.v4i16.v4f16(<4 x half> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %v4f16s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f16(<4 x half> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %v4f16s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f16(<4 x half> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %v4f16u32 = call <4 x i32> @llvm.fptoui.sat.v4i32.v4f16(<4 x half> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 37 for instruction: %v4f16s64 = call <4 x i64> @llvm.fptosi.sat.v4i64.v4f16(<4 x half> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %v4f16u64 = call <4 x i64> @llvm.fptoui.sat.v4i64.v4f16(<4 x half> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %v8f16s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f16(<8 x half> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: %v8f16s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f16(<8 x half> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %v8f16u1 = call <8 x i1> @llvm.fptoui.sat.v8i1.v8f16(<8 x half> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %v8f16s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f16(<8 x half> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 77 for instruction: %v8f16s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f16(<8 x half> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 67 for instruction: %v8f16u8 = call <8 x i8> @llvm.fptoui.sat.v8i8.v8f16(<8 x half> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %v8f16s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f16(<8 x half> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 75 for instruction: %v8f16s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f16(<8 x half> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 65 for instruction: %v8f16u16 = call <8 x i16> @llvm.fptoui.sat.v8i16.v8f16(<8 x half> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 73 for instruction: %v8f16s32 = call <8 x i32> @llvm.fptosi.sat.v8i32.v8f16(<8 x half> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 63 for instruction: %v8f16u32 = call <8 x i32> @llvm.fptoui.sat.v8i32.v8f16(<8 x half> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %v8f16s64 = call <8 x i64> @llvm.fptosi.sat.v8i64.v8f16(<8 x half> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %v8f16u64 = call <8 x i64> @llvm.fptoui.sat.v8i64.v8f16(<8 x half> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 144 for instruction: %v16f16s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f16(<16 x half> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 145 for instruction: %v16f16s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f16(<16 x half> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 127 for instruction: %v16f16u1 = call <16 x i1> @llvm.fptoui.sat.v16i1.v16f16(<16 x half> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 147 for instruction: %v16f16s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f16(<16 x half> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %v16f16s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f16(<16 x half> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: %v16f16u8 = call <16 x i8> @llvm.fptoui.sat.v16i8.v16f16(<16 x half> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 148 for instruction: %v16f16s16 = call <16 x i16> @llvm.fptosi.sat.v16i16.v16f16(<16 x half> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 130 for instruction: %v16f16u16 = call <16 x i16> @llvm.fptoui.sat.v16i16.v16f16(<16 x half> undef)
@@ -1122,45 +1122,45 @@ define void @fp16() {
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %f16u32 = call i32 @llvm.fptoui.sat.i32.f16(half undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %f16s64 = call i64 @llvm.fptosi.sat.i64.f16(half undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %f16u64 = call i64 @llvm.fptoui.sat.i64.f16(half undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %v2f16s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f16(<2 x half> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %v2f16s1 = call <2 x i1> @llvm.fptosi.sat.v2i1.v2f16(<2 x half> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %v2f16u1 = call <2 x i1> @llvm.fptoui.sat.v2i1.v2f16(<2 x half> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v2f16s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f16(<2 x half> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %v2f16s8 = call <2 x i8> @llvm.fptosi.sat.v2i8.v2f16(<2 x half> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %v2f16u8 = call <2 x i8> @llvm.fptoui.sat.v2i8.v2f16(<2 x half> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %v2f16s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f16(<2 x half> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v2f16s16 = call <2 x i16> @llvm.fptosi.sat.v2i16.v2f16(<2 x half> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v2f16u16 = call <2 x i16> @llvm.fptoui.sat.v2i16.v2f16(<2 x half> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %v2f16s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f16(<2 x half> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %v2f16s32 = call <2 x i32> @llvm.fptosi.sat.v2i32.v2f16(<2 x half> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %v2f16u32 = call <2 x i32> @llvm.fptoui.sat.v2i32.v2f16(<2 x half> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %v2f16s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f16(<2 x half> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %v2f16s64 = call <2 x i64> @llvm.fptosi.sat.v2i64.v2f16(<2 x half> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %v2f16u64 = call <2 x i64> @llvm.fptoui.sat.v2i64.v2f16(<2 x half> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %v4f16s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f16(<4 x half> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %v4f16s1 = call <4 x i1> @llvm.fptosi.sat.v4i1.v4f16(<4 x half> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %v4f16u1 = call <4 x i1> @llvm.fptoui.sat.v4i1.v4f16(<4 x half> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %v4f16s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f16(<4 x half> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %v4f16s8 = call <4 x i8> @llvm.fptosi.sat.v4i8.v4f16(<4 x half> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %v4f16u8 = call <4 x i8> @llvm.fptoui.sat.v4i8.v4f16(<4 x half> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %v4f16s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f16(<4 x half> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 47 for instruction: %v4f16s16 = call <4 x i16> @llvm.fptosi.sat.v4i16.v4f16(<4 x half> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 41 for instruction: %v4f16u16 = call <4 x i16> @llvm.fptoui.sat.v4i16.v4f16(<4 x half> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %v4f16s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f16(<4 x half> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %v4f16s32 = call <4 x i32> @llvm.fptosi.sat.v4i32.v4f16(<4 x half> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 39 for instruction: %v4f16u32 = call <4 x i32> @llvm.fptoui.sat.v4i32.v4f16(<4 x half> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %v4f16s64 = call <4 x i64> @llvm.fptosi.sat.v4i64.v4f16(<4 x half> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %v4f16s64 = call <4 x i64> @llvm.fptosi.sat.v4i64.v4f16(<4 x half> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %v4f16u64 = call <4 x i64> @llvm.fptoui.sat.v4i64.v4f16(<4 x half> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %v8f16s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f16(<8 x half> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 89 for instruction: %v8f16s1 = call <8 x i1> @llvm.fptosi.sat.v8i1.v8f16(<8 x half> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %v8f16u1 = call <8 x i1> @llvm.fptoui.sat.v8i1.v8f16(<8 x half> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %v8f16s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f16(<8 x half> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %v8f16s8 = call <8 x i8> @llvm.fptosi.sat.v8i8.v8f16(<8 x half> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 82 for instruction: %v8f16u8 = call <8 x i8> @llvm.fptoui.sat.v8i8.v8f16(<8 x half> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %v8f16s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f16(<8 x half> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 93 for instruction: %v8f16s16 = call <8 x i16> @llvm.fptosi.sat.v8i16.v8f16(<8 x half> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 83 for instruction: %v8f16u16 = call <8 x i16> @llvm.fptoui.sat.v8i16.v8f16(<8 x half> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %v8f16s32 = call <8 x i32> @llvm.fptosi.sat.v8i32.v8f16(<8 x half> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 90 for instruction: %v8f16s32 = call <8 x i32> @llvm.fptosi.sat.v8i32.v8f16(<8 x half> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 78 for instruction: %v8f16u32 = call <8 x i32> @llvm.fptoui.sat.v8i32.v8f16(<8 x half> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %v8f16s64 = call <8 x i64> @llvm.fptosi.sat.v8i64.v8f16(<8 x half> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %v8f16s64 = call <8 x i64> @llvm.fptosi.sat.v8i64.v8f16(<8 x half> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %v8f16u64 = call <8 x i64> @llvm.fptoui.sat.v8i64.v8f16(<8 x half> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 176 for instruction: %v16f16s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f16(<16 x half> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 177 for instruction: %v16f16s1 = call <16 x i1> @llvm.fptosi.sat.v16i1.v16f16(<16 x half> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 159 for instruction: %v16f16u1 = call <16 x i1> @llvm.fptoui.sat.v16i1.v16f16(<16 x half> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 180 for instruction: %v16f16s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f16(<16 x half> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 181 for instruction: %v16f16s8 = call <16 x i8> @llvm.fptosi.sat.v16i8.v16f16(<16 x half> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 163 for instruction: %v16f16u8 = call <16 x i8> @llvm.fptoui.sat.v16i8.v16f16(<16 x half> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 184 for instruction: %v16f16s16 = call <16 x i16> @llvm.fptosi.sat.v16i16.v16f16(<16 x half> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 186 for instruction: %v16f16s16 = call <16 x i16> @llvm.fptosi.sat.v16i16.v16f16(<16 x half> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 166 for instruction: %v16f16u16 = call <16 x i16> @llvm.fptoui.sat.v16i16.v16f16(<16 x half> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 176 for instruction: %v16f16s32 = call <16 x i32> @llvm.fptosi.sat.v16i32.v16f16(<16 x half> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 180 for instruction: %v16f16s32 = call <16 x i32> @llvm.fptosi.sat.v16i32.v16f16(<16 x half> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 156 for instruction: %v16f16u32 = call <16 x i32> @llvm.fptoui.sat.v16i32.v16f16(<16 x half> undef)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 176 for instruction: %v16f16s64 = call <16 x i64> @llvm.fptosi.sat.v16i64.v16f16(<16 x half> undef)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 184 for instruction: %v16f16s64 = call <16 x i64> @llvm.fptosi.sat.v16i64.v16f16(<16 x half> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 200 for instruction: %v16f16u64 = call <16 x i64> @llvm.fptoui.sat.v16i64.v16f16(<16 x half> undef)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
diff --git a/llvm/test/Analysis/CostModel/X86/fshl.ll b/llvm/test/Analysis/CostModel/X86/fshl.ll
index 0272e159f9c6d..4eb7c05681534 100644
--- a/llvm/test/Analysis/CostModel/X86/fshl.ll
+++ b/llvm/test/Analysis/CostModel/X86/fshl.ll
@@ -29,21 +29,21 @@ define void @var_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8 x i64
 ;
 ; SSE42-LABEL: 'var_funnel_i64'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 %c64)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'var_funnel_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 %c64)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'var_funnel_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 %c64)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -57,21 +57,21 @@ define void @var_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8 x i64
 ;
 ; SLM-LABEL: 'var_funnel_i64'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 %c64)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'var_funnel_i64'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 %c64)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'var_funnel_i64'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 %c64)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -93,21 +93,21 @@ define void @var_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i3
 ;
 ; SSE42-LABEL: 'var_funnel_i32'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 %c32)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'var_funnel_i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 %c32)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'var_funnel_i32'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 %c32)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -121,21 +121,21 @@ define void @var_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i3
 ;
 ; SLM-LABEL: 'var_funnel_i32'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 %c32)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'var_funnel_i32'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 %c32)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'var_funnel_i32'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 %c32)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -157,21 +157,21 @@ define void @var_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i
 ;
 ; SSE42-LABEL: 'var_funnel_i16'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 %c16)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 120 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'var_funnel_i16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 %c16)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 108 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'var_funnel_i16'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 %c16)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -199,21 +199,21 @@ define void @var_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i
 ;
 ; SLM-LABEL: 'var_funnel_i16'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 %c16)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 120 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'var_funnel_i16'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 %c16)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 120 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'var_funnel_i16'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 %c16)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -235,21 +235,21 @@ define void @var_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x i8> %
 ;
 ; SSE42-LABEL: 'var_funnel_i8'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 %c8)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 108 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'var_funnel_i8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 %c8)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'var_funnel_i8'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 %c8)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -277,21 +277,21 @@ define void @var_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x i8> %
 ;
 ; SLM-LABEL: 'var_funnel_i8'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 %c8)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 108 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'var_funnel_i8'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 %c8)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 108 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'var_funnel_i8'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 %c8)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -321,16 +321,16 @@ define void @splatvar_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatvar_funnel_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -339,7 +339,7 @@ define void @splatvar_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -357,25 +357,25 @@ define void @splatvar_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatvar_funnel_i64'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatvar_funnel_i64'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; XOP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -390,29 +390,20 @@ define void @splatvar_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
 }
 
 define void @splatvar_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i32> %a512, i32 %b32, <4 x i32> %b128, <8 x i32> %b256, <16 x i32> %b512, i32 %c32, <4 x i32> %c128, <8 x i32> %c256, <16 x i32> %c512) {
-; SSSE3-LABEL: 'splatvar_funnel_i32'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
-;
-; SSE42-LABEL: 'splatvar_funnel_i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+; SSE-LABEL: 'splatvar_funnel_i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatvar_funnel_i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -421,7 +412,7 @@ define void @splatvar_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -439,25 +430,25 @@ define void @splatvar_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatvar_funnel_i32'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatvar_funnel_i32'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; XOP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -472,29 +463,20 @@ define void @splatvar_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16
 }
 
 define void @splatvar_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i16> %a512, i16 %b16, <8 x i16> %b128, <16 x i16> %b256, <32 x i16> %b512, i16 %c16, <8 x i16> %c128, <16 x i16> %c256, <32 x i16> %c512) {
-; SSSE3-LABEL: 'splatvar_funnel_i16'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
-;
-; SSE42-LABEL: 'splatvar_funnel_i16'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+; SSE-LABEL: 'splatvar_funnel_i16'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatvar_funnel_i16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -503,7 +485,7 @@ define void @splatvar_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -539,25 +521,25 @@ define void @splatvar_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatvar_funnel_i16'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatvar_funnel_i16'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; XOP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -585,16 +567,16 @@ define void @splatvar_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 108 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatvar_funnel_i8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -603,7 +585,7 @@ define void @splatvar_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -639,25 +621,25 @@ define void @splatvar_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 108 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatvar_funnel_i8'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; GLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 108 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatvar_funnel_i8'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; XOP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -685,21 +667,21 @@ define void @constant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
 ;
 ; SSE42-LABEL: 'constant_funnel_i64'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'constant_funnel_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'constant_funnel_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -713,21 +695,21 @@ define void @constant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
 ;
 ; SLM-LABEL: 'constant_funnel_i64'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'constant_funnel_i64'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'constant_funnel_i64'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -740,30 +722,23 @@ define void @constant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
 }
 
 define void @constant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i32> %a512, i32 %b32, <4 x i32> %b128, <8 x i32> %b256, <16 x i32> %b512) {
-; SSSE3-LABEL: 'constant_funnel_i32'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 7)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
-;
-; SSE42-LABEL: 'constant_funnel_i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 7)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+; SSE-LABEL: 'constant_funnel_i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 7)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'constant_funnel_i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 7)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'constant_funnel_i32'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 7)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -777,21 +752,21 @@ define void @constant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16
 ;
 ; SLM-LABEL: 'constant_funnel_i32'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 7)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'constant_funnel_i32'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 7)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'constant_funnel_i32'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 7)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -813,21 +788,21 @@ define void @constant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
 ;
 ; SSE42-LABEL: 'constant_funnel_i16'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 7)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'constant_funnel_i16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 7)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'constant_funnel_i16'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 7)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -855,21 +830,21 @@ define void @constant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
 ;
 ; SLM-LABEL: 'constant_funnel_i16'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 7)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'constant_funnel_i16'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 7)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'constant_funnel_i16'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 7)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -891,21 +866,21 @@ define void @constant_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
 ;
 ; SSE42-LABEL: 'constant_funnel_i8'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 7)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'constant_funnel_i8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 7)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 114 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'constant_funnel_i8'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 7)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -933,21 +908,21 @@ define void @constant_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
 ;
 ; SLM-LABEL: 'constant_funnel_i8'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 7)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'constant_funnel_i8'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 7)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'constant_funnel_i8'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 7)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -973,21 +948,21 @@ define void @splatconstant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256
 ;
 ; SSE42-LABEL: 'splatconstant_funnel_i64'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatconstant_funnel_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatconstant_funnel_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -1001,21 +976,21 @@ define void @splatconstant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256
 ;
 ; SLM-LABEL: 'splatconstant_funnel_i64'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatconstant_funnel_i64'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatconstant_funnel_i64'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshl.i64(i64 %a64, i64 %b64, i64 7)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshl.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.fshl.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.fshl.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -1028,30 +1003,23 @@ define void @splatconstant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256
 }
 
 define void @splatconstant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i32> %a512, i32 %b32, <4 x i32> %b128, <8 x i32> %b256, <16 x i32> %b512) {
-; SSSE3-LABEL: 'splatconstant_funnel_i32'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 5)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
-;
-; SSE42-LABEL: 'splatconstant_funnel_i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 5)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+; SSE-LABEL: 'splatconstant_funnel_i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 5)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatconstant_funnel_i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 5)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatconstant_funnel_i32'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 5)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -1065,21 +1033,21 @@ define void @splatconstant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256
 ;
 ; SLM-LABEL: 'splatconstant_funnel_i32'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 5)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatconstant_funnel_i32'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 5)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatconstant_funnel_i32'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshl.i32(i32 %a32, i32 %b32, i32 5)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <8 x i32> @llvm.fshl.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <16 x i32> @llvm.fshl.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -1092,30 +1060,23 @@ define void @splatconstant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256
 }
 
 define void @splatconstant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i16> %a512, i16 %b16, <8 x i16> %b128, <16 x i16> %b256, <32 x i16> %b512) {
-; SSSE3-LABEL: 'splatconstant_funnel_i16'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 3)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
-;
-; SSE42-LABEL: 'splatconstant_funnel_i16'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 3)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+; SSE-LABEL: 'splatconstant_funnel_i16'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 3)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatconstant_funnel_i16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 3)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatconstant_funnel_i16'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 3)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -1143,21 +1104,21 @@ define void @splatconstant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a25
 ;
 ; SLM-LABEL: 'splatconstant_funnel_i16'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 3)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatconstant_funnel_i16'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 3)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatconstant_funnel_i16'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshl.i16(i16 %a16, i16 %b16, i16 3)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshl.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I16 = call <16 x i16> @llvm.fshl.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32I16 = call <32 x i16> @llvm.fshl.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -1170,30 +1131,23 @@ define void @splatconstant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a25
 }
 
 define void @splatconstant_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x i8> %a512, i8 %b8, <16 x i8> %b128, <32 x i8> %b256, <64 x i8> %b512) {
-; SSSE3-LABEL: 'splatconstant_funnel_i8'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 3)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
-;
-; SSE42-LABEL: 'splatconstant_funnel_i8'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 3)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+; SSE-LABEL: 'splatconstant_funnel_i8'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 3)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatconstant_funnel_i8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 3)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatconstant_funnel_i8'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 3)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -1221,21 +1175,21 @@ define void @splatconstant_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <
 ;
 ; SLM-LABEL: 'splatconstant_funnel_i8'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 3)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatconstant_funnel_i8'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 3)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatconstant_funnel_i8'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshl.i8(i8 %a8, i8 %b8, i8 3)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16I8 = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I8 = call <32 x i8> @llvm.fshl.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V64I8 = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
diff --git a/llvm/test/Analysis/CostModel/X86/fshr.ll b/llvm/test/Analysis/CostModel/X86/fshr.ll
index 14c4da0cb7b34..3aef33a7b6630 100644
--- a/llvm/test/Analysis/CostModel/X86/fshr.ll
+++ b/llvm/test/Analysis/CostModel/X86/fshr.ll
@@ -29,21 +29,21 @@ define void @var_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8 x i64
 ;
 ; SSE42-LABEL: 'var_funnel_i64'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 %c64)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'var_funnel_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 %c64)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'var_funnel_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 %c64)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -57,21 +57,21 @@ define void @var_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8 x i64
 ;
 ; SLM-LABEL: 'var_funnel_i64'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 %c64)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'var_funnel_i64'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 %c64)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'var_funnel_i64'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 %c64)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %c128)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %c256)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %c512)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -93,21 +93,21 @@ define void @var_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i3
 ;
 ; SSE42-LABEL: 'var_funnel_i32'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 %c32)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'var_funnel_i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 %c32)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'var_funnel_i32'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 %c32)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -121,21 +121,21 @@ define void @var_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i3
 ;
 ; SLM-LABEL: 'var_funnel_i32'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 %c32)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'var_funnel_i32'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 %c32)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'var_funnel_i32'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 %c32)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %c128)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %c256)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %c512)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -157,21 +157,21 @@ define void @var_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i
 ;
 ; SSE42-LABEL: 'var_funnel_i16'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 %c16)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 120 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'var_funnel_i16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 %c16)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 108 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'var_funnel_i16'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 %c16)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -199,21 +199,21 @@ define void @var_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i
 ;
 ; SLM-LABEL: 'var_funnel_i16'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 %c16)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 120 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'var_funnel_i16'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 %c16)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 29 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 30 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 120 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'var_funnel_i16'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 %c16)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %c128)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %c256)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %c512)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -235,21 +235,21 @@ define void @var_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x i8> %
 ;
 ; SSE42-LABEL: 'var_funnel_i8'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 %c8)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 108 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'var_funnel_i8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 %c8)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'var_funnel_i8'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 %c8)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -277,21 +277,21 @@ define void @var_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x i8> %
 ;
 ; SLM-LABEL: 'var_funnel_i8'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 %c8)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 108 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'var_funnel_i8'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 %c8)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 108 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'var_funnel_i8'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 %c8)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %c128)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %c256)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %c512)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -321,16 +321,16 @@ define void @splatvar_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatvar_funnel_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -339,7 +339,7 @@ define void @splatvar_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -357,25 +357,25 @@ define void @splatvar_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatvar_funnel_i64'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatvar_funnel_i64'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <2 x i64> %c128, <2 x i64> undef, <2 x i32> zeroinitializer
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u256 = shufflevector <4 x i64> %c256, <4 x i64> undef, <4 x i32> zeroinitializer
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u512 = shufflevector <8 x i64> %c512, <8 x i64> undef, <8 x i32> zeroinitializer
-; XOP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> %u128)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> %u256)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> %u512)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -390,29 +390,20 @@ define void @splatvar_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
 }
 
 define void @splatvar_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i32> %a512, i32 %b32, <4 x i32> %b128, <8 x i32> %b256, <16 x i32> %b512, i32 %c32, <4 x i32> %c128, <8 x i32> %c256, <16 x i32> %c512) {
-; SSSE3-LABEL: 'splatvar_funnel_i32'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
-;
-; SSE42-LABEL: 'splatvar_funnel_i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+; SSE-LABEL: 'splatvar_funnel_i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatvar_funnel_i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -421,7 +412,7 @@ define void @splatvar_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -439,25 +430,25 @@ define void @splatvar_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatvar_funnel_i32'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatvar_funnel_i32'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <4 x i32> %c128, <4 x i32> undef, <4 x i32> zeroinitializer
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u256 = shufflevector <8 x i32> %c256, <8 x i32> undef, <8 x i32> zeroinitializer
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u512 = shufflevector <16 x i32> %c512, <16 x i32> undef, <16 x i32> zeroinitializer
-; XOP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> %u128)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> %u256)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> %u512)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -472,29 +463,20 @@ define void @splatvar_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16
 }
 
 define void @splatvar_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i16> %a512, i16 %b16, <8 x i16> %b128, <16 x i16> %b256, <32 x i16> %b512, i16 %c16, <8 x i16> %c128, <16 x i16> %c256, <32 x i16> %c512) {
-; SSSE3-LABEL: 'splatvar_funnel_i16'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
-;
-; SSE42-LABEL: 'splatvar_funnel_i16'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+; SSE-LABEL: 'splatvar_funnel_i16'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatvar_funnel_i16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -503,7 +485,7 @@ define void @splatvar_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -539,25 +521,25 @@ define void @splatvar_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatvar_funnel_i16'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatvar_funnel_i16'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <8 x i16> %c128, <8 x i16> undef, <8 x i32> zeroinitializer
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %u256 = shufflevector <16 x i16> %c256, <16 x i16> undef, <16 x i32> zeroinitializer
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %u512 = shufflevector <32 x i16> %c512, <32 x i16> undef, <32 x i32> zeroinitializer
-; XOP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> %u128)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> %u256)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> %u512)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -585,16 +567,16 @@ define void @splatvar_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 108 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatvar_funnel_i8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -603,7 +585,7 @@ define void @splatvar_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -639,25 +621,25 @@ define void @splatvar_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; SLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 108 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatvar_funnel_i8'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; GLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 27 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 108 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatvar_funnel_i8'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %u128 = shufflevector <16 x i8> %c128, <16 x i8> undef, <16 x i32> zeroinitializer
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u256 = shufflevector <32 x i8> %c256, <32 x i8> undef, <32 x i32> zeroinitializer
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %u512 = shufflevector <64 x i8> %c512, <64 x i8> undef, <64 x i32> zeroinitializer
-; XOP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> %u128)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> %u256)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> %u512)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -685,21 +667,21 @@ define void @constant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
 ;
 ; SSE42-LABEL: 'constant_funnel_i64'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'constant_funnel_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'constant_funnel_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -713,21 +695,21 @@ define void @constant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
 ;
 ; SLM-LABEL: 'constant_funnel_i64'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'constant_funnel_i64'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'constant_funnel_i64'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 1, i64 7>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 1, i64 7, i64 15, i64 31>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -740,30 +722,23 @@ define void @constant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256, <8
 }
 
 define void @constant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i32> %a512, i32 %b32, <4 x i32> %b128, <8 x i32> %b256, <16 x i32> %b512) {
-; SSSE3-LABEL: 'constant_funnel_i32'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 7)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
-;
-; SSE42-LABEL: 'constant_funnel_i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 7)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+; SSE-LABEL: 'constant_funnel_i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 7)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'constant_funnel_i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 7)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'constant_funnel_i32'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 7)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -777,21 +752,21 @@ define void @constant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16
 ;
 ; SLM-LABEL: 'constant_funnel_i32'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 7)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'constant_funnel_i32'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 7)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 92 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'constant_funnel_i32'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 7)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 4, i32 5, i32 6, i32 7>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -813,21 +788,21 @@ define void @constant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
 ;
 ; SSE42-LABEL: 'constant_funnel_i16'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 7)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'constant_funnel_i16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 7)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 88 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'constant_funnel_i16'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 7)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -855,21 +830,21 @@ define void @constant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <3
 ;
 ; SLM-LABEL: 'constant_funnel_i16'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 7)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'constant_funnel_i16'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 7)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 76 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'constant_funnel_i16'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 7)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -891,21 +866,21 @@ define void @constant_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
 ;
 ; SSE42-LABEL: 'constant_funnel_i8'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 7)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'constant_funnel_i8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 7)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 57 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 114 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'constant_funnel_i8'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 7)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -933,21 +908,21 @@ define void @constant_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x
 ;
 ; SLM-LABEL: 'constant_funnel_i8'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 7)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'constant_funnel_i8'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 7)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 52 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 104 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'constant_funnel_i8'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 7)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -973,21 +948,21 @@ define void @splatconstant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256
 ;
 ; SSE42-LABEL: 'splatconstant_funnel_i64'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatconstant_funnel_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatconstant_funnel_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -1001,21 +976,21 @@ define void @splatconstant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256
 ;
 ; SLM-LABEL: 'splatconstant_funnel_i64'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatconstant_funnel_i64'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatconstant_funnel_i64'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I64 = call i64 @llvm.fshr.i64(i64 %a64, i64 %b64, i64 7)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I64 = call <2 x i64> @llvm.fshr.v2i64(<2 x i64> %a128, <2 x i64> %b128, <2 x i64> <i64 7, i64 7>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I64 = call <4 x i64> @llvm.fshr.v4i64(<4 x i64> %a256, <4 x i64> %b256, <4 x i64> <i64 7, i64 7, i64 7, i64 7>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I64 = call <8 x i64> @llvm.fshr.v8i64(<8 x i64> %a512, <8 x i64> %b512, <8 x i64> <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -1028,30 +1003,23 @@ define void @splatconstant_funnel_i64(i64 %a64, <2 x i64> %a128, <4 x i64> %a256
 }
 
 define void @splatconstant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256, <16 x i32> %a512, i32 %b32, <4 x i32> %b128, <8 x i32> %b256, <16 x i32> %b512) {
-; SSSE3-LABEL: 'splatconstant_funnel_i32'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 5)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
-;
-; SSE42-LABEL: 'splatconstant_funnel_i32'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 5)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+; SSE-LABEL: 'splatconstant_funnel_i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 5)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatconstant_funnel_i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 5)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatconstant_funnel_i32'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 5)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -1065,21 +1033,21 @@ define void @splatconstant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256
 ;
 ; SLM-LABEL: 'splatconstant_funnel_i32'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 5)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatconstant_funnel_i32'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 5)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatconstant_funnel_i32'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.fshr.i32(i32 %a32, i32 %b32, i32 5)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2I32 = call <4 x i32> @llvm.fshr.v4i32(<4 x i32> %a128, <4 x i32> %b128, <4 x i32> <i32 5, i32 5, i32 5, i32 5>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <8 x i32> @llvm.fshr.v8i32(<8 x i32> %a256, <8 x i32> %b256, <8 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <16 x i32> @llvm.fshr.v16i32(<16 x i32> %a512, <16 x i32> %b512, <16 x i32> <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -1092,30 +1060,23 @@ define void @splatconstant_funnel_i32(i32 %a32, <4 x i32> %a128, <8 x i32> %a256
 }
 
 define void @splatconstant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a256, <32 x i16> %a512, i16 %b16, <8 x i16> %b128, <16 x i16> %b256, <32 x i16> %b512) {
-; SSSE3-LABEL: 'splatconstant_funnel_i16'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 3)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
-;
-; SSE42-LABEL: 'splatconstant_funnel_i16'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 3)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+; SSE-LABEL: 'splatconstant_funnel_i16'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 3)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatconstant_funnel_i16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 3)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatconstant_funnel_i16'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 3)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -1143,21 +1104,21 @@ define void @splatconstant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a25
 ;
 ; SLM-LABEL: 'splatconstant_funnel_i16'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 3)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatconstant_funnel_i16'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 3)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatconstant_funnel_i16'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.fshr.i16(i16 %a16, i16 %b16, i16 3)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V8I16 = call <8 x i16> @llvm.fshr.v8i16(<8 x i16> %a128, <8 x i16> %b128, <8 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I16 = call <16 x i16> @llvm.fshr.v16i16(<16 x i16> %a256, <16 x i16> %b256, <16 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32I16 = call <32 x i16> @llvm.fshr.v32i16(<32 x i16> %a512, <32 x i16> %b512, <32 x i16> <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -1170,30 +1131,23 @@ define void @splatconstant_funnel_i16(i16 %a16, <8 x i16> %a128, <16 x i16> %a25
 }
 
 define void @splatconstant_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <64 x i8> %a512, i8 %b8, <16 x i8> %b128, <32 x i8> %b256, <64 x i8> %b512) {
-; SSSE3-LABEL: 'splatconstant_funnel_i8'
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 3)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SSSE3-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
-;
-; SSE42-LABEL: 'splatconstant_funnel_i8'
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 3)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
+; SSE-LABEL: 'splatconstant_funnel_i8'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 3)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX1-LABEL: 'splatconstant_funnel_i8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 3)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; AVX2-LABEL: 'splatconstant_funnel_i8'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 3)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
@@ -1221,21 +1175,21 @@ define void @splatconstant_funnel_i8(i8 %a8, <16 x i8> %a128, <32 x i8> %a256, <
 ;
 ; SLM-LABEL: 'splatconstant_funnel_i8'
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 3)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; SLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; SLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; SLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; GLM-LABEL: 'splatconstant_funnel_i8'
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 3)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
-; GLM-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; GLM-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; GLM-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
 ;
 ; XOP-LABEL: 'splatconstant_funnel_i8'
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.fshr.i8(i8 %a8, i8 %b8, i8 3)
-; XOP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
+; XOP-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V16I8 = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %a128, <16 x i8> %b128, <16 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V32I8 = call <32 x i8> @llvm.fshr.v32i8(<32 x i8> %a256, <32 x i8> %b256, <32 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V64I8 = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %a512, <64 x i8> %b512, <64 x i8> <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3>)
 ; XOP-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret void
diff --git a/llvm/test/Analysis/CostModel/X86/reduce-smax.ll b/llvm/test/Analysis/CostModel/X86/reduce-smax.ll
index 44f12c803a0d7..b5ba92f1b681f 100644
--- a/llvm/test/Analysis/CostModel/X86/reduce-smax.ll
+++ b/llvm/test/Analysis/CostModel/X86/reduce-smax.ll
@@ -28,34 +28,34 @@ define i32 @reduce_i64(i32 %arg) {
 ;
 ; SSE41-LABEL: 'reduce_i64'
 ; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.smax.v1i64(<1 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2 = call i64 @llvm.vector.reduce.smax.v2i64(<2 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4 = call i64 @llvm.vector.reduce.smax.v4i64(<4 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8 = call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V16 = call i64 @llvm.vector.reduce.smax.v16i64(<16 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2 = call i64 @llvm.vector.reduce.smax.v2i64(<2 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i64 @llvm.vector.reduce.smax.v4i64(<4 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V8 = call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 82 for instruction: %V16 = call i64 @llvm.vector.reduce.smax.v16i64(<16 x i64> undef)
 ; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i64'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.smax.v1i64(<1 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i64 @llvm.vector.reduce.smax.v2i64(<2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4 = call i64 @llvm.vector.reduce.smax.v4i64(<4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8 = call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V16 = call i64 @llvm.vector.reduce.smax.v16i64(<16 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2 = call i64 @llvm.vector.reduce.smax.v2i64(<2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4 = call i64 @llvm.vector.reduce.smax.v4i64(<4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8 = call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V16 = call i64 @llvm.vector.reduce.smax.v16i64(<16 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.smax.v1i64(<1 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i64 @llvm.vector.reduce.smax.v2i64(<2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4 = call i64 @llvm.vector.reduce.smax.v4i64(<4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8 = call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V16 = call i64 @llvm.vector.reduce.smax.v16i64(<16 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2 = call i64 @llvm.vector.reduce.smax.v2i64(<2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V4 = call i64 @llvm.vector.reduce.smax.v4i64(<4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8 = call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V16 = call i64 @llvm.vector.reduce.smax.v16i64(<16 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.smax.v1i64(<1 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i64 @llvm.vector.reduce.smax.v2i64(<2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4 = call i64 @llvm.vector.reduce.smax.v4i64(<4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8 = call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16 = call i64 @llvm.vector.reduce.smax.v16i64(<16 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2 = call i64 @llvm.vector.reduce.smax.v2i64(<2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V4 = call i64 @llvm.vector.reduce.smax.v4i64(<4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V8 = call i64 @llvm.vector.reduce.smax.v8i64(<8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16 = call i64 @llvm.vector.reduce.smax.v16i64(<16 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512-LABEL: 'reduce_i64'
diff --git a/llvm/test/Analysis/CostModel/X86/reduce-smin.ll b/llvm/test/Analysis/CostModel/X86/reduce-smin.ll
index 98f1c589486e8..c82dac371c810 100644
--- a/llvm/test/Analysis/CostModel/X86/reduce-smin.ll
+++ b/llvm/test/Analysis/CostModel/X86/reduce-smin.ll
@@ -28,34 +28,34 @@ define i32 @reduce_i64(i32 %arg) {
 ;
 ; SSE41-LABEL: 'reduce_i64'
 ; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.smin.v1i64(<1 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2 = call i64 @llvm.vector.reduce.smin.v2i64(<2 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4 = call i64 @llvm.vector.reduce.smin.v4i64(<4 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8 = call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V16 = call i64 @llvm.vector.reduce.smin.v16i64(<16 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2 = call i64 @llvm.vector.reduce.smin.v2i64(<2 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i64 @llvm.vector.reduce.smin.v4i64(<4 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V8 = call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 82 for instruction: %V16 = call i64 @llvm.vector.reduce.smin.v16i64(<16 x i64> undef)
 ; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i64'
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.smin.v1i64(<1 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i64 @llvm.vector.reduce.smin.v2i64(<2 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V4 = call i64 @llvm.vector.reduce.smin.v4i64(<4 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V8 = call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> undef)
-; SSE42-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V16 = call i64 @llvm.vector.reduce.smin.v16i64(<16 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2 = call i64 @llvm.vector.reduce.smin.v2i64(<2 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V4 = call i64 @llvm.vector.reduce.smin.v4i64(<4 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 26 for instruction: %V8 = call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> undef)
+; SSE42-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V16 = call i64 @llvm.vector.reduce.smin.v16i64(<16 x i64> undef)
 ; SSE42-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'reduce_i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.smin.v1i64(<1 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i64 @llvm.vector.reduce.smin.v2i64(<2 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4 = call i64 @llvm.vector.reduce.smin.v4i64(<4 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 23 for instruction: %V8 = call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> undef)
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V16 = call i64 @llvm.vector.reduce.smin.v16i64(<16 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2 = call i64 @llvm.vector.reduce.smin.v2i64(<2 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V4 = call i64 @llvm.vector.reduce.smin.v4i64(<4 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V8 = call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> undef)
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 45 for instruction: %V16 = call i64 @llvm.vector.reduce.smin.v16i64(<16 x i64> undef)
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'reduce_i64'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.smin.v1i64(<1 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i64 @llvm.vector.reduce.smin.v2i64(<2 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V4 = call i64 @llvm.vector.reduce.smin.v4i64(<4 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V8 = call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> undef)
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 31 for instruction: %V16 = call i64 @llvm.vector.reduce.smin.v16i64(<16 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2 = call i64 @llvm.vector.reduce.smin.v2i64(<2 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %V4 = call i64 @llvm.vector.reduce.smin.v4i64(<4 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 21 for instruction: %V8 = call i64 @llvm.vector.reduce.smin.v8i64(<8 x i64> undef)
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %V16 = call i64 @llvm.vector.reduce.smin.v16i64(<16 x i64> undef)
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX512-LABEL: 'reduce_i64'
diff --git a/llvm/test/Analysis/CostModel/X86/reduce-umax.ll b/llvm/test/Analysis/CostModel/X86/reduce-umax.ll
index 67abb39d6be81..610beb4352ef5 100644
--- a/llvm/test/Analysis/CostModel/X86/reduce-umax.ll
+++ b/llvm/test/Analysis/CostModel/X86/reduce-umax.ll
@@ -28,10 +28,10 @@ define i32 @reduce_i64(i32 %arg) {
 ;
 ; SSE41-LABEL: 'reduce_i64'
 ; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.umax.v1i64(<1 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2 = call i64 @llvm.vector.reduce.umax.v2i64(<2 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4 = call i64 @llvm.vector.reduce.umax.v4i64(<4 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8 = call i64 @llvm.vector.reduce.umax.v8i64(<8 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V16 = call i64 @llvm.vector.reduce.umax.v16i64(<16 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2 = call i64 @llvm.vector.reduce.umax.v2i64(<2 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i64 @llvm.vector.reduce.umax.v4i64(<4 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V8 = call i64 @llvm.vector.reduce.umax.v8i64(<8 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 82 for instruction: %V16 = call i64 @llvm.vector.reduce.umax.v16i64(<16 x i64> undef)
 ; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i64'
diff --git a/llvm/test/Analysis/CostModel/X86/reduce-umin.ll b/llvm/test/Analysis/CostModel/X86/reduce-umin.ll
index 6a016b051988c..6215b510fb766 100644
--- a/llvm/test/Analysis/CostModel/X86/reduce-umin.ll
+++ b/llvm/test/Analysis/CostModel/X86/reduce-umin.ll
@@ -28,10 +28,10 @@ define i32 @reduce_i64(i32 %arg) {
 ;
 ; SSE41-LABEL: 'reduce_i64'
 ; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: %V1 = call i64 @llvm.vector.reduce.umin.v1i64(<1 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 11 for instruction: %V2 = call i64 @llvm.vector.reduce.umin.v2i64(<2 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4 = call i64 @llvm.vector.reduce.umin.v4i64(<4 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V8 = call i64 @llvm.vector.reduce.umin.v8i64(<8 x i64> undef)
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 74 for instruction: %V16 = call i64 @llvm.vector.reduce.umin.v16i64(<16 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V2 = call i64 @llvm.vector.reduce.umin.v2i64(<2 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i64 @llvm.vector.reduce.umin.v4i64(<4 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V8 = call i64 @llvm.vector.reduce.umin.v8i64(<8 x i64> undef)
+; SSE41-NEXT:  Cost Model: Found an estimated cost of 82 for instruction: %V16 = call i64 @llvm.vector.reduce.umin.v16i64(<16 x i64> undef)
 ; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; SSE42-LABEL: 'reduce_i64'
diff --git a/llvm/test/Analysis/CostModel/X86/vselect-cost.ll b/llvm/test/Analysis/CostModel/X86/vselect-cost.ll
index 9c475f9d6fc5e..e94bf31b5c6df 100644
--- a/llvm/test/Analysis/CostModel/X86/vselect-cost.ll
+++ b/llvm/test/Analysis/CostModel/X86/vselect-cost.ll
@@ -1,86 +1,67 @@
 ; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
-; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -mtriple=x86_64-unknown-linux-gnu -mattr=+sse2 | FileCheck %s -check-prefixes=SSE2
-; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -mtriple=x86_64-unknown-linux-gnu -mattr=+sse4.1 | FileCheck %s -check-prefixes=SSE41
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -mtriple=x86_64-unknown-linux-gnu -mattr=+sse2 | FileCheck %s -check-prefixes=SSE
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -mtriple=x86_64-unknown-linux-gnu -mattr=+sse4.1 | FileCheck %s -check-prefixes=SSE
 ; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -mtriple=x86_64-unknown-linux-gnu -mattr=+avx | FileCheck %s -check-prefixes=AVX,AVX1
 ; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -mtriple=x86_64-unknown-linux-gnu -mattr=+avx2 | FileCheck %s -check-prefixes=AVX,AVX2
-; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512vl  | FileCheck %s -check-prefixes=AVX,AVX512,AVX512F
-; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512bw,+avx512vl | FileCheck %s -check-prefixes=AVX,AVX512,AVX512BW
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f,+avx512vl  | FileCheck %s -check-prefixes=AVX512,AVX512F
+; RUN: opt < %s -passes="print<cost-model>" 2>&1 -disable-output -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512bw,+avx512vl | FileCheck %s -check-prefixes=AVX512,AVX512BW
 
 ; Verify the cost of vector select instructions.
 
 define i32 @test_select() {
-; SSE2-LABEL: 'test_select'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = select i1 undef, i64 undef, i64 undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = select <2 x i1> undef, <2 x i64> undef, <2 x i64> undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I64 = select <4 x i1> undef, <4 x i64> undef, <4 x i64> undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I64 = select <8 x i1> undef, <8 x i64> undef, <8 x i64> undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = select i1 undef, i32 undef, i32 undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = select <4 x i1> undef, <4 x i32> undef, <4 x i32> undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = select <8 x i1> undef, <8 x i32> undef, <8 x i32> undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I32 = select <16 x i1> undef, <16 x i32> undef, <16 x i32> undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = select i1 undef, i16 undef, i16 undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = select <8 x i1> undef, <8 x i16> undef, <8 x i16> undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = select <16 x i1> undef, <16 x i16> undef, <16 x i16> undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I16 = select <32 x i1> undef, <32 x i16> undef, <32 x i16> undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = select i1 undef, i8 undef, i8 undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = select <16 x i1> undef, <16 x i8> undef, <16 x i8> undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I8 = select <32 x i1> undef, <32 x i8> undef, <32 x i8> undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V64I8 = select <64 x i1> undef, <64 x i8> undef, <64 x i8> undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
-;
-; SSE41-LABEL: 'test_select'
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = select i1 undef, i64 undef, i64 undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = select <2 x i1> undef, <2 x i64> undef, <2 x i64> undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = select <4 x i1> undef, <4 x i64> undef, <4 x i64> undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = select <8 x i1> undef, <8 x i64> undef, <8 x i64> undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = select i1 undef, i32 undef, i32 undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = select <4 x i1> undef, <4 x i32> undef, <4 x i32> undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = select <8 x i1> undef, <8 x i32> undef, <8 x i32> undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = select <16 x i1> undef, <16 x i32> undef, <16 x i32> undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = select i1 undef, i16 undef, i16 undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = select <8 x i1> undef, <8 x i16> undef, <8 x i16> undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = select <16 x i1> undef, <16 x i16> undef, <16 x i16> undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I16 = select <32 x i1> undef, <32 x i16> undef, <32 x i16> undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = select i1 undef, i8 undef, i8 undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = select <16 x i1> undef, <16 x i8> undef, <16 x i8> undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32I8 = select <32 x i1> undef, <32 x i8> undef, <32 x i8> undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = select <64 x i1> undef, <64 x i8> undef, <64 x i8> undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+; SSE-LABEL: 'test_select'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = select i1 undef, i64 undef, i64 undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = select <2 x i1> undef, <2 x i64> undef, <2 x i64> undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4I64 = select <4 x i1> undef, <4 x i64> undef, <4 x i64> undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8I64 = select <8 x i1> undef, <8 x i64> undef, <8 x i64> undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = select i1 undef, i32 undef, i32 undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = select <4 x i1> undef, <4 x i32> undef, <4 x i32> undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I32 = select <8 x i1> undef, <8 x i32> undef, <8 x i32> undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16I32 = select <16 x i1> undef, <16 x i32> undef, <16 x i32> undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = select i1 undef, i16 undef, i16 undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = select <8 x i1> undef, <8 x i16> undef, <8 x i16> undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I16 = select <16 x i1> undef, <16 x i16> undef, <16 x i16> undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V32I16 = select <32 x i1> undef, <32 x i16> undef, <32 x i16> undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = select i1 undef, i8 undef, i8 undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = select <16 x i1> undef, <16 x i8> undef, <16 x i8> undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I8 = select <32 x i1> undef, <32 x i8> undef, <32 x i8> undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V64I8 = select <64 x i1> undef, <64 x i8> undef, <64 x i8> undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'test_select'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = select i1 undef, i64 undef, i64 undef
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = select <2 x i1> undef, <2 x i64> undef, <2 x i64> undef
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = select <2 x i1> undef, <2 x i64> undef, <2 x i64> undef
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4I64 = select <4 x i1> undef, <4 x i64> undef, <4 x i64> undef
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I64 = select <8 x i1> undef, <8 x i64> undef, <8 x i64> undef
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = select i1 undef, i32 undef, i32 undef
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = select <4 x i1> undef, <4 x i32> undef, <4 x i32> undef
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = select <4 x i1> undef, <4 x i32> undef, <4 x i32> undef
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8I32 = select <8 x i1> undef, <8 x i32> undef, <8 x i32> undef
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16I32 = select <16 x i1> undef, <16 x i32> undef, <16 x i32> undef
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = select i1 undef, i16 undef, i16 undef
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = select <8 x i1> undef, <8 x i16> undef, <8 x i16> undef
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = select <8 x i1> undef, <8 x i16> undef, <8 x i16> undef
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V16I16 = select <16 x i1> undef, <16 x i16> undef, <16 x i16> undef
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V32I16 = select <32 x i1> undef, <32 x i16> undef, <32 x i16> undef
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = select i1 undef, i8 undef, i8 undef
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = select <16 x i1> undef, <16 x i8> undef, <16 x i8> undef
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = select <16 x i1> undef, <16 x i8> undef, <16 x i8> undef
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V32I8 = select <32 x i1> undef, <32 x i8> undef, <32 x i8> undef
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V64I8 = select <64 x i1> undef, <64 x i8> undef, <64 x i8> undef
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'test_select'
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I64 = select i1 undef, i64 undef, i64 undef
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2I64 = select <2 x i1> undef, <2 x i64> undef, <2 x i64> undef
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2I64 = select <2 x i1> undef, <2 x i64> undef, <2 x i64> undef
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I64 = select <4 x i1> undef, <4 x i64> undef, <4 x i64> undef
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8I64 = select <8 x i1> undef, <8 x i64> undef, <8 x i64> undef
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I32 = select i1 undef, i32 undef, i32 undef
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4I32 = select <4 x i1> undef, <4 x i32> undef, <4 x i32> undef
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4I32 = select <4 x i1> undef, <4 x i32> undef, <4 x i32> undef
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I32 = select <8 x i1> undef, <8 x i32> undef, <8 x i32> undef
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16I32 = select <16 x i1> undef, <16 x i32> undef, <16 x i32> undef
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I16 = select i1 undef, i16 undef, i16 undef
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V8I16 = select <8 x i1> undef, <8 x i16> undef, <8 x i16> undef
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8I16 = select <8 x i1> undef, <8 x i16> undef, <8 x i16> undef
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I16 = select <16 x i1> undef, <16 x i16> undef, <16 x i16> undef
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V32I16 = select <32 x i1> undef, <32 x i16> undef, <32 x i16> undef
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %I8 = select i1 undef, i8 undef, i8 undef
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V16I8 = select <16 x i1> undef, <16 x i8> undef, <16 x i8> undef
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V16I8 = select <16 x i1> undef, <16 x i8> undef, <16 x i8> undef
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V32I8 = select <32 x i1> undef, <32 x i8> undef, <32 x i8> undef
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V64I8 = select <64 x i1> undef, <64 x i8> undef, <64 x i8> undef
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
@@ -147,46 +128,35 @@ define i32 @test_select() {
 }
 
 define i32 @test_select_fp() {
-; SSE2-LABEL: 'test_select_fp'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = select i1 undef, double undef, double undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = select <2 x i1> undef, <2 x double> undef, <2 x double> undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = select <4 x i1> undef, <4 x double> undef, <4 x double> undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = select <8 x i1> undef, <8 x double> undef, <8 x double> undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = select i1 undef, float undef, float undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = select <4 x i1> undef, <4 x float> undef, <4 x float> undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = select <8 x i1> undef, <8 x float> undef, <8 x float> undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = select <16 x i1> undef, <16 x float> undef, <16 x float> undef
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
-;
-; SSE41-LABEL: 'test_select_fp'
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = select i1 undef, double undef, double undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = select <2 x i1> undef, <2 x double> undef, <2 x double> undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = select <4 x i1> undef, <4 x double> undef, <4 x double> undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = select <8 x i1> undef, <8 x double> undef, <8 x double> undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = select i1 undef, float undef, float undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = select <4 x i1> undef, <4 x float> undef, <4 x float> undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = select <8 x i1> undef, <8 x float> undef, <8 x float> undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = select <16 x i1> undef, <16 x float> undef, <16 x float> undef
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
+; SSE-LABEL: 'test_select_fp'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = select i1 undef, double undef, double undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = select <2 x i1> undef, <2 x double> undef, <2 x double> undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V4F64 = select <4 x i1> undef, <4 x double> undef, <4 x double> undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V8F64 = select <8 x i1> undef, <8 x double> undef, <8 x double> undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = select i1 undef, float undef, float undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = select <4 x i1> undef, <4 x float> undef, <4 x float> undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F32 = select <8 x i1> undef, <8 x float> undef, <8 x float> undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V16F32 = select <16 x i1> undef, <16 x float> undef, <16 x float> undef
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX1-LABEL: 'test_select_fp'
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = select i1 undef, double undef, double undef
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = select <2 x i1> undef, <2 x double> undef, <2 x double> undef
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = select i1 undef, double undef, double undef
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = select <2 x i1> undef, <2 x double> undef, <2 x double> undef
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V4F64 = select <4 x i1> undef, <4 x double> undef, <4 x double> undef
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8F64 = select <8 x i1> undef, <8 x double> undef, <8 x double> undef
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = select i1 undef, float undef, float undef
-; AVX1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = select <4 x i1> undef, <4 x float> undef, <4 x float> undef
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = select i1 undef, float undef, float undef
+; AVX1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = select <4 x i1> undef, <4 x float> undef, <4 x float> undef
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %V8F32 = select <8 x i1> undef, <8 x float> undef, <8 x float> undef
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16F32 = select <16 x i1> undef, <16 x float> undef, <16 x float> undef
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
 ;
 ; AVX2-LABEL: 'test_select_fp'
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F64 = select i1 undef, double undef, double undef
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = select <2 x i1> undef, <2 x double> undef, <2 x double> undef
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F64 = select i1 undef, double undef, double undef
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V2F64 = select <2 x i1> undef, <2 x double> undef, <2 x double> undef
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = select <4 x i1> undef, <4 x double> undef, <4 x double> undef
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = select <8 x i1> undef, <8 x double> undef, <8 x double> undef
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %F32 = select i1 undef, float undef, float undef
-; AVX2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = select <4 x i1> undef, <4 x float> undef, <4 x float> undef
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %F32 = select i1 undef, float undef, float undef
+; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V4F32 = select <4 x i1> undef, <4 x float> undef, <4 x float> undef
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = select <8 x i1> undef, <8 x float> undef, <8 x float> undef
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = select <16 x i1> undef, <16 x float> undef, <16 x float> undef
 ; AVX2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret i32 undef
@@ -219,85 +189,85 @@ define i32 @test_select_fp() {
 ; Integers of the same size should also use those instructions.
 
 define <2 x i64> @test_2i64(<2 x i64> %a, <2 x i64> %b) {
-; SSE2-LABEL: 'test_2i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %sel = select <2 x i1> <i1 true, i1 false>, <2 x i64> %a, <2 x i64> %b
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %sel
-;
-; SSE41-LABEL: 'test_2i64'
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %sel = select <2 x i1> <i1 true, i1 false>, <2 x i64> %a, <2 x i64> %b
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %sel
+; SSE-LABEL: 'test_2i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %sel = select <2 x i1> <i1 true, i1 false>, <2 x i64> %a, <2 x i64> %b
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %sel
 ;
 ; AVX-LABEL: 'test_2i64'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %sel = select <2 x i1> <i1 true, i1 false>, <2 x i64> %a, <2 x i64> %b
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %sel = select <2 x i1> <i1 true, i1 false>, <2 x i64> %a, <2 x i64> %b
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %sel
+;
+; AVX512-LABEL: 'test_2i64'
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %sel = select <2 x i1> <i1 true, i1 false>, <2 x i64> %a, <2 x i64> %b
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x i64> %sel
 ;
   %sel = select <2 x i1> <i1 true, i1 false>, <2 x i64> %a, <2 x i64> %b
   ret <2 x i64> %sel
 }
 
 define <2 x double> @test_2double(<2 x double> %a, <2 x double> %b) {
-; SSE2-LABEL: 'test_2double'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %sel = select <2 x i1> <i1 true, i1 false>, <2 x double> %a, <2 x double> %b
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x double> %sel
-;
-; SSE41-LABEL: 'test_2double'
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %sel = select <2 x i1> <i1 true, i1 false>, <2 x double> %a, <2 x double> %b
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x double> %sel
+; SSE-LABEL: 'test_2double'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %sel = select <2 x i1> <i1 true, i1 false>, <2 x double> %a, <2 x double> %b
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x double> %sel
 ;
 ; AVX-LABEL: 'test_2double'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %sel = select <2 x i1> <i1 true, i1 false>, <2 x double> %a, <2 x double> %b
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %sel = select <2 x i1> <i1 true, i1 false>, <2 x double> %a, <2 x double> %b
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x double> %sel
+;
+; AVX512-LABEL: 'test_2double'
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %sel = select <2 x i1> <i1 true, i1 false>, <2 x double> %a, <2 x double> %b
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <2 x double> %sel
 ;
   %sel = select <2 x i1> <i1 true, i1 false>, <2 x double> %a, <2 x double> %b
   ret <2 x double> %sel
 }
 
 define <4 x i32> @test_4i32(<4 x i32> %a, <4 x i32> %b) {
-; SSE2-LABEL: 'test_4i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %sel = select <4 x i1> <i1 true, i1 false, i1 true, i1 false>, <4 x i32> %a, <4 x i32> %b
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %sel
-;
-; SSE41-LABEL: 'test_4i32'
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %sel = select <4 x i1> <i1 true, i1 false, i1 true, i1 false>, <4 x i32> %a, <4 x i32> %b
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %sel
+; SSE-LABEL: 'test_4i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %sel = select <4 x i1> <i1 true, i1 false, i1 true, i1 false>, <4 x i32> %a, <4 x i32> %b
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %sel
 ;
 ; AVX-LABEL: 'test_4i32'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %sel = select <4 x i1> <i1 true, i1 false, i1 true, i1 false>, <4 x i32> %a, <4 x i32> %b
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %sel = select <4 x i1> <i1 true, i1 false, i1 true, i1 false>, <4 x i32> %a, <4 x i32> %b
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %sel
+;
+; AVX512-LABEL: 'test_4i32'
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %sel = select <4 x i1> <i1 true, i1 false, i1 true, i1 false>, <4 x i32> %a, <4 x i32> %b
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i32> %sel
 ;
   %sel = select <4 x i1> <i1 true, i1 false, i1 true, i1 false>, <4 x i32> %a, <4 x i32> %b
   ret <4 x i32> %sel
 }
 
 define <4 x float> @test_4float(<4 x float> %a, <4 x float> %b) {
-; SSE2-LABEL: 'test_4float'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %sel = select <4 x i1> <i1 true, i1 false, i1 true, i1 true>, <4 x float> %a, <4 x float> %b
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %sel
-;
-; SSE41-LABEL: 'test_4float'
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %sel = select <4 x i1> <i1 true, i1 false, i1 true, i1 true>, <4 x float> %a, <4 x float> %b
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %sel
+; SSE-LABEL: 'test_4float'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %sel = select <4 x i1> <i1 true, i1 false, i1 true, i1 true>, <4 x float> %a, <4 x float> %b
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %sel
 ;
 ; AVX-LABEL: 'test_4float'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %sel = select <4 x i1> <i1 true, i1 false, i1 true, i1 true>, <4 x float> %a, <4 x float> %b
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %sel = select <4 x i1> <i1 true, i1 false, i1 true, i1 true>, <4 x float> %a, <4 x float> %b
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %sel
+;
+; AVX512-LABEL: 'test_4float'
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %sel = select <4 x i1> <i1 true, i1 false, i1 true, i1 true>, <4 x float> %a, <4 x float> %b
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %sel
 ;
   %sel = select <4 x i1> <i1 true, i1 false, i1 true, i1 true>, <4 x float> %a, <4 x float> %b
   ret <4 x float> %sel
 }
 
 define <16 x i8> @test_16i8(<16 x i8> %a, <16 x i8> %b) {
-; SSE2-LABEL: 'test_16i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %sel = select <16 x i1> <i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true>, <16 x i8> %a, <16 x i8> %b
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %sel
-;
-; SSE41-LABEL: 'test_16i8'
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %sel = select <16 x i1> <i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true>, <16 x i8> %a, <16 x i8> %b
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %sel
+; SSE-LABEL: 'test_16i8'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %sel = select <16 x i1> <i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true>, <16 x i8> %a, <16 x i8> %b
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %sel
 ;
 ; AVX-LABEL: 'test_16i8'
-; AVX-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %sel = select <16 x i1> <i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true>, <16 x i8> %a, <16 x i8> %b
+; AVX-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %sel = select <16 x i1> <i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true>, <16 x i8> %a, <16 x i8> %b
 ; AVX-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %sel
+;
+; AVX512-LABEL: 'test_16i8'
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %sel = select <16 x i1> <i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true>, <16 x i8> %a, <16 x i8> %b
+; AVX512-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i8> %sel
 ;
   %sel = select <16 x i1> <i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true>, <16 x i8> %a, <16 x i8> %b
   ret <16 x i8> %sel
@@ -307,13 +277,9 @@ define <16 x i8> @test_16i8(<16 x i8> %a, <16 x i8> %b) {
 ; Integers of the same size should also use those instructions.
 
 define <4 x i64> @test_4i64(<4 x i64> %a, <4 x i64> %b) {
-; SSE2-LABEL: 'test_4i64'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %sel = select <4 x i1> <i1 true, i1 false, i1 false, i1 true>, <4 x i64> %a, <4 x i64> %b
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %sel
-;
-; SSE41-LABEL: 'test_4i64'
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %sel = select <4 x i1> <i1 true, i1 false, i1 false, i1 true>, <4 x i64> %a, <4 x i64> %b
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %sel
+; SSE-LABEL: 'test_4i64'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %sel = select <4 x i1> <i1 true, i1 false, i1 false, i1 true>, <4 x i64> %a, <4 x i64> %b
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x i64> %sel
 ;
 ; AVX1-LABEL: 'test_4i64'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %sel = select <4 x i1> <i1 true, i1 false, i1 false, i1 true>, <4 x i64> %a, <4 x i64> %b
@@ -332,13 +298,9 @@ define <4 x i64> @test_4i64(<4 x i64> %a, <4 x i64> %b) {
 }
 
 define <4 x double> @test_4double(<4 x double> %a, <4 x double> %b) {
-; SSE2-LABEL: 'test_4double'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %sel = select <4 x i1> <i1 true, i1 false, i1 true, i1 false>, <4 x double> %a, <4 x double> %b
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x double> %sel
-;
-; SSE41-LABEL: 'test_4double'
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %sel = select <4 x i1> <i1 true, i1 false, i1 true, i1 false>, <4 x double> %a, <4 x double> %b
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x double> %sel
+; SSE-LABEL: 'test_4double'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %sel = select <4 x i1> <i1 true, i1 false, i1 true, i1 false>, <4 x double> %a, <4 x double> %b
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <4 x double> %sel
 ;
 ; AVX1-LABEL: 'test_4double'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %sel = select <4 x i1> <i1 true, i1 false, i1 true, i1 false>, <4 x double> %a, <4 x double> %b
@@ -357,13 +319,9 @@ define <4 x double> @test_4double(<4 x double> %a, <4 x double> %b) {
 }
 
 define <8 x i32> @test_8i32(<8 x i32> %a, <8 x i32> %b) {
-; SSE2-LABEL: 'test_8i32'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %sel = select <8 x i1> <i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 false>, <8 x i32> %a, <8 x i32> %b
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %sel
-;
-; SSE41-LABEL: 'test_8i32'
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %sel = select <8 x i1> <i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 false>, <8 x i32> %a, <8 x i32> %b
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %sel
+; SSE-LABEL: 'test_8i32'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %sel = select <8 x i1> <i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 false>, <8 x i32> %a, <8 x i32> %b
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x i32> %sel
 ;
 ; AVX1-LABEL: 'test_8i32'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %sel = select <8 x i1> <i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 false>, <8 x i32> %a, <8 x i32> %b
@@ -382,13 +340,9 @@ define <8 x i32> @test_8i32(<8 x i32> %a, <8 x i32> %b) {
 }
 
 define <8 x float> @test_8float(<8 x float> %a, <8 x float> %b) {
-; SSE2-LABEL: 'test_8float'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %sel = select <8 x i1> <i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 false, i1 false>, <8 x float> %a, <8 x float> %b
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x float> %sel
-;
-; SSE41-LABEL: 'test_8float'
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %sel = select <8 x i1> <i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 false, i1 false>, <8 x float> %a, <8 x float> %b
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x float> %sel
+; SSE-LABEL: 'test_8float'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %sel = select <8 x i1> <i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 false, i1 false>, <8 x float> %a, <8 x float> %b
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <8 x float> %sel
 ;
 ; AVX1-LABEL: 'test_8float'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %sel = select <8 x i1> <i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 false, i1 false>, <8 x float> %a, <8 x float> %b
@@ -407,13 +361,9 @@ define <8 x float> @test_8float(<8 x float> %a, <8 x float> %b) {
 }
 
 define <16 x i16> @test_16i16(<16 x i16> %a, <16 x i16> %b) {
-; SSE2-LABEL: 'test_16i16'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %sel = select <16 x i1> <i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 false, i1 false>, <16 x i16> %a, <16 x i16> %b
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %sel
-;
-; SSE41-LABEL: 'test_16i16'
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %sel = select <16 x i1> <i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 false, i1 false>, <16 x i16> %a, <16 x i16> %b
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %sel
+; SSE-LABEL: 'test_16i16'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %sel = select <16 x i1> <i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 false, i1 false>, <16 x i16> %a, <16 x i16> %b
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <16 x i16> %sel
 ;
 ; AVX1-LABEL: 'test_16i16'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %sel = select <16 x i1> <i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 false, i1 false>, <16 x i16> %a, <16 x i16> %b
@@ -432,13 +382,9 @@ define <16 x i16> @test_16i16(<16 x i16> %a, <16 x i16> %b) {
 }
 
 define <32 x i8> @test_32i8(<32 x i8> %a, <32 x i8> %b) {
-; SSE2-LABEL: 'test_32i8'
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %sel = select <32 x i1> <i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true>, <32 x i8> %a, <32 x i8> %b
-; SSE2-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %sel
-;
-; SSE41-LABEL: 'test_32i8'
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %sel = select <32 x i1> <i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true>, <32 x i8> %a, <32 x i8> %b
-; SSE41-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %sel
+; SSE-LABEL: 'test_32i8'
+; SSE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %sel = select <32 x i1> <i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true>, <32 x i8> %a, <32 x i8> %b
+; SSE-NEXT:  Cost Model: Found an estimated cost of 0 for instruction: ret <32 x i8> %sel
 ;
 ; AVX1-LABEL: 'test_32i8'
 ; AVX1-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %sel = select <32 x i1> <i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true, i1 true, i1 false, i1 true, i1 true>, <32 x i8> %a, <32 x i8> %b
diff --git a/llvm/test/Transforms/LoopVectorize/X86/vector-scalar-select-cost.ll b/llvm/test/Transforms/LoopVectorize/X86/vector-scalar-select-cost.ll
index 4faa9f205cd45..f69b88e804b34 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/vector-scalar-select-cost.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/vector-scalar-select-cost.ll
@@ -22,8 +22,9 @@ define void @scalarselect(i1 %cond) {
   %6 = add nsw i32 %5, %3
   %7 = getelementptr inbounds [2048 x i32], [2048 x i32]* @a, i64 0, i64 %indvars.iv
 
-; A scalar select has a cost of 1 on core2
-; CHECK: cost of 1 for VF 2 {{.*}}  select i1 %cond, i32 %6, i32 0
+; CHECK: cost of 1 for VF 1 {{.*}}  select i1 %cond, i32 %6, i32 0
+; CHECK: cost of 2 for VF 2 {{.*}}  select i1 %cond, i32 %6, i32 0
+; CHECK: cost of 2 for VF 4 {{.*}}  select i1 %cond, i32 %6, i32 0
 
   %sel = select i1 %cond, i32 %6, i32 zeroinitializer
   store i32 %sel, i32* %7, align 4
@@ -50,8 +51,9 @@ define void @vectorselect(i1 %cond) {
   %7 = getelementptr inbounds [2048 x i32], [2048 x i32]* @a, i64 0, i64 %indvars.iv
   %8 = icmp ult i64 %indvars.iv, 8
 
-; A vector select has a cost of 1 on core2
-; CHECK: cost of 1 for VF 2 {{.*}}  select i1 %8, i32 %6, i32 0
+; CHECK: cost of 1 for VF 1 {{.*}}  select i1 %8, i32 %6, i32 0
+; CHECK: cost of 2 for VF 2 {{.*}}  select i1 %8, i32 %6, i32 0
+; CHECK: cost of 2 for VF 4 {{.*}}  select i1 %8, i32 %6, i32 0
 
   %sel = select i1 %8, i32 %6, i32 zeroinitializer
   store i32 %sel, i32* %7, align 4