[RISCV][TTI] Adjust costing in getPartialReductionCost for zvqdotq #141430
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Two changes: 1) Handle fixed vector cases now that 77a3f8 has landed. 2) Fix a mistake in the original costing - the VF passed in is the input VF, not the output VF. Given that we should be costing the accumulator type with VF/4. Note that (2) does not cause any visibile test differences as the vectorizer (outside of maximize-bandwidth mode) does not consider wide enough VF for the costing difference to matter.
|
@llvm/pr-subscribers-backend-risc-v @llvm/pr-subscribers-llvm-transforms Author: Philip Reames (preames) ChangesTwo changes:
Note that (2) does not cause any visible test differences as the vectorizer (outside of maximize-bandwidth mode) does not consider wide enough VF for the costing difference to matter. Full diff: https://github.com/llvm/llvm-project/pull/141430.diff 2 Files Affected:
diff --git a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
index d54ad63404578..860d787111ce4 100644
--- a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
+++ b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
@@ -306,10 +306,10 @@ InstructionCost RISCVTTIImpl::getPartialReductionCost(
Opcode != Instruction::Add || !BinOp || *BinOp != Instruction::Mul ||
InputTypeA != InputTypeB || !InputTypeA->isIntegerTy(8) ||
OpAExtend != OpBExtend || !AccumType->isIntegerTy(32) ||
- !VF.isKnownMultipleOf(4) || !VF.isScalable())
+ !VF.isKnownMultipleOf(4))
return InstructionCost::getInvalid();
- Type *Tp = VectorType::get(AccumType, VF);
+ Type *Tp = VectorType::get(AccumType, VF.divideCoefficientBy(4));
std::pair<InstructionCost, MVT> LT = getTypeLegalizationCost(Tp);
// Note: Asuming all vqdot* variants are equal cost
// TODO: Thread CostKind through this API
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/partial-reduce-dot-product.ll b/llvm/test/Transforms/LoopVectorize/RISCV/partial-reduce-dot-product.ll
index 23534143ed3a9..847c4ba0bebfc 100644
--- a/llvm/test/Transforms/LoopVectorize/RISCV/partial-reduce-dot-product.ll
+++ b/llvm/test/Transforms/LoopVectorize/RISCV/partial-reduce-dot-product.ll
@@ -1,6 +1,8 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --filter-out-after "^scalar.ph:" --version 4
; RUN: opt -passes=loop-vectorize -mattr=+v -S < %s | FileCheck %s --check-prefixes=CHECK,V
; RUN: opt -passes=loop-vectorize -mattr=+v,+experimental-zvqdotq -S < %s | FileCheck %s --check-prefixes=CHECK,ZVQDOTQ
+; RUN: opt -passes=loop-vectorize -mattr=+v -scalable-vectorization=off -S < %s | FileCheck %s --check-prefixes=FIXED,FIXED-V
+; RUN: opt -passes=loop-vectorize -mattr=+v,+experimental-zvqdotq -scalable-vectorization=off -S < %s | FileCheck %s --check-prefixes=FIXED,FIXED-ZVQDOTQ
target triple = "riscv64-none-unknown-elf"
@@ -79,6 +81,80 @@ define i32 @vqdot(ptr %a, ptr %b) #0 {
; ZVQDOTQ-NEXT: br i1 [[CMP_N]], label [[FOR_EXIT:%.*]], label [[SCALAR_PH]]
; ZVQDOTQ: scalar.ph:
;
+; FIXED-V-LABEL: define i32 @vqdot(
+; FIXED-V-SAME: ptr [[A:%.*]], ptr [[B:%.*]]) #[[ATTR0:[0-9]+]] {
+; FIXED-V-NEXT: entry:
+; FIXED-V-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; FIXED-V: vector.ph:
+; FIXED-V-NEXT: br label [[VECTOR_BODY:%.*]]
+; FIXED-V: vector.body:
+; FIXED-V-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; FIXED-V-NEXT: [[VEC_PHI:%.*]] = phi <8 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP12:%.*]], [[VECTOR_BODY]] ]
+; FIXED-V-NEXT: [[VEC_PHI1:%.*]] = phi <8 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP13:%.*]], [[VECTOR_BODY]] ]
+; FIXED-V-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[A]], i64 [[INDEX]]
+; FIXED-V-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[TMP0]], i32 0
+; FIXED-V-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[TMP0]], i32 8
+; FIXED-V-NEXT: [[WIDE_LOAD:%.*]] = load <8 x i8>, ptr [[TMP1]], align 1
+; FIXED-V-NEXT: [[WIDE_LOAD2:%.*]] = load <8 x i8>, ptr [[TMP2]], align 1
+; FIXED-V-NEXT: [[TMP3:%.*]] = sext <8 x i8> [[WIDE_LOAD]] to <8 x i32>
+; FIXED-V-NEXT: [[TMP4:%.*]] = sext <8 x i8> [[WIDE_LOAD2]] to <8 x i32>
+; FIXED-V-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[B]], i64 [[INDEX]]
+; FIXED-V-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[TMP5]], i32 0
+; FIXED-V-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[TMP5]], i32 8
+; FIXED-V-NEXT: [[WIDE_LOAD3:%.*]] = load <8 x i8>, ptr [[TMP6]], align 1
+; FIXED-V-NEXT: [[WIDE_LOAD4:%.*]] = load <8 x i8>, ptr [[TMP7]], align 1
+; FIXED-V-NEXT: [[TMP8:%.*]] = sext <8 x i8> [[WIDE_LOAD3]] to <8 x i32>
+; FIXED-V-NEXT: [[TMP9:%.*]] = sext <8 x i8> [[WIDE_LOAD4]] to <8 x i32>
+; FIXED-V-NEXT: [[TMP10:%.*]] = mul <8 x i32> [[TMP8]], [[TMP3]]
+; FIXED-V-NEXT: [[TMP11:%.*]] = mul <8 x i32> [[TMP9]], [[TMP4]]
+; FIXED-V-NEXT: [[TMP12]] = add <8 x i32> [[TMP10]], [[VEC_PHI]]
+; FIXED-V-NEXT: [[TMP13]] = add <8 x i32> [[TMP11]], [[VEC_PHI1]]
+; FIXED-V-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
+; FIXED-V-NEXT: [[TMP14:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
+; FIXED-V-NEXT: br i1 [[TMP14]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; FIXED-V: middle.block:
+; FIXED-V-NEXT: [[BIN_RDX:%.*]] = add <8 x i32> [[TMP13]], [[TMP12]]
+; FIXED-V-NEXT: [[TMP15:%.*]] = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> [[BIN_RDX]])
+; FIXED-V-NEXT: br i1 true, label [[FOR_EXIT:%.*]], label [[SCALAR_PH]]
+; FIXED-V: scalar.ph:
+;
+; FIXED-ZVQDOTQ-LABEL: define i32 @vqdot(
+; FIXED-ZVQDOTQ-SAME: ptr [[A:%.*]], ptr [[B:%.*]]) #[[ATTR0:[0-9]+]] {
+; FIXED-ZVQDOTQ-NEXT: entry:
+; FIXED-ZVQDOTQ-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; FIXED-ZVQDOTQ: vector.ph:
+; FIXED-ZVQDOTQ-NEXT: br label [[VECTOR_BODY:%.*]]
+; FIXED-ZVQDOTQ: vector.body:
+; FIXED-ZVQDOTQ-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; FIXED-ZVQDOTQ-NEXT: [[VEC_PHI:%.*]] = phi <2 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE:%.*]], [[VECTOR_BODY]] ]
+; FIXED-ZVQDOTQ-NEXT: [[VEC_PHI1:%.*]] = phi <2 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE5:%.*]], [[VECTOR_BODY]] ]
+; FIXED-ZVQDOTQ-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[A]], i64 [[INDEX]]
+; FIXED-ZVQDOTQ-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[TMP0]], i32 0
+; FIXED-ZVQDOTQ-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[TMP0]], i32 8
+; FIXED-ZVQDOTQ-NEXT: [[WIDE_LOAD:%.*]] = load <8 x i8>, ptr [[TMP1]], align 1
+; FIXED-ZVQDOTQ-NEXT: [[WIDE_LOAD2:%.*]] = load <8 x i8>, ptr [[TMP2]], align 1
+; FIXED-ZVQDOTQ-NEXT: [[TMP3:%.*]] = sext <8 x i8> [[WIDE_LOAD]] to <8 x i32>
+; FIXED-ZVQDOTQ-NEXT: [[TMP4:%.*]] = sext <8 x i8> [[WIDE_LOAD2]] to <8 x i32>
+; FIXED-ZVQDOTQ-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[B]], i64 [[INDEX]]
+; FIXED-ZVQDOTQ-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[TMP5]], i32 0
+; FIXED-ZVQDOTQ-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[TMP5]], i32 8
+; FIXED-ZVQDOTQ-NEXT: [[WIDE_LOAD3:%.*]] = load <8 x i8>, ptr [[TMP6]], align 1
+; FIXED-ZVQDOTQ-NEXT: [[WIDE_LOAD4:%.*]] = load <8 x i8>, ptr [[TMP7]], align 1
+; FIXED-ZVQDOTQ-NEXT: [[TMP8:%.*]] = sext <8 x i8> [[WIDE_LOAD3]] to <8 x i32>
+; FIXED-ZVQDOTQ-NEXT: [[TMP9:%.*]] = sext <8 x i8> [[WIDE_LOAD4]] to <8 x i32>
+; FIXED-ZVQDOTQ-NEXT: [[TMP10:%.*]] = mul <8 x i32> [[TMP8]], [[TMP3]]
+; FIXED-ZVQDOTQ-NEXT: [[TMP11:%.*]] = mul <8 x i32> [[TMP9]], [[TMP4]]
+; FIXED-ZVQDOTQ-NEXT: [[PARTIAL_REDUCE]] = call <2 x i32> @llvm.experimental.vector.partial.reduce.add.v2i32.v8i32(<2 x i32> [[VEC_PHI]], <8 x i32> [[TMP10]])
+; FIXED-ZVQDOTQ-NEXT: [[PARTIAL_REDUCE5]] = call <2 x i32> @llvm.experimental.vector.partial.reduce.add.v2i32.v8i32(<2 x i32> [[VEC_PHI1]], <8 x i32> [[TMP11]])
+; FIXED-ZVQDOTQ-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
+; FIXED-ZVQDOTQ-NEXT: [[TMP14:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
+; FIXED-ZVQDOTQ-NEXT: br i1 [[TMP14]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; FIXED-ZVQDOTQ: middle.block:
+; FIXED-ZVQDOTQ-NEXT: [[BIN_RDX:%.*]] = add <2 x i32> [[PARTIAL_REDUCE5]], [[PARTIAL_REDUCE]]
+; FIXED-ZVQDOTQ-NEXT: [[TMP13:%.*]] = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> [[BIN_RDX]])
+; FIXED-ZVQDOTQ-NEXT: br i1 true, label [[FOR_EXIT:%.*]], label [[SCALAR_PH]]
+; FIXED-ZVQDOTQ: scalar.ph:
+;
entry:
br label %for.body
@@ -177,6 +253,80 @@ define i32 @vqdotu(ptr %a, ptr %b) #0 {
; ZVQDOTQ-NEXT: br i1 [[CMP_N]], label [[FOR_EXIT:%.*]], label [[SCALAR_PH]]
; ZVQDOTQ: scalar.ph:
;
+; FIXED-V-LABEL: define i32 @vqdotu(
+; FIXED-V-SAME: ptr [[A:%.*]], ptr [[B:%.*]]) #[[ATTR0]] {
+; FIXED-V-NEXT: entry:
+; FIXED-V-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; FIXED-V: vector.ph:
+; FIXED-V-NEXT: br label [[VECTOR_BODY:%.*]]
+; FIXED-V: vector.body:
+; FIXED-V-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; FIXED-V-NEXT: [[VEC_PHI:%.*]] = phi <8 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP12:%.*]], [[VECTOR_BODY]] ]
+; FIXED-V-NEXT: [[VEC_PHI1:%.*]] = phi <8 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP13:%.*]], [[VECTOR_BODY]] ]
+; FIXED-V-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[A]], i64 [[INDEX]]
+; FIXED-V-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[TMP0]], i32 0
+; FIXED-V-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[TMP0]], i32 8
+; FIXED-V-NEXT: [[WIDE_LOAD:%.*]] = load <8 x i8>, ptr [[TMP1]], align 1
+; FIXED-V-NEXT: [[WIDE_LOAD2:%.*]] = load <8 x i8>, ptr [[TMP2]], align 1
+; FIXED-V-NEXT: [[TMP3:%.*]] = zext <8 x i8> [[WIDE_LOAD]] to <8 x i32>
+; FIXED-V-NEXT: [[TMP4:%.*]] = zext <8 x i8> [[WIDE_LOAD2]] to <8 x i32>
+; FIXED-V-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[B]], i64 [[INDEX]]
+; FIXED-V-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[TMP5]], i32 0
+; FIXED-V-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[TMP5]], i32 8
+; FIXED-V-NEXT: [[WIDE_LOAD3:%.*]] = load <8 x i8>, ptr [[TMP6]], align 1
+; FIXED-V-NEXT: [[WIDE_LOAD4:%.*]] = load <8 x i8>, ptr [[TMP7]], align 1
+; FIXED-V-NEXT: [[TMP8:%.*]] = zext <8 x i8> [[WIDE_LOAD3]] to <8 x i32>
+; FIXED-V-NEXT: [[TMP9:%.*]] = zext <8 x i8> [[WIDE_LOAD4]] to <8 x i32>
+; FIXED-V-NEXT: [[TMP10:%.*]] = mul <8 x i32> [[TMP8]], [[TMP3]]
+; FIXED-V-NEXT: [[TMP11:%.*]] = mul <8 x i32> [[TMP9]], [[TMP4]]
+; FIXED-V-NEXT: [[TMP12]] = add <8 x i32> [[TMP10]], [[VEC_PHI]]
+; FIXED-V-NEXT: [[TMP13]] = add <8 x i32> [[TMP11]], [[VEC_PHI1]]
+; FIXED-V-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
+; FIXED-V-NEXT: [[TMP14:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
+; FIXED-V-NEXT: br i1 [[TMP14]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
+; FIXED-V: middle.block:
+; FIXED-V-NEXT: [[BIN_RDX:%.*]] = add <8 x i32> [[TMP13]], [[TMP12]]
+; FIXED-V-NEXT: [[TMP15:%.*]] = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> [[BIN_RDX]])
+; FIXED-V-NEXT: br i1 true, label [[FOR_EXIT:%.*]], label [[SCALAR_PH]]
+; FIXED-V: scalar.ph:
+;
+; FIXED-ZVQDOTQ-LABEL: define i32 @vqdotu(
+; FIXED-ZVQDOTQ-SAME: ptr [[A:%.*]], ptr [[B:%.*]]) #[[ATTR0]] {
+; FIXED-ZVQDOTQ-NEXT: entry:
+; FIXED-ZVQDOTQ-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; FIXED-ZVQDOTQ: vector.ph:
+; FIXED-ZVQDOTQ-NEXT: br label [[VECTOR_BODY:%.*]]
+; FIXED-ZVQDOTQ: vector.body:
+; FIXED-ZVQDOTQ-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; FIXED-ZVQDOTQ-NEXT: [[VEC_PHI:%.*]] = phi <2 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE:%.*]], [[VECTOR_BODY]] ]
+; FIXED-ZVQDOTQ-NEXT: [[VEC_PHI1:%.*]] = phi <2 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE5:%.*]], [[VECTOR_BODY]] ]
+; FIXED-ZVQDOTQ-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[A]], i64 [[INDEX]]
+; FIXED-ZVQDOTQ-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[TMP0]], i32 0
+; FIXED-ZVQDOTQ-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[TMP0]], i32 8
+; FIXED-ZVQDOTQ-NEXT: [[WIDE_LOAD:%.*]] = load <8 x i8>, ptr [[TMP1]], align 1
+; FIXED-ZVQDOTQ-NEXT: [[WIDE_LOAD2:%.*]] = load <8 x i8>, ptr [[TMP2]], align 1
+; FIXED-ZVQDOTQ-NEXT: [[TMP3:%.*]] = zext <8 x i8> [[WIDE_LOAD]] to <8 x i32>
+; FIXED-ZVQDOTQ-NEXT: [[TMP4:%.*]] = zext <8 x i8> [[WIDE_LOAD2]] to <8 x i32>
+; FIXED-ZVQDOTQ-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[B]], i64 [[INDEX]]
+; FIXED-ZVQDOTQ-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[TMP5]], i32 0
+; FIXED-ZVQDOTQ-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[TMP5]], i32 8
+; FIXED-ZVQDOTQ-NEXT: [[WIDE_LOAD3:%.*]] = load <8 x i8>, ptr [[TMP6]], align 1
+; FIXED-ZVQDOTQ-NEXT: [[WIDE_LOAD4:%.*]] = load <8 x i8>, ptr [[TMP7]], align 1
+; FIXED-ZVQDOTQ-NEXT: [[TMP8:%.*]] = zext <8 x i8> [[WIDE_LOAD3]] to <8 x i32>
+; FIXED-ZVQDOTQ-NEXT: [[TMP9:%.*]] = zext <8 x i8> [[WIDE_LOAD4]] to <8 x i32>
+; FIXED-ZVQDOTQ-NEXT: [[TMP10:%.*]] = mul <8 x i32> [[TMP8]], [[TMP3]]
+; FIXED-ZVQDOTQ-NEXT: [[TMP11:%.*]] = mul <8 x i32> [[TMP9]], [[TMP4]]
+; FIXED-ZVQDOTQ-NEXT: [[PARTIAL_REDUCE]] = call <2 x i32> @llvm.experimental.vector.partial.reduce.add.v2i32.v8i32(<2 x i32> [[VEC_PHI]], <8 x i32> [[TMP10]])
+; FIXED-ZVQDOTQ-NEXT: [[PARTIAL_REDUCE5]] = call <2 x i32> @llvm.experimental.vector.partial.reduce.add.v2i32.v8i32(<2 x i32> [[VEC_PHI1]], <8 x i32> [[TMP11]])
+; FIXED-ZVQDOTQ-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
+; FIXED-ZVQDOTQ-NEXT: [[TMP14:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
+; FIXED-ZVQDOTQ-NEXT: br i1 [[TMP14]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
+; FIXED-ZVQDOTQ: middle.block:
+; FIXED-ZVQDOTQ-NEXT: [[BIN_RDX:%.*]] = add <2 x i32> [[PARTIAL_REDUCE5]], [[PARTIAL_REDUCE]]
+; FIXED-ZVQDOTQ-NEXT: [[TMP13:%.*]] = call i32 @llvm.vector.reduce.add.v2i32(<2 x i32> [[BIN_RDX]])
+; FIXED-ZVQDOTQ-NEXT: br i1 true, label [[FOR_EXIT:%.*]], label [[SCALAR_PH]]
+; FIXED-ZVQDOTQ: scalar.ph:
+;
entry:
br label %for.body
@@ -238,6 +388,43 @@ define i32 @vqdotsu(ptr %a, ptr %b) #0 {
; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_EXIT:%.*]], label [[SCALAR_PH]]
; CHECK: scalar.ph:
;
+; FIXED-LABEL: define i32 @vqdotsu(
+; FIXED-SAME: ptr [[A:%.*]], ptr [[B:%.*]]) #[[ATTR0:[0-9]+]] {
+; FIXED-NEXT: entry:
+; FIXED-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; FIXED: vector.ph:
+; FIXED-NEXT: br label [[VECTOR_BODY:%.*]]
+; FIXED: vector.body:
+; FIXED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; FIXED-NEXT: [[VEC_PHI:%.*]] = phi <8 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP12:%.*]], [[VECTOR_BODY]] ]
+; FIXED-NEXT: [[VEC_PHI1:%.*]] = phi <8 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP13:%.*]], [[VECTOR_BODY]] ]
+; FIXED-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[A]], i64 [[INDEX]]
+; FIXED-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[TMP0]], i32 0
+; FIXED-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[TMP0]], i32 8
+; FIXED-NEXT: [[WIDE_LOAD:%.*]] = load <8 x i8>, ptr [[TMP1]], align 1
+; FIXED-NEXT: [[WIDE_LOAD2:%.*]] = load <8 x i8>, ptr [[TMP2]], align 1
+; FIXED-NEXT: [[TMP3:%.*]] = zext <8 x i8> [[WIDE_LOAD]] to <8 x i32>
+; FIXED-NEXT: [[TMP4:%.*]] = zext <8 x i8> [[WIDE_LOAD2]] to <8 x i32>
+; FIXED-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[B]], i64 [[INDEX]]
+; FIXED-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[TMP5]], i32 0
+; FIXED-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[TMP5]], i32 8
+; FIXED-NEXT: [[WIDE_LOAD3:%.*]] = load <8 x i8>, ptr [[TMP6]], align 1
+; FIXED-NEXT: [[WIDE_LOAD4:%.*]] = load <8 x i8>, ptr [[TMP7]], align 1
+; FIXED-NEXT: [[TMP8:%.*]] = sext <8 x i8> [[WIDE_LOAD3]] to <8 x i32>
+; FIXED-NEXT: [[TMP9:%.*]] = sext <8 x i8> [[WIDE_LOAD4]] to <8 x i32>
+; FIXED-NEXT: [[TMP10:%.*]] = mul <8 x i32> [[TMP8]], [[TMP3]]
+; FIXED-NEXT: [[TMP11:%.*]] = mul <8 x i32> [[TMP9]], [[TMP4]]
+; FIXED-NEXT: [[TMP12]] = add <8 x i32> [[TMP10]], [[VEC_PHI]]
+; FIXED-NEXT: [[TMP13]] = add <8 x i32> [[TMP11]], [[VEC_PHI1]]
+; FIXED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
+; FIXED-NEXT: [[TMP14:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
+; FIXED-NEXT: br i1 [[TMP14]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
+; FIXED: middle.block:
+; FIXED-NEXT: [[BIN_RDX:%.*]] = add <8 x i32> [[TMP13]], [[TMP12]]
+; FIXED-NEXT: [[TMP15:%.*]] = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> [[BIN_RDX]])
+; FIXED-NEXT: br i1 true, label [[FOR_EXIT:%.*]], label [[SCALAR_PH]]
+; FIXED: scalar.ph:
+;
entry:
br label %for.body
@@ -298,6 +485,43 @@ define i32 @vqdotsu2(ptr %a, ptr %b) #0 {
; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_EXIT:%.*]], label [[SCALAR_PH]]
; CHECK: scalar.ph:
;
+; FIXED-LABEL: define i32 @vqdotsu2(
+; FIXED-SAME: ptr [[A:%.*]], ptr [[B:%.*]]) #[[ATTR0]] {
+; FIXED-NEXT: entry:
+; FIXED-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; FIXED: vector.ph:
+; FIXED-NEXT: br label [[VECTOR_BODY:%.*]]
+; FIXED: vector.body:
+; FIXED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; FIXED-NEXT: [[VEC_PHI:%.*]] = phi <8 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP12:%.*]], [[VECTOR_BODY]] ]
+; FIXED-NEXT: [[VEC_PHI1:%.*]] = phi <8 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP13:%.*]], [[VECTOR_BODY]] ]
+; FIXED-NEXT: [[TMP0:%.*]] = getelementptr i8, ptr [[A]], i64 [[INDEX]]
+; FIXED-NEXT: [[TMP1:%.*]] = getelementptr i8, ptr [[TMP0]], i32 0
+; FIXED-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[TMP0]], i32 8
+; FIXED-NEXT: [[WIDE_LOAD:%.*]] = load <8 x i8>, ptr [[TMP1]], align 1
+; FIXED-NEXT: [[WIDE_LOAD2:%.*]] = load <8 x i8>, ptr [[TMP2]], align 1
+; FIXED-NEXT: [[TMP3:%.*]] = sext <8 x i8> [[WIDE_LOAD]] to <8 x i32>
+; FIXED-NEXT: [[TMP4:%.*]] = sext <8 x i8> [[WIDE_LOAD2]] to <8 x i32>
+; FIXED-NEXT: [[TMP5:%.*]] = getelementptr i8, ptr [[B]], i64 [[INDEX]]
+; FIXED-NEXT: [[TMP6:%.*]] = getelementptr i8, ptr [[TMP5]], i32 0
+; FIXED-NEXT: [[TMP7:%.*]] = getelementptr i8, ptr [[TMP5]], i32 8
+; FIXED-NEXT: [[WIDE_LOAD3:%.*]] = load <8 x i8>, ptr [[TMP6]], align 1
+; FIXED-NEXT: [[WIDE_LOAD4:%.*]] = load <8 x i8>, ptr [[TMP7]], align 1
+; FIXED-NEXT: [[TMP8:%.*]] = zext <8 x i8> [[WIDE_LOAD3]] to <8 x i32>
+; FIXED-NEXT: [[TMP9:%.*]] = zext <8 x i8> [[WIDE_LOAD4]] to <8 x i32>
+; FIXED-NEXT: [[TMP10:%.*]] = mul <8 x i32> [[TMP8]], [[TMP3]]
+; FIXED-NEXT: [[TMP11:%.*]] = mul <8 x i32> [[TMP9]], [[TMP4]]
+; FIXED-NEXT: [[TMP12]] = add <8 x i32> [[TMP10]], [[VEC_PHI]]
+; FIXED-NEXT: [[TMP13]] = add <8 x i32> [[TMP11]], [[VEC_PHI1]]
+; FIXED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
+; FIXED-NEXT: [[TMP14:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1024
+; FIXED-NEXT: br i1 [[TMP14]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
+; FIXED: middle.block:
+; FIXED-NEXT: [[BIN_RDX:%.*]] = add <8 x i32> [[TMP13]], [[TMP12]]
+; FIXED-NEXT: [[TMP15:%.*]] = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> [[BIN_RDX]])
+; FIXED-NEXT: br i1 true, label [[FOR_EXIT:%.*]], label [[SCALAR_PH]]
+; FIXED: scalar.ph:
+;
entry:
br label %for.body
|
lukel97
approved these changes
May 26, 2025
sivan-shani
pushed a commit
to sivan-shani/llvm-project
that referenced
this pull request
Jun 3, 2025
…lvm#141430) Two changes: 1) Handle fixed vector cases now that 77a3f8 has landed. 2) Fix a mistake in the original costing - the VF passed in is the input VF, not the output VF. Given that we should be costing the accumulator type with VF/4. Note that (2) does not cause any visible test differences as the vectorizer (outside of maximize-bandwidth mode) does not consider wide enough VF for the costing difference to matter.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Two changes:
input VF, not the output VF. Given that we should be costing the
accumulator type with VF/4.
Note that (2) does not cause any visible test differences as the vectorizer (outside of maximize-bandwidth mode) does not consider wide enough VF for the costing difference to matter.