[VPlan] Introduce chainUsesScalarValues #158377
Open
+122
−138
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
|
@llvm/pr-subscribers-llvm-transforms @llvm/pr-subscribers-backend-risc-v Author: Ramkumar Ramachandra (artagnon) ChangesIntroduce chainUsesScalarValues to dig through a recipe-chain, skipping widening decisions, and determine if the final leaves use only scalar values of the given root. Demonstrate its utility in narrowToSingleScalarRecipes, showing that it is essentially a drop-in replacement for onlyScalarValuesUsed for the purposes of optimizations. Patch is 34.54 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/158377.diff 13 Files Affected:
diff --git a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
index 2cac5557daeee..92add4a96a5ed 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanTransforms.cpp
@@ -1322,15 +1322,11 @@ static void narrowToSingleScalarRecipes(VPlan &Plan) {
continue;
}
- // Skip recipes that aren't single scalars or don't have only their
- // scalar results used. In the latter case, we would introduce extra
- // broadcasts.
+ // Only consider recipes that are single scalars whose scalar value is
+ // used by the final leaves in a recipe-chain walk: if the final leaves
+ // don't use the scalar value, it could introduce extra broadcasts.
if (!vputils::isSingleScalar(RepOrWidenR) ||
- !all_of(RepOrWidenR->users(), [RepOrWidenR](const VPUser *U) {
- return U->usesScalars(RepOrWidenR) ||
- match(cast<VPRecipeBase>(U),
- m_ExtractLastElement(m_VPValue()));
- }))
+ !vputils::chainUsesScalarValues(RepOrWidenR))
continue;
auto *Clone = new VPReplicateRecipe(RepOrWidenR->getUnderlyingInstr(),
diff --git a/llvm/lib/Transforms/Vectorize/VPlanUtils.cpp b/llvm/lib/Transforms/Vectorize/VPlanUtils.cpp
index c6c1ef3369825..3f21063330b03 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanUtils.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanUtils.cpp
@@ -29,6 +29,31 @@ bool vputils::onlyScalarValuesUsed(const VPValue *Def) {
[Def](const VPUser *U) { return U->usesScalars(Def); });
}
+bool vputils::chainUsesScalarValues(const VPValue *Root) {
+ SmallVector<std::pair<const VPValue *, const VPUser *>> Worklist;
+ for (const VPUser *V : Root->users())
+ Worklist.emplace_back(Root, V);
+ while (!Worklist.empty()) {
+ auto [Op, U] = Worklist.pop_back_val();
+ if (isa<VPWidenRecipe, VPWidenCastRecipe, VPWidenCallRecipe,
+ VPWidenGEPRecipe, VPWidenSelectRecipe, VPWidenIntrinsicRecipe>(U)) {
+ const VPValue *Def = cast<VPRecipeBase>(U)->getVPSingleValue();
+ for (const VPUser *V : Def->users())
+ Worklist.emplace_back(Def, V);
+ continue;
+ }
+ if (auto *Store = dyn_cast<VPWidenStoreRecipe>(U))
+ if (vputils::isSingleScalar(Store->getStoredValue()))
+ continue;
+ if (auto *VPI = dyn_cast<VPInstruction>(U))
+ if (VPI->isVectorToScalar())
+ continue;
+ if (!U->usesScalars(Op))
+ return false;
+ }
+ return true;
+}
+
VPValue *vputils::getOrCreateVPValueForSCEVExpr(VPlan &Plan, const SCEV *Expr) {
if (auto *Expanded = Plan.getSCEVExpansion(Expr))
return Expanded;
diff --git a/llvm/lib/Transforms/Vectorize/VPlanUtils.h b/llvm/lib/Transforms/Vectorize/VPlanUtils.h
index 77c099b271717..8fcb51ec98ade 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanUtils.h
+++ b/llvm/lib/Transforms/Vectorize/VPlanUtils.h
@@ -28,6 +28,11 @@ bool onlyFirstPartUsed(const VPValue *Def);
/// Returns true if only scalar values of \p Def are used by all users.
bool onlyScalarValuesUsed(const VPValue *Def);
+/// Digs through a chain of recipes starting from \p Root, skipping widening
+/// decisions, and determines if the final leaves use only scalar values of \p
+/// Root.
+bool chainUsesScalarValues(const VPValue *Root);
+
/// Get or create a VPValue that corresponds to the expansion of \p Expr. If \p
/// Expr is a SCEVConstant or SCEVUnknown, return a VPValue wrapping the live-in
/// value. Otherwise return a VPExpandSCEVRecipe to expand \p Expr. If \p Plan's
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/conditional-branches-cost.ll b/llvm/test/Transforms/LoopVectorize/AArch64/conditional-branches-cost.ll
index 884eeac09e1e3..c38f80a994e89 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/conditional-branches-cost.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/conditional-branches-cost.ll
@@ -428,48 +428,36 @@ define i32 @header_mask_and_invariant_compare(ptr %A, ptr %B, ptr %C, ptr %D, pt
; DEFAULT-NEXT: [[N_VEC:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF]]
; DEFAULT-NEXT: br label %[[VECTOR_BODY:.*]]
; DEFAULT: [[VECTOR_BODY]]:
-; DEFAULT-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[PRED_STORE_CONTINUE37:.*]] ]
-; DEFAULT-NEXT: [[TMP9:%.*]] = load i32, ptr [[A]], align 4, !alias.scope [[META8:![0-9]+]]
-; DEFAULT-NEXT: [[BROADCAST_SPLATINSERT28:%.*]] = insertelement <4 x i32> poison, i32 [[TMP9]], i64 0
-; DEFAULT-NEXT: [[BROADCAST_SPLAT29:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT28]], <4 x i32> poison, <4 x i32> zeroinitializer
-; DEFAULT-NEXT: [[TMP19:%.*]] = load i32, ptr [[B]], align 4, !alias.scope [[META11:![0-9]+]]
-; DEFAULT-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[TMP19]], i64 0
-; DEFAULT-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
-; DEFAULT-NEXT: [[TMP6:%.*]] = or <4 x i32> [[BROADCAST_SPLAT]], [[BROADCAST_SPLAT29]]
-; DEFAULT-NEXT: [[TMP7:%.*]] = load i32, ptr [[C]], align 4, !alias.scope [[META13:![0-9]+]]
-; DEFAULT-NEXT: [[BROADCAST_SPLATINSERT30:%.*]] = insertelement <4 x i32> poison, i32 [[TMP7]], i64 0
-; DEFAULT-NEXT: [[BROADCAST_SPLAT31:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT30]], <4 x i32> poison, <4 x i32> zeroinitializer
-; DEFAULT-NEXT: [[TMP8:%.*]] = icmp ugt <4 x i32> [[BROADCAST_SPLAT31]], [[TMP6]]
+; DEFAULT-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[PRED_STORE_CONTINUE33:.*]] ]
+; DEFAULT-NEXT: [[TMP3:%.*]] = load i32, ptr [[A]], align 4, !alias.scope [[META8:![0-9]+]]
+; DEFAULT-NEXT: [[TMP4:%.*]] = load i32, ptr [[B]], align 4, !alias.scope [[META11:![0-9]+]]
+; DEFAULT-NEXT: [[TMP5:%.*]] = or i32 [[TMP4]], [[TMP3]]
+; DEFAULT-NEXT: [[TMP6:%.*]] = load i32, ptr [[C]], align 4, !alias.scope [[META13:![0-9]+]]
+; DEFAULT-NEXT: [[TMP7:%.*]] = icmp ugt i32 [[TMP6]], [[TMP5]]
+; DEFAULT-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i1> poison, i1 [[TMP7]], i64 0
+; DEFAULT-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i1> [[BROADCAST_SPLATINSERT]], <4 x i1> poison, <4 x i32> zeroinitializer
; DEFAULT-NEXT: [[TMP16:%.*]] = getelementptr i32, ptr [[D]], i64 [[INDEX]]
-; DEFAULT-NEXT: [[TMP20:%.*]] = extractelement <4 x i1> [[TMP8]], i32 0
-; DEFAULT-NEXT: br i1 [[TMP20]], label %[[PRED_STORE_IF:.*]], label %[[PRED_STORE_CONTINUE:.*]]
+; DEFAULT-NEXT: br i1 [[TMP7]], label %[[PRED_STORE_IF:.*]], label %[[PRED_STORE_CONTINUE:.*]]
; DEFAULT: [[PRED_STORE_IF]]:
-; DEFAULT-NEXT: [[TMP11:%.*]] = extractelement <4 x i32> [[TMP6]], i32 0
-; DEFAULT-NEXT: store i32 [[TMP11]], ptr [[E]], align 4, !alias.scope [[META15:![0-9]+]], !noalias [[META17:![0-9]+]]
+; DEFAULT-NEXT: store i32 [[TMP5]], ptr [[E]], align 4, !alias.scope [[META15:![0-9]+]], !noalias [[META17:![0-9]+]]
; DEFAULT-NEXT: br label %[[PRED_STORE_CONTINUE]]
; DEFAULT: [[PRED_STORE_CONTINUE]]:
-; DEFAULT-NEXT: [[TMP12:%.*]] = extractelement <4 x i1> [[TMP8]], i32 1
-; DEFAULT-NEXT: br i1 [[TMP12]], label %[[PRED_STORE_IF32:.*]], label %[[PRED_STORE_CONTINUE33:.*]]
+; DEFAULT-NEXT: br i1 [[TMP7]], label %[[PRED_STORE_IF28:.*]], label %[[PRED_STORE_CONTINUE29:.*]]
+; DEFAULT: [[PRED_STORE_IF28]]:
+; DEFAULT-NEXT: store i32 [[TMP5]], ptr [[E]], align 4, !alias.scope [[META15]], !noalias [[META17]]
+; DEFAULT-NEXT: br label %[[PRED_STORE_CONTINUE29]]
+; DEFAULT: [[PRED_STORE_CONTINUE29]]:
+; DEFAULT-NEXT: br i1 [[TMP7]], label %[[PRED_STORE_IF30:.*]], label %[[PRED_STORE_CONTINUE31:.*]]
+; DEFAULT: [[PRED_STORE_IF30]]:
+; DEFAULT-NEXT: store i32 [[TMP5]], ptr [[E]], align 4, !alias.scope [[META15]], !noalias [[META17]]
+; DEFAULT-NEXT: br label %[[PRED_STORE_CONTINUE31]]
+; DEFAULT: [[PRED_STORE_CONTINUE31]]:
+; DEFAULT-NEXT: br i1 [[TMP7]], label %[[PRED_STORE_IF32:.*]], label %[[PRED_STORE_CONTINUE33]]
; DEFAULT: [[PRED_STORE_IF32]]:
-; DEFAULT-NEXT: [[TMP13:%.*]] = extractelement <4 x i32> [[TMP6]], i32 0
-; DEFAULT-NEXT: store i32 [[TMP13]], ptr [[E]], align 4, !alias.scope [[META15]], !noalias [[META17]]
+; DEFAULT-NEXT: store i32 [[TMP5]], ptr [[E]], align 4, !alias.scope [[META15]], !noalias [[META17]]
; DEFAULT-NEXT: br label %[[PRED_STORE_CONTINUE33]]
; DEFAULT: [[PRED_STORE_CONTINUE33]]:
-; DEFAULT-NEXT: [[TMP14:%.*]] = extractelement <4 x i1> [[TMP8]], i32 2
-; DEFAULT-NEXT: br i1 [[TMP14]], label %[[PRED_STORE_IF34:.*]], label %[[PRED_STORE_CONTINUE35:.*]]
-; DEFAULT: [[PRED_STORE_IF34]]:
-; DEFAULT-NEXT: [[TMP15:%.*]] = extractelement <4 x i32> [[TMP6]], i32 0
-; DEFAULT-NEXT: store i32 [[TMP15]], ptr [[E]], align 4, !alias.scope [[META15]], !noalias [[META17]]
-; DEFAULT-NEXT: br label %[[PRED_STORE_CONTINUE35]]
-; DEFAULT: [[PRED_STORE_CONTINUE35]]:
-; DEFAULT-NEXT: [[TMP21:%.*]] = extractelement <4 x i1> [[TMP8]], i32 3
-; DEFAULT-NEXT: br i1 [[TMP21]], label %[[PRED_STORE_IF36:.*]], label %[[PRED_STORE_CONTINUE37]]
-; DEFAULT: [[PRED_STORE_IF36]]:
-; DEFAULT-NEXT: [[TMP22:%.*]] = extractelement <4 x i32> [[TMP6]], i32 0
-; DEFAULT-NEXT: store i32 [[TMP22]], ptr [[E]], align 4, !alias.scope [[META15]], !noalias [[META17]]
-; DEFAULT-NEXT: br label %[[PRED_STORE_CONTINUE37]]
-; DEFAULT: [[PRED_STORE_CONTINUE37]]:
-; DEFAULT-NEXT: call void @llvm.masked.store.v4i32.p0(<4 x i32> zeroinitializer, ptr [[TMP16]], i32 4, <4 x i1> [[TMP8]]), !alias.scope [[META19:![0-9]+]], !noalias [[META20:![0-9]+]]
+; DEFAULT-NEXT: call void @llvm.masked.store.v4i32.p0(<4 x i32> zeroinitializer, ptr [[TMP16]], i32 4, <4 x i1> [[BROADCAST_SPLAT]]), !alias.scope [[META19:![0-9]+]], !noalias [[META20:![0-9]+]]
; DEFAULT-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; DEFAULT-NEXT: [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; DEFAULT-NEXT: br i1 [[TMP18]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP21:![0-9]+]]
@@ -727,10 +715,10 @@ define void @test_conditional_interleave_group (ptr noalias %src.1, ptr noalias
; DEFAULT-NEXT: [[BROADCAST_SPLATINSERT8:%.*]] = insertelement <8 x float> poison, float [[TMP15]], i64 0
; DEFAULT-NEXT: [[BROADCAST_SPLAT9:%.*]] = shufflevector <8 x float> [[BROADCAST_SPLATINSERT8]], <8 x float> poison, <8 x i32> zeroinitializer
; DEFAULT-NEXT: [[TMP16:%.*]] = load float, ptr [[SRC_2]], align 4
-; DEFAULT-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <8 x float> poison, float [[TMP16]], i64 0
+; DEFAULT-NEXT: [[TMP17:%.*]] = fmul float [[TMP16]], 0.000000e+00
+; DEFAULT-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <8 x float> poison, float [[TMP17]], i64 0
; DEFAULT-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <8 x float> [[BROADCAST_SPLATINSERT]], <8 x float> poison, <8 x i32> zeroinitializer
-; DEFAULT-NEXT: [[TMP17:%.*]] = fmul <8 x float> [[BROADCAST_SPLAT]], zeroinitializer
-; DEFAULT-NEXT: [[TMP18:%.*]] = call <8 x float> @llvm.fmuladd.v8f32(<8 x float> [[BROADCAST_SPLAT9]], <8 x float> zeroinitializer, <8 x float> [[TMP17]])
+; DEFAULT-NEXT: [[TMP18:%.*]] = call <8 x float> @llvm.fmuladd.v8f32(<8 x float> [[BROADCAST_SPLAT9]], <8 x float> zeroinitializer, <8 x float> [[BROADCAST_SPLAT]])
; DEFAULT-NEXT: [[TMP19:%.*]] = load float, ptr [[SRC_3]], align 4
; DEFAULT-NEXT: [[BROADCAST_SPLATINSERT10:%.*]] = insertelement <8 x float> poison, float [[TMP19]], i64 0
; DEFAULT-NEXT: [[BROADCAST_SPLAT11:%.*]] = shufflevector <8 x float> [[BROADCAST_SPLATINSERT10]], <8 x float> poison, <8 x i32> zeroinitializer
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/extractvalue-no-scalarization-required.ll b/llvm/test/Transforms/LoopVectorize/AArch64/extractvalue-no-scalarization-required.ll
index efcd810203a44..379825bbbab88 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/extractvalue-no-scalarization-required.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/extractvalue-no-scalarization-required.ll
@@ -16,7 +16,7 @@
; CM: vector.ph:
; CM: CLONE ir<%a> = extractvalue ir<%sv>
; CM: CLONE ir<%b> = extractvalue ir<%sv>
-; CM: WIDEN ir<%add> = add ir<%a>, ir<%b>
+; CM: CLONE ir<%add> = add ir<%a>, ir<%b>
; CM: Successor(s): vector loop
; CM: LV: Scalar loop costs: 5.
@@ -30,17 +30,15 @@ define void @test1(ptr %dst, {i64, i64} %sv) {
; FORCED-NEXT: br i1 false, label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; FORCED: [[VECTOR_PH]]:
; FORCED-NEXT: [[TMP0:%.*]] = extractvalue { i64, i64 } [[SV]], 0
-; FORCED-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[TMP0]], i64 0
-; FORCED-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT]], <2 x i64> poison, <2 x i32> zeroinitializer
; FORCED-NEXT: [[TMP4:%.*]] = extractvalue { i64, i64 } [[SV]], 1
-; FORCED-NEXT: [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <2 x i64> poison, i64 [[TMP4]], i64 0
+; FORCED-NEXT: [[TMP5:%.*]] = add i64 [[TMP0]], [[TMP4]]
+; FORCED-NEXT: [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <2 x i64> poison, i64 [[TMP5]], i64 0
; FORCED-NEXT: [[BROADCAST_SPLAT2:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT1]], <2 x i64> poison, <2 x i32> zeroinitializer
-; FORCED-NEXT: [[TMP1:%.*]] = add <2 x i64> [[BROADCAST_SPLAT]], [[BROADCAST_SPLAT2]]
; FORCED-NEXT: br label %[[VECTOR_BODY:.*]]
; FORCED: [[VECTOR_BODY]]:
; FORCED-NEXT: [[INDEX:%.*]] = phi i32 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; FORCED-NEXT: [[TMP2:%.*]] = getelementptr i64, ptr [[DST]], i32 [[INDEX]]
-; FORCED-NEXT: store <2 x i64> [[TMP1]], ptr [[TMP2]], align 4
+; FORCED-NEXT: store <2 x i64> [[BROADCAST_SPLAT2]], ptr [[TMP2]], align 4
; FORCED-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 2
; FORCED-NEXT: [[TMP3:%.*]] = icmp eq i32 [[INDEX_NEXT]], 1000
; FORCED-NEXT: br i1 [[TMP3]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sve-widen-extractvalue.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sve-widen-extractvalue.ll
index 0c6a490ddf4ba..eceda0897b174 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sve-widen-extractvalue.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sve-widen-extractvalue.ll
@@ -17,17 +17,15 @@ define void @widen_extractvalue(ptr %dst, {i64, i64} %sv) #0 {
; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i32 1000, [[TMP3]]
; CHECK-NEXT: [[N_VEC:%.*]] = sub i32 1000, [[N_MOD_VF]]
; CHECK-NEXT: [[EXTRACT0:%.*]] = extractvalue { i64, i64 } [[SV]], 0
-; CHECK-NEXT: [[DOTSPLATINSERT1:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[EXTRACT0]], i64 0
-; CHECK-NEXT: [[DOTSPLAT2:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT1]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP10:%.*]] = extractvalue { i64, i64 } [[SV]], 1
-; CHECK-NEXT: [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[TMP10]], i64 0
+; CHECK-NEXT: [[TMP6:%.*]] = add i64 [[EXTRACT0]], [[TMP10]]
+; CHECK-NEXT: [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[TMP6]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT2:%.*]] = shufflevector <vscale x 2 x i64> [[BROADCAST_SPLATINSERT1]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
-; CHECK-NEXT: [[TMP7:%.*]] = add <vscale x 2 x i64> [[DOTSPLAT2]], [[BROADCAST_SPLAT2]]
; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
; CHECK: [[VECTOR_BODY]]:
; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i64, ptr [[DST]], i32 [[INDEX]]
-; CHECK-NEXT: store <vscale x 2 x i64> [[TMP7]], ptr [[TMP8]], align 8
+; CHECK-NEXT: store <vscale x 2 x i64> [[BROADCAST_SPLAT2]], ptr [[TMP8]], align 8
; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], [[TMP3]]
; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT: br i1 [[TMP9]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
diff --git a/llvm/test/Transforms/LoopVectorize/RISCV/dead-ops-cost.ll b/llvm/test/Transforms/LoopVectorize/RISCV/dead-ops-cost.ll
index e014b9ddbcd0e..95e68d6cc300a 100644
--- a/llvm/test/Transforms/LoopVectorize/RISCV/dead-ops-cost.ll
+++ b/llvm/test/Transforms/LoopVectorize/RISCV/dead-ops-cost.ll
@@ -304,9 +304,9 @@ define void @test_phi_in_latch_redundant(ptr %dst, i32 %a) {
; CHECK-NEXT: [[ENTRY:.*:]]
; CHECK-NEXT: br i1 false, label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
; CHECK: [[VECTOR_PH]]:
-; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[A]], i64 0
+; CHECK-NEXT: [[TMP0:%.*]] = xor i32 [[A]], -1
+; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[TMP0]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 4 x i32> [[BROADCAST_SPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
-; CHECK-NEXT: [[TMP19:%.*]] = xor <vscale x 4 x i32> [[BROADCAST_SPLAT]], splat (i32 -1)
; CHECK-NEXT: [[TMP6:%.*]] = call <vscale x 4 x i64> @llvm.stepvector.nxv4i64()
; CHECK-NEXT: [[TMP7:%.*]] = mul <vscale x 4 x i64> [[TMP6]], splat (i64 9)
; CHECK-NEXT: [[INDUCTION:%.*]] = add <vscale x 4 x i64> zeroinitializer, [[TMP7]]
@@ -320,7 +320,7 @@ define void @test_phi_in_latch_redundant(ptr %dst, i32 %a) {
; CHECK-NEXT: [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <vscale x 4 x i64> poison, i64 [[TMP9]], i64 0
; CHECK-NEXT: [[BROADCAST_SPLAT2:%.*]] = shufflevector <vscale x 4 x i64> [[BROADCAST_SPLATINSERT1]], <vscale x 4 x i64> poison, <vscale x 4 x i32> zeroinitializer
; CHECK-NEXT: [[TMP16:%.*]] = getelementptr i32, ptr [[DST]], <vscale x 4 x i64> [[VEC_IND]]
-; CHECK-NEXT: call void @llvm.vp.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[TMP19]], <vscale x 4 x ptr> align 4 [[TMP16]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP8]])
+; CHECK-NEXT: call void @llvm.vp.scatter.nxv4i32.nxv4p0(<vscale x 4 x i32> [[BROADCAST_SPLAT]], <vscale x 4 x ptr> align 4 [[TMP16]], <vscale x 4 x i1> splat (i1 true), i32 [[TMP8]])
; CHECK-NEXT: [[TMP17:%.*]] = zext i32 [[TMP8]] to i64
; CHECK-NEXT: [[AVL_NEXT]] = sub nuw i64 [[AVL]], [[TMP17]]
; CHECK-NEXT: [[VEC_IND_NEXT]] = add <vscale x 4 x i64> [[VEC_IND]], [[BROADCAST_SPLAT2]]
diff --git a/llvm/test/Transforms/LoopVectorize/X86/cost-model.ll b/llvm/test/Transforms/LoopVectorize/X86/cost-model.ll
index 391653a2efe34..1b171dac3c655 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/cost-model.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/cost-model.ll
@@ -335,7 +335,7 @@ define void @multi_exit(ptr %dst, ptr %src.1, ptr %src.2, i64 %A, i64 %B) #0 {
; CHECK-NEXT: [[TMP1:%.*]] = freeze i64 [[TMP0]]
; CHECK-NEXT: [[UMIN7:%.*]] = call i64 @llvm.umin.i64(i64 [[TMP1]], i64 [[A:%.*]])
; CHECK-NEXT: [[TMP2:%.*]] = add nuw i64 [[UMIN7]], 1
-; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ule i64 [[TMP2]], 28
+; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ule i64 [[TMP2]], 14
; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_SCEVCHECK:%.*]]
; CHECK: vector.scevcheck:
; CHECK-NEXT: [[UMAX:%.*]] = call i64 @llvm.umax.i64(i64 [[B]], i64 1)
@@ -370,14 +370,12 @@ define void @multi_exit(ptr %dst, ptr %src.1, ptr %src.2, i64 %A, i64 %B) #0 {
; CHECK: vector.body:
; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT: [[TMP13:%.*]] = load i64, ptr [[SRC_1]], align 8, !alias.scope [[META6:![0-9]+]]
-; CHECK-NEXT: [[BROADCAST_SPLATINSERT:%.*]] = insertelement <2 x i64> poison, i64 [[TMP13]], i64 0
-; CHECK-NEXT: [[BROADCAST_SPLAT:%.*]] = shufflevector <2 x i64> [[BROADCAST_SPLATINSERT]], <2 x i64> poison, <2 x i32> zeroinitializer
; CHECK-NEXT: [[TMP14:%.*]] = load i64, ptr [[SRC_2]], align 8, !alias.scope [[META9:![0-9]+]]
-; CHECK-NEXT: [[BROADCAST_SPLATINSERT9:%.*]] = insertelement <2 x i64> poison, i64 [[TMP14]], i64 0
-; CHECK-NE...
[truncated]
|
artagnon
force-pushed
the
vplan-chain-uses-scalar-values
branch
from
3b562a9 to
a95f5ca
Compare
lukel97
reviewed
Sep 15, 2025
| match(cast<VPRecipeBase>(U), | ||
| m_ExtractLastElement(m_VPValue())); | ||
| })) | ||
| !vputils::chainUsesScalarValues(RepOrWidenR)) |
There was a problem hiding this comment.
Aren't we already iterating backwards through the VPBB? My understanding is that because we're narrowing the uses before the defs anyway we wouldn't need to look through the chain.
Are we seeing the the diff because we're now checking isVectorToScalar?
There was a problem hiding this comment.
We are iterating backwards through the VPBB, but there is no way to narrow certain Widen-like recipes, as they don't have an underlying value (eg. WidenCast might not have an underlying value). The VPI/Store handling in chainUsesScalarValues is also necessary. Another reason is that the chain may flow into a VPBB we don't handle in our shallow-walk.
artagnon
force-pushed
the
vplan-chain-uses-scalar-values
branch
from
a95f5ca to
5de7b41
Compare
artagnon
force-pushed
the
vplan-chain-uses-scalar-values
branch
from
5de7b41 to
b455f56
Compare
|
Gentle ping. |
david-arm
reviewed
Sep 29, 2025
| Worklist.emplace_back(Root, V); | ||
| while (!Worklist.empty()) { | ||
| auto [Op, U] = Worklist.pop_back_val(); | ||
| if (isa<VPWidenRecipe, VPWidenCastRecipe, VPWidenCallRecipe, |
There was a problem hiding this comment.
Perhaps I'm missing something here, but I don't understand why we're ignoring these. Are you trying to work around a problem where U would return false for usesScalars, but all the users of U would return true? If so, wouldn't it be better to fix usesScalars directly?
There was a problem hiding this comment.
Certain widens cannot be narrowed as they don't have an underlying value that would permit us to turn them into single-scalars: for example, VPWidenCast and VPWidenIntrinsic. All the usesScalars are correct as they are written; I think you meant onlyScalarvaluesUsed in VPlanUtils, which is used by more than just optimizations: it is also used in the execution of recipes, and chainUsesScalars is a version that's suitable for optimizations like narrowToSingleScalars.
| Worklist.emplace_back(Def, V); | ||
| continue; | ||
| } | ||
| if (isa<VPWidenMemoryRecipe>(U) && vputils::isSingleScalar(Op)) |
There was a problem hiding this comment.
Why the special case here for VPWidenMemoryRecipe? Wouldn't you just always want to do:
if (vputils::isSingleScalar(Op))
continue;
? Could you add some comments explaining the code?
There was a problem hiding this comment.
Unfortunately, vputils::isSingleScalar is a bit too powerful, and returns true when the argument is indirectly a single-scalar. Sorry, we're specifically checking for VPWidenStoreRecipe: will fix, but the reason is that it is always profitable to widen stores, even if it's uniform across all lanes; it is an important leaf condition in chainUsesScalarValues, as it has no users, and therefore, any single-scalar addresses or stored-values should be narrowed, allowing the entire chain of uses to be narrowed.
| if (auto *VPI = dyn_cast<VPInstruction>(U)) | ||
| if (VPI->isVectorToScalar() || VPI->isSingleScalar()) | ||
| continue; | ||
| if (!U->usesScalars(Op)) |
There was a problem hiding this comment.
Looks like really you could just update the version of usesScalars in the VPInstruction class to return true when isVectorToScalar or isSingleScalar returns true?
There was a problem hiding this comment.
Nope, that unfortunately doesn't work: usesScalars is intentionally narrower, as it is used in the execution of VPInstructions; isVectorToScalar and isSingleScalar are special cases, which is why they're different routines in the first place.
Introduce chainUsesScalarValues to dig through a recipe-chain, skipping widening decisions, and determine if the final leaves use only scalar values of the given root. Demonstrate its utility in narrowToSingleScalarRecipes, showing that it is essentially a drop-in replacement for onlyScalarValuesUsed for the purposes of optimizations.
artagnon
force-pushed
the
vplan-chain-uses-scalar-values
branch
from
0b322cd to
26b82fa
Compare
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.
Introduce chainUsesScalarValues to dig through a recipe-chain, skipping widening decisions, and determine if the final leaves use only scalar values of the given root. Demonstrate its utility in narrowToSingleScalarRecipes, showing that it is essentially a drop-in replacement for onlyScalarValuesUsed for the purposes of optimizations.