[llvm] 68210c7 - [VPlan] Only generate first lane for VPPredInstPHI if no others used.
Florian Hahn via llvm-commits
llvm-commits at lists.llvm.org
Sat Oct 5 11:15:27 PDT 2024
Author: Florian Hahn
Date: 2024-10-05T19:15:05+01:00
New Revision: 68210c7c269b7b7325dedaea7cea9fe9d979fab5
URL: https://github.com/llvm/llvm-project/commit/68210c7c269b7b7325dedaea7cea9fe9d979fab5
DIFF: https://github.com/llvm/llvm-project/commit/68210c7c269b7b7325dedaea7cea9fe9d979fab5.diff
LOG: [VPlan] Only generate first lane for VPPredInstPHI if no others used.
IF only the first lane of the result is used, only generate the first
lane.
Fixes https://github.com/llvm/llvm-project/issues/111042.
Added:
llvm/test/Transforms/LoopVectorize/X86/replicate-recipe-with-only-first-lane-used.ll
Modified:
llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
llvm/test/Transforms/LoopVectorize/X86/divs-with-tail-folding.ll
llvm/test/Transforms/LoopVectorize/X86/drop-poison-generating-flags.ll
Removed:
################################################################################
diff --git a/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp b/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
index 0d092b9c10acc8..376b3d33be769e 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp
@@ -2185,6 +2185,9 @@ void VPPredInstPHIRecipe::execute(VPTransformState &State) {
// predicated iteration inserts its generated value in the correct vector.
State.reset(getOperand(0), VPhi);
} else {
+ if (vputils::onlyFirstLaneUsed(this) && !State.Lane->isFirstLane())
+ return;
+
Type *PredInstType = getOperand(0)->getUnderlyingValue()->getType();
PHINode *Phi = State.Builder.CreatePHI(PredInstType, 2);
Phi->addIncoming(PoisonValue::get(ScalarPredInst->getType()),
diff --git a/llvm/test/Transforms/LoopVectorize/X86/divs-with-tail-folding.ll b/llvm/test/Transforms/LoopVectorize/X86/divs-with-tail-folding.ll
index 229bdc341a2c34..4c3c2e89cd204b 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/divs-with-tail-folding.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/divs-with-tail-folding.ll
@@ -133,21 +133,18 @@ define void @sdiv_feeding_gep_predicated(ptr %dst, i32 %x, i64 %M, i64 %conv6, i
; CHECK-NEXT: [[TMP13:%.*]] = sdiv i64 [[M]], [[CONV6]]
; CHECK-NEXT: br label %[[PRED_SDIV_CONTINUE4]]
; CHECK: [[PRED_SDIV_CONTINUE4]]:
-; CHECK-NEXT: [[TMP14:%.*]] = phi i64 [ poison, %[[PRED_SDIV_CONTINUE]] ], [ [[TMP13]], %[[PRED_SDIV_IF3]] ]
; CHECK-NEXT: [[TMP15:%.*]] = extractelement <4 x i1> [[TMP8]], i32 2
; CHECK-NEXT: br i1 [[TMP15]], label %[[PRED_SDIV_IF5:.*]], label %[[PRED_SDIV_CONTINUE6:.*]]
; CHECK: [[PRED_SDIV_IF5]]:
; CHECK-NEXT: [[TMP16:%.*]] = sdiv i64 [[M]], [[CONV6]]
; CHECK-NEXT: br label %[[PRED_SDIV_CONTINUE6]]
; CHECK: [[PRED_SDIV_CONTINUE6]]:
-; CHECK-NEXT: [[TMP17:%.*]] = phi i64 [ poison, %[[PRED_SDIV_CONTINUE4]] ], [ [[TMP16]], %[[PRED_SDIV_IF5]] ]
; CHECK-NEXT: [[TMP18:%.*]] = extractelement <4 x i1> [[TMP8]], i32 3
; CHECK-NEXT: br i1 [[TMP18]], label %[[PRED_SDIV_IF7:.*]], label %[[PRED_SDIV_CONTINUE8]]
; CHECK: [[PRED_SDIV_IF7]]:
; CHECK-NEXT: [[TMP19:%.*]] = sdiv i64 [[M]], [[CONV6]]
; CHECK-NEXT: br label %[[PRED_SDIV_CONTINUE8]]
; CHECK: [[PRED_SDIV_CONTINUE8]]:
-; CHECK-NEXT: [[TMP20:%.*]] = phi i64 [ poison, %[[PRED_SDIV_CONTINUE6]] ], [ [[TMP19]], %[[PRED_SDIV_IF7]] ]
; CHECK-NEXT: [[TMP21:%.*]] = trunc i64 [[TMP11]] to i32
; CHECK-NEXT: [[TMP22:%.*]] = mul i64 [[TMP11]], [[CONV61]]
; CHECK-NEXT: [[TMP23:%.*]] = sub i64 [[TMP5]], [[TMP22]]
diff --git a/llvm/test/Transforms/LoopVectorize/X86/drop-poison-generating-flags.ll b/llvm/test/Transforms/LoopVectorize/X86/drop-poison-generating-flags.ll
index fcc3864a7aebe5..7b7b36e15cd5b2 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/drop-poison-generating-flags.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/drop-poison-generating-flags.ll
@@ -509,21 +509,18 @@ define void @pr70590_recipe_without_underlying_instr(i64 %n, ptr noalias %dst) {
; CHECK-NEXT: [[TMP7:%.*]] = srem i64 3, 0
; CHECK-NEXT: br label [[PRED_SREM_CONTINUE2]]
; CHECK: pred.srem.continue2:
-; CHECK-NEXT: [[TMP8:%.*]] = phi i64 [ poison, [[PRED_SREM_CONTINUE]] ], [ [[TMP7]], [[PRED_SREM_IF1]] ]
; CHECK-NEXT: [[TMP9:%.*]] = extractelement <4 x i1> [[TMP2]], i32 2
; CHECK-NEXT: br i1 [[TMP9]], label [[PRED_SREM_IF3:%.*]], label [[PRED_SREM_CONTINUE4:%.*]]
; CHECK: pred.srem.if3:
; CHECK-NEXT: [[TMP10:%.*]] = srem i64 3, 0
; CHECK-NEXT: br label [[PRED_SREM_CONTINUE4]]
; CHECK: pred.srem.continue4:
-; CHECK-NEXT: [[TMP11:%.*]] = phi i64 [ poison, [[PRED_SREM_CONTINUE2]] ], [ [[TMP10]], [[PRED_SREM_IF3]] ]
; CHECK-NEXT: [[TMP12:%.*]] = extractelement <4 x i1> [[TMP2]], i32 3
; CHECK-NEXT: br i1 [[TMP12]], label [[PRED_SREM_IF5:%.*]], label [[PRED_SREM_CONTINUE6]]
; CHECK: pred.srem.if5:
; CHECK-NEXT: [[TMP13:%.*]] = srem i64 3, 0
; CHECK-NEXT: br label [[PRED_SREM_CONTINUE6]]
; CHECK: pred.srem.continue6:
-; CHECK-NEXT: [[TMP14:%.*]] = phi i64 [ poison, [[PRED_SREM_CONTINUE4]] ], [ [[TMP13]], [[PRED_SREM_IF5]] ]
; CHECK-NEXT: [[TMP15:%.*]] = add i64 [[TMP5]], -3
; CHECK-NEXT: [[TMP16:%.*]] = add i64 [[TMP0]], [[TMP15]]
; CHECK-NEXT: [[TMP17:%.*]] = getelementptr [5 x i8], ptr @c, i64 0, i64 [[TMP16]]
diff --git a/llvm/test/Transforms/LoopVectorize/X86/replicate-recipe-with-only-first-lane-used.ll b/llvm/test/Transforms/LoopVectorize/X86/replicate-recipe-with-only-first-lane-used.ll
new file mode 100644
index 00000000000000..e6c9ce3381f73f
--- /dev/null
+++ b/llvm/test/Transforms/LoopVectorize/X86/replicate-recipe-with-only-first-lane-used.ll
@@ -0,0 +1,118 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
+; RUN: opt -p loop-vectorize -force-vector-width=4 -force-vector-interleave=2 -S %s | FileCheck %s
+
+target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-i128:128-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
+; Test case for https://github.com/llvm/llvm-project/issues/111042.
+define void @replicate_udiv_with_only_first_lane_used(i32 %x, ptr %dst, i64 %d) {
+; CHECK-LABEL: define void @replicate_udiv_with_only_first_lane_used(
+; CHECK-SAME: i32 [[X:%.*]], ptr [[DST:%.*]], i64 [[D:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*]]:
+; CHECK-NEXT: [[C:%.*]] = icmp eq i32 [[X]], 10
+; CHECK-NEXT: br i1 false, label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
+; CHECK: [[VECTOR_PH]]:
+; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
+; CHECK: [[VECTOR_BODY]]:
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[PRED_UDIV_CONTINUE14:.*]] ]
+; CHECK-NEXT: br i1 false, label %[[PRED_UDIV_IF:.*]], label %[[PRED_UDIV_CONTINUE:.*]]
+; CHECK: [[PRED_UDIV_IF]]:
+; CHECK-NEXT: [[TMP0:%.*]] = udiv i64 99, [[D]]
+; CHECK-NEXT: br label %[[PRED_UDIV_CONTINUE]]
+; CHECK: [[PRED_UDIV_CONTINUE]]:
+; CHECK-NEXT: [[TMP1:%.*]] = phi i64 [ poison, %[[VECTOR_BODY]] ], [ [[TMP0]], %[[PRED_UDIV_IF]] ]
+; CHECK-NEXT: br i1 false, label %[[PRED_UDIV_IF1:.*]], label %[[PRED_UDIV_CONTINUE2:.*]]
+; CHECK: [[PRED_UDIV_IF1]]:
+; CHECK-NEXT: [[TMP2:%.*]] = udiv i64 99, [[D]]
+; CHECK-NEXT: br label %[[PRED_UDIV_CONTINUE2]]
+; CHECK: [[PRED_UDIV_CONTINUE2]]:
+; CHECK-NEXT: br i1 false, label %[[PRED_UDIV_IF3:.*]], label %[[PRED_UDIV_CONTINUE4:.*]]
+; CHECK: [[PRED_UDIV_IF3]]:
+; CHECK-NEXT: [[TMP3:%.*]] = udiv i64 99, [[D]]
+; CHECK-NEXT: br label %[[PRED_UDIV_CONTINUE4]]
+; CHECK: [[PRED_UDIV_CONTINUE4]]:
+; CHECK-NEXT: br i1 false, label %[[PRED_UDIV_IF5:.*]], label %[[PRED_UDIV_CONTINUE6:.*]]
+; CHECK: [[PRED_UDIV_IF5]]:
+; CHECK-NEXT: [[TMP4:%.*]] = udiv i64 99, [[D]]
+; CHECK-NEXT: br label %[[PRED_UDIV_CONTINUE6]]
+; CHECK: [[PRED_UDIV_CONTINUE6]]:
+; CHECK-NEXT: br i1 false, label %[[PRED_UDIV_IF7:.*]], label %[[PRED_UDIV_CONTINUE8:.*]]
+; CHECK: [[PRED_UDIV_IF7]]:
+; CHECK-NEXT: [[TMP5:%.*]] = udiv i64 99, [[D]]
+; CHECK-NEXT: br label %[[PRED_UDIV_CONTINUE8]]
+; CHECK: [[PRED_UDIV_CONTINUE8]]:
+; CHECK-NEXT: [[TMP6:%.*]] = phi i64 [ poison, %[[PRED_UDIV_CONTINUE6]] ], [ [[TMP5]], %[[PRED_UDIV_IF7]] ]
+; CHECK-NEXT: br i1 false, label %[[PRED_UDIV_IF9:.*]], label %[[PRED_UDIV_CONTINUE10:.*]]
+; CHECK: [[PRED_UDIV_IF9]]:
+; CHECK-NEXT: [[TMP7:%.*]] = udiv i64 99, [[D]]
+; CHECK-NEXT: br label %[[PRED_UDIV_CONTINUE10]]
+; CHECK: [[PRED_UDIV_CONTINUE10]]:
+; CHECK-NEXT: br i1 false, label %[[PRED_UDIV_IF11:.*]], label %[[PRED_UDIV_CONTINUE12:.*]]
+; CHECK: [[PRED_UDIV_IF11]]:
+; CHECK-NEXT: [[TMP8:%.*]] = udiv i64 99, [[D]]
+; CHECK-NEXT: br label %[[PRED_UDIV_CONTINUE12]]
+; CHECK: [[PRED_UDIV_CONTINUE12]]:
+; CHECK-NEXT: br i1 false, label %[[PRED_UDIV_IF13:.*]], label %[[PRED_UDIV_CONTINUE14]]
+; CHECK: [[PRED_UDIV_IF13]]:
+; CHECK-NEXT: [[TMP9:%.*]] = udiv i64 99, [[D]]
+; CHECK-NEXT: br label %[[PRED_UDIV_CONTINUE14]]
+; CHECK: [[PRED_UDIV_CONTINUE14]]:
+; CHECK-NEXT: [[PREDPHI:%.*]] = select i1 true, i64 0, i64 [[TMP1]]
+; CHECK-NEXT: [[PREDPHI15:%.*]] = select i1 true, i64 0, i64 [[TMP6]]
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i16, ptr [[DST]], i64 [[PREDPHI]]
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i16, ptr [[DST]], i64 [[PREDPHI15]]
+; CHECK-NEXT: store i16 0, ptr [[TMP10]], align 2
+; CHECK-NEXT: store i16 0, ptr [[TMP11]], align 2
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
+; CHECK-NEXT: [[TMP12:%.*]] = icmp eq i64 [[INDEX_NEXT]], 96
+; CHECK-NEXT: br i1 [[TMP12]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; CHECK: [[MIDDLE_BLOCK]]:
+; CHECK-NEXT: br i1 false, label %[[EXIT:.*]], label %[[SCALAR_PH]]
+; CHECK: [[SCALAR_PH]]:
+; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ 96, %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ]
+; CHECK-NEXT: br label %[[LOOP_HEADER:.*]]
+; CHECK: [[LOOP_HEADER]]:
+; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[LOOP_LATCH:.*]] ]
+; CHECK-NEXT: br i1 true, label %[[LOOP_LATCH]], label %[[ELSE:.*]]
+; CHECK: [[ELSE]]:
+; CHECK-NEXT: [[DIV_I:%.*]] = udiv i64 99, [[D]]
+; CHECK-NEXT: br label %[[LOOP_LATCH]]
+; CHECK: [[LOOP_LATCH]]:
+; CHECK-NEXT: [[RETVAL_0_I:%.*]] = phi i64 [ [[DIV_I]], %[[ELSE]] ], [ 0, %[[LOOP_HEADER]] ]
+; CHECK-NEXT: [[GEP:%.*]] = getelementptr i16, ptr [[DST]], i64 [[RETVAL_0_I]]
+; CHECK-NEXT: store i16 0, ptr [[GEP]], align 2
+; CHECK-NEXT: [[IV_NEXT]] = add i64 [[IV]], 1
+; CHECK-NEXT: [[CMP:%.*]] = icmp ult i64 [[IV_NEXT]], 101
+; CHECK-NEXT: br i1 [[CMP]], label %[[LOOP_HEADER]], label %[[EXIT]], !llvm.loop [[LOOP3:![0-9]+]]
+; CHECK: [[EXIT]]:
+; CHECK-NEXT: ret void
+;
+entry:
+ %c = icmp eq i32 %x, 10
+ br label %loop.header
+
+loop.header: ; preds = %loop.latch, %entry
+ %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
+ br i1 true, label %loop.latch, label %else
+
+else:
+ %div.i = udiv i64 99, %d
+ br label %loop.latch
+
+loop.latch:
+ %retval.0.i = phi i64 [ %div.i, %else ], [ 0, %loop.header ]
+ %gep = getelementptr i16, ptr %dst, i64 %retval.0.i
+ store i16 0, ptr %gep, align 2
+ %iv.next = add i64 %iv, 1
+ %cmp = icmp ult i64 %iv.next, 101
+ br i1 %cmp, label %loop.header, label %exit
+
+exit:
+ ret void
+}
+;.
+; CHECK: [[LOOP0]] = distinct !{[[LOOP0]], [[META1:![0-9]+]], [[META2:![0-9]+]]}
+; CHECK: [[META1]] = !{!"llvm.loop.isvectorized", i32 1}
+; CHECK: [[META2]] = !{!"llvm.loop.unroll.runtime.disable"}
+; CHECK: [[LOOP3]] = distinct !{[[LOOP3]], [[META2]], [[META1]]}
+;.
More information about the llvm-commits
mailing list