[llvm] [LV] Regen a test with UTC (PR #139235)
via llvm-commits
llvm-commits at lists.llvm.org
Fri May 9 02:55:57 PDT 2025
llvmbot wrote:
<!--LLVM PR SUMMARY COMMENT-->
@llvm/pr-subscribers-llvm-transforms
Author: Ramkumar Ramachandra (artagnon)
<details>
<summary>Changes</summary>
---
Patch is 46.44 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/139235.diff
1 Files Affected:
- (modified) llvm/test/Transforms/LoopVectorize/X86/drop-poison-generating-flags.ll (+351-233)
``````````diff
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 8beb467572826..207f3e8ba915e 100644
--- a/llvm/test/Transforms/LoopVectorize/X86/drop-poison-generating-flags.ll
+++ b/llvm/test/Transforms/LoopVectorize/X86/drop-poison-generating-flags.ll
@@ -1,3 +1,4 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --filter-out-after "^middle.block" --version 5
; RUN: opt %s -passes=loop-vectorize -force-vector-width=4 -force-vector-interleave=1 -S | FileCheck %s
; Make sure that integer poison-generating flags (i.e., nuw/nsw, exact and inbounds)
@@ -18,18 +19,31 @@ target triple = "x86_64-pc-linux-gnu"
; Drop poison-generating flags from 'sub' and 'getelementptr' feeding a masked load.
; Test for PR52111.
-define void @drop_scalar_nuw_nsw(ptr noalias nocapture readonly %input,
- ptr %output) local_unnamed_addr #0 {
-; CHECK-LABEL: @drop_scalar_nuw_nsw(
-; CHECK: vector.body:
-; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, {{.*}} ]
-; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, {{.*}} ]
-; CHECK: [[TMP4:%.*]] = icmp eq <4 x i64> [[VEC_IND]], zeroinitializer
-; CHECK-NEXT: [[TMP7:%.*]] = xor <4 x i1> [[TMP4]], splat (i1 true)
-; CHECK-NEXT: [[TMP5:%.*]] = sub i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr float, ptr [[INPUT:%.*]], i64 [[TMP5]]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr float, ptr [[TMP6]], i32 0
-; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x float> @llvm.masked.load.v4f32.p0(ptr [[TMP8]], i32 4, <4 x i1> [[TMP7]], <4 x float> poison), !invariant.load !0
+define void @drop_scalar_nuw_nsw(ptr noalias nocapture readonly %input, ptr %output) local_unnamed_addr #0 {
+; CHECK-LABEL: define void @drop_scalar_nuw_nsw(
+; CHECK-SAME: ptr noalias readonly captures(none) [[INPUT:%.*]], ptr [[OUTPUT:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: br i1 false, [[SCALAR_PH:label %.*]], 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:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP0:%.*]] = icmp eq <4 x i64> [[VEC_IND]], zeroinitializer
+; CHECK-NEXT: [[TMP1:%.*]] = xor <4 x i1> [[TMP0]], splat (i1 true)
+; CHECK-NEXT: [[TMP2:%.*]] = sub i64 [[INDEX]], 1
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr float, ptr [[INPUT]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr float, ptr [[TMP3]], i32 0
+; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x float> @llvm.masked.load.v4f32.p0(ptr [[TMP4]], i32 4, <4 x i1> [[TMP1]], <4 x float> poison), !invariant.load [[META0:![0-9]+]]
+; CHECK-NEXT: [[PREDPHI:%.*]] = select <4 x i1> [[TMP0]], <4 x float> zeroinitializer, <4 x float> [[WIDE_MASKED_LOAD]]
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds float, ptr [[OUTPUT]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[TMP5]], i32 0
+; CHECK-NEXT: store <4 x float> [[PREDPHI]], ptr [[TMP6]], align 4
+; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], splat (i64 4)
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; CHECK-NEXT: br i1 true, label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP1:![0-9]+]]
+; CHECK: [[MIDDLE_BLOCK]]:
+;
entry:
br label %loop.header
@@ -57,18 +71,31 @@ loop.exit:
}
; Variant with getelementptr nusw.
-define void @drop_scalar_gep_nusw(ptr noalias nocapture readonly %input,
- ptr %output) local_unnamed_addr #0 {
-; CHECK-LABEL: @drop_scalar_gep_nusw(
-; CHECK: vector.body:
-; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, {{.*}} ]
-; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, {{.*}} ]
-; CHECK: [[TMP4:%.*]] = icmp eq <4 x i64> [[VEC_IND]], zeroinitializer
-; CHECK-NEXT: [[TMP7:%.*]] = xor <4 x i1> [[TMP4]], splat (i1 true)
-; CHECK-NEXT: [[TMP5:%.*]] = sub i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr float, ptr [[INPUT:%.*]], i64 [[TMP5]]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr float, ptr [[TMP6]], i32 0
-; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x float> @llvm.masked.load.v4f32.p0(ptr [[TMP8]], i32 4, <4 x i1> [[TMP7]], <4 x float> poison), !invariant.load !0
+define void @drop_scalar_gep_nusw(ptr noalias nocapture readonly %input, ptr %output) local_unnamed_addr #0 {
+; CHECK-LABEL: define void @drop_scalar_gep_nusw(
+; CHECK-SAME: ptr noalias readonly captures(none) [[INPUT:%.*]], ptr [[OUTPUT:%.*]]) local_unnamed_addr #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: br i1 false, [[SCALAR_PH:label %.*]], 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:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP0:%.*]] = icmp eq <4 x i64> [[VEC_IND]], zeroinitializer
+; CHECK-NEXT: [[TMP1:%.*]] = xor <4 x i1> [[TMP0]], splat (i1 true)
+; CHECK-NEXT: [[TMP2:%.*]] = sub i64 [[INDEX]], 1
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr float, ptr [[INPUT]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr float, ptr [[TMP3]], i32 0
+; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x float> @llvm.masked.load.v4f32.p0(ptr [[TMP4]], i32 4, <4 x i1> [[TMP1]], <4 x float> poison), !invariant.load [[META0]]
+; CHECK-NEXT: [[PREDPHI:%.*]] = select <4 x i1> [[TMP0]], <4 x float> zeroinitializer, <4 x float> [[WIDE_MASKED_LOAD]]
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr nusw float, ptr [[OUTPUT]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr nusw float, ptr [[TMP5]], i32 0
+; CHECK-NEXT: store <4 x float> [[PREDPHI]], ptr [[TMP6]], align 4
+; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], splat (i64 4)
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; CHECK-NEXT: br i1 true, label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
+; CHECK: [[MIDDLE_BLOCK]]:
+;
entry:
br label %loop.header
@@ -96,18 +123,31 @@ loop.exit:
}
; Variant with getelementptr nuw.
-define void @drop_scalar_gep_nuw(ptr noalias nocapture readonly %input,
- ptr %output) local_unnamed_addr #0 {
-; CHECK-LABEL: @drop_scalar_gep_nuw(
-; CHECK: vector.body:
-; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, {{.*}} ]
-; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, {{.*}} ]
-; CHECK: [[TMP4:%.*]] = icmp eq <4 x i64> [[VEC_IND]], zeroinitializer
-; CHECK-NEXT: [[TMP7:%.*]] = xor <4 x i1> [[TMP4]], splat (i1 true)
-; CHECK-NEXT: [[TMP5:%.*]] = sub i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr float, ptr [[INPUT:%.*]], i64 [[TMP5]]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr float, ptr [[TMP6]], i32 0
-; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x float> @llvm.masked.load.v4f32.p0(ptr [[TMP8]], i32 4, <4 x i1> [[TMP7]], <4 x float> poison), !invariant.load !0
+define void @drop_scalar_gep_nuw(ptr noalias nocapture readonly %input, ptr %output) local_unnamed_addr #0 {
+; CHECK-LABEL: define void @drop_scalar_gep_nuw(
+; CHECK-SAME: ptr noalias readonly captures(none) [[INPUT:%.*]], ptr [[OUTPUT:%.*]]) local_unnamed_addr #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: br i1 false, [[SCALAR_PH:label %.*]], 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:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP0:%.*]] = icmp eq <4 x i64> [[VEC_IND]], zeroinitializer
+; CHECK-NEXT: [[TMP1:%.*]] = xor <4 x i1> [[TMP0]], splat (i1 true)
+; CHECK-NEXT: [[TMP2:%.*]] = sub i64 [[INDEX]], 1
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr float, ptr [[INPUT]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr float, ptr [[TMP3]], i32 0
+; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x float> @llvm.masked.load.v4f32.p0(ptr [[TMP4]], i32 4, <4 x i1> [[TMP1]], <4 x float> poison), !invariant.load [[META0]]
+; CHECK-NEXT: [[PREDPHI:%.*]] = select <4 x i1> [[TMP0]], <4 x float> zeroinitializer, <4 x float> [[WIDE_MASKED_LOAD]]
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr nuw float, ptr [[OUTPUT]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr nuw float, ptr [[TMP5]], i32 0
+; CHECK-NEXT: store <4 x float> [[PREDPHI]], ptr [[TMP6]], align 4
+; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], splat (i64 4)
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; CHECK-NEXT: br i1 true, label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
+; CHECK: [[MIDDLE_BLOCK]]:
+;
entry:
br label %loop.header
@@ -136,18 +176,31 @@ loop.exit:
; Drop poison-generating flags from 'sub' and 'getelementptr' feeding a masked load.
; In this case, 'sub' and 'getelementptr' are not guarded by the predicate.
-define void @drop_nonpred_scalar_nuw_nsw(ptr noalias nocapture readonly %input,
- ptr %output) local_unnamed_addr #0 {
-; CHECK-LABEL: @drop_nonpred_scalar_nuw_nsw(
-; CHECK: vector.body:
-; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, {{.*}} ]
-; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, {{.*}} ]
-; CHECK: [[TMP5:%.*]] = sub i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr float, ptr [[INPUT:%.*]], i64 [[TMP5]]
-; CHECK-NEXT: [[TMP4:%.*]] = icmp eq <4 x i64> [[VEC_IND]], zeroinitializer
-; CHECK-NEXT: [[TMP7:%.*]] = xor <4 x i1> [[TMP4]], splat (i1 true)
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr float, ptr [[TMP6]], i32 0
-; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x float> @llvm.masked.load.v4f32.p0(ptr [[TMP8]], i32 4, <4 x i1> [[TMP7]], <4 x float> poison), !invariant.load !0
+define void @drop_nonpred_scalar_nuw_nsw(ptr noalias nocapture readonly %input, ptr %output) local_unnamed_addr #0 {
+; CHECK-LABEL: define void @drop_nonpred_scalar_nuw_nsw(
+; CHECK-SAME: ptr noalias readonly captures(none) [[INPUT:%.*]], ptr [[OUTPUT:%.*]]) local_unnamed_addr #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: br i1 false, [[SCALAR_PH:label %.*]], 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:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP0:%.*]] = sub i64 [[INDEX]], 1
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr float, ptr [[INPUT]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP2:%.*]] = icmp eq <4 x i64> [[VEC_IND]], zeroinitializer
+; CHECK-NEXT: [[TMP3:%.*]] = xor <4 x i1> [[TMP2]], splat (i1 true)
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr float, ptr [[TMP1]], i32 0
+; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x float> @llvm.masked.load.v4f32.p0(ptr [[TMP4]], i32 4, <4 x i1> [[TMP3]], <4 x float> poison), !invariant.load [[META0]]
+; CHECK-NEXT: [[PREDPHI:%.*]] = select <4 x i1> [[TMP2]], <4 x float> zeroinitializer, <4 x float> [[WIDE_MASKED_LOAD]]
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds float, ptr [[OUTPUT]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[TMP5]], i32 0
+; CHECK-NEXT: store <4 x float> [[PREDPHI]], ptr [[TMP6]], align 4
+; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], splat (i64 4)
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; CHECK-NEXT: br i1 true, label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
+; CHECK: [[MIDDLE_BLOCK]]:
+;
entry:
br label %loop.header
@@ -175,18 +228,31 @@ loop.exit:
}
; Preserve poison-generating flags from vector 'sub', 'mul' and 'getelementptr' feeding a masked gather.
-define void @preserve_vector_nuw_nsw(ptr noalias nocapture readonly %input,
- ptr %output) local_unnamed_addr #0 {
-; CHECK-LABEL: @preserve_vector_nuw_nsw(
-; CHECK: vector.body:
-; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, {{.*}} ]
-; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, {{.*}} ]
-; CHECK: [[TMP4:%.*]] = icmp eq <4 x i64> [[VEC_IND]], zeroinitializer
-; CHECK-NEXT: [[TMP8:%.*]] = xor <4 x i1> [[TMP4]], splat (i1 true)
-; CHECK-NEXT: [[TMP5:%.*]] = sub nuw nsw <4 x i64> [[VEC_IND]], splat (i64 1)
-; CHECK-NEXT: [[TMP6:%.*]] = mul nuw nsw <4 x i64> [[TMP5]], splat (i64 2)
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds float, ptr [[INPUT:%.*]], <4 x i64> [[TMP6]]
-; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> [[TMP7]], i32 4, <4 x i1> [[TMP8]], <4 x float> poison), !invariant.load !0
+define void @preserve_vector_nuw_nsw(ptr noalias nocapture readonly %input, ptr %output) local_unnamed_addr #0 {
+; CHECK-LABEL: define void @preserve_vector_nuw_nsw(
+; CHECK-SAME: ptr noalias readonly captures(none) [[INPUT:%.*]], ptr [[OUTPUT:%.*]]) local_unnamed_addr #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: br i1 false, [[SCALAR_PH:label %.*]], 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:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP0:%.*]] = icmp eq <4 x i64> [[VEC_IND]], zeroinitializer
+; CHECK-NEXT: [[TMP1:%.*]] = xor <4 x i1> [[TMP0]], splat (i1 true)
+; CHECK-NEXT: [[TMP2:%.*]] = sub nuw nsw <4 x i64> [[VEC_IND]], splat (i64 1)
+; CHECK-NEXT: [[TMP3:%.*]] = mul nuw nsw <4 x i64> [[TMP2]], splat (i64 2)
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds float, ptr [[INPUT]], <4 x i64> [[TMP3]]
+; CHECK-NEXT: [[WIDE_MASKED_GATHER:%.*]] = call <4 x float> @llvm.masked.gather.v4f32.v4p0(<4 x ptr> [[TMP4]], i32 4, <4 x i1> [[TMP1]], <4 x float> poison), !invariant.load [[META0]]
+; CHECK-NEXT: [[PREDPHI:%.*]] = select <4 x i1> [[TMP0]], <4 x float> zeroinitializer, <4 x float> [[WIDE_MASKED_GATHER]]
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds float, ptr [[OUTPUT]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds float, ptr [[TMP5]], i32 0
+; CHECK-NEXT: store <4 x float> [[PREDPHI]], ptr [[TMP6]], align 4
+; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], splat (i64 4)
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; CHECK-NEXT: br i1 true, label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP11:![0-9]+]]
+; CHECK: [[MIDDLE_BLOCK]]:
+;
entry:
br label %loop.header
@@ -215,20 +281,35 @@ loop.exit:
}
; Drop poison-generating flags from vector 'sub' and 'gep' feeding a masked load.
-define void @drop_vector_nuw_nsw(ptr noalias nocapture readonly %input,
- ptr %output, ptr noalias %ptrs) local_unnamed_addr #0 {
-; CHECK-LABEL: @drop_vector_nuw_nsw(
-; CHECK: vector.body:
-; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, {{.*}} ]
-; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, {{.*}} ]
-; CHECK: [[TMP4:%.*]] = icmp eq <4 x i64> [[VEC_IND]], zeroinitializer
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds ptr, ptr [[PTRS:%.*]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP6:%.*]] = sub <4 x i64> [[VEC_IND]], splat (i64 1)
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr float, ptr [[INPUT:%.*]], <4 x i64> [[TMP6]]
-; CHECK: [[TMP10:%.*]] = xor <4 x i1> [[TMP4]], splat (i1 true)
-; CHECK-NEXT: [[TMP11:%.*]] = extractelement <4 x ptr> [[TMP7]], i32 0
-; CHECK-NEXT: [[TMP12:%.*]] = getelementptr float, ptr [[TMP11]], i32 0
-; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x float> @llvm.masked.load.v4f32.p0(ptr [[TMP12]], i32 4, <4 x i1> [[TMP10]], <4 x float> poison), !invariant.load !0
+define void @drop_vector_nuw_nsw(ptr noalias nocapture readonly %input, ptr %output, ptr noalias %ptrs) local_unnamed_addr #0 {
+; CHECK-LABEL: define void @drop_vector_nuw_nsw(
+; CHECK-SAME: ptr noalias readonly captures(none) [[INPUT:%.*]], ptr [[OUTPUT:%.*]], ptr noalias [[PTRS:%.*]]) local_unnamed_addr #[[ATTR0]] {
+; CHECK-NEXT: [[ENTRY:.*:]]
+; CHECK-NEXT: br i1 false, [[SCALAR_PH:label %.*]], 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:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP0:%.*]] = icmp eq <4 x i64> [[VEC_IND]], zeroinitializer
+; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds ptr, ptr [[PTRS]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP2:%.*]] = sub <4 x i64> [[VEC_IND]], splat (i64 1)
+; CHECK-NEXT: [[TMP3:%.*]] = getelementptr float, ptr [[INPUT]], <4 x i64> [[TMP2]]
+; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i32 0
+; CHECK-NEXT: store <4 x ptr> [[TMP3]], ptr [[TMP4]], align 8
+; CHECK-NEXT: [[TMP5:%.*]] = xor <4 x i1> [[TMP0]], splat (i1 true)
+; CHECK-NEXT: [[TMP6:%.*]] = extractelement <4 x ptr> [[TMP3]], i32 0
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr float, ptr [[TMP6]], i32 0
+; CHECK-NEXT: [[WIDE_MASKED_LOAD:%.*]] = call <4 x float> @llvm.masked.load.v4f32.p0(ptr [[TMP7]], i32 4, <4 x i1> [[TMP5]], <4 x float> poison), !invariant.load [[META0]]
+; CHECK-NEXT: [[PREDPHI:%.*]] = select <4 x i1> [[TMP0]], <4 x float> zeroinitializer, <4 x float> [[WIDE_MASKED_LOAD]]
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, ptr [[OUTPUT]], i64 [[INDEX]]
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds float, ptr [[TMP8]], i32 0
+; CHECK-NEXT: store <4 x float> [[PREDPHI]], ptr [[TMP9]], align 4
+; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], splat (i64 4)
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; CHECK-NEXT: br i1 true, label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP13:![0-9]+]]
+; CHECK: [[MIDDLE_BLOCK]]:
+;
entry:
br label %loop.header
@@ -260,16 +341,26 @@ loop.exit:
; Preserve poison-generating flags from 'sub', which is not contributing to any address computation
; of any masked load/store/gather/scatter.
define void @preserve_nuw_nsw_no_addr(ptr %output) local_unnamed_addr #0 {
-; CHECK-LABEL: @preserve_nuw_nsw_no_addr(
-; CHECK: vector.body:
-; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, {{.*}} ]
-; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, {{.*}} ]
-; CHECK: [[TMP4:%.*]] = icmp eq <4 x i64> [[VEC_IND]], zeroinitializer
-; CHECK-NEXT: [[TMP5:%.*]] = sub nuw nsw <4 x i64> [[VEC_IND]], splat (i64 1)
-; CHECK-NEXT: [[PREDPHI:%.*]] = select <4 x i1> [[TMP4]], <4 x i64> zeroinitializer, <4 x i64> [[TMP5]]
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[OUTPUT:%.*]], i64 [[INDEX]]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[TMP7]], i32 0
-; CHECK-NEXT: store <4 x i64> [[PREDPHI]], ptr [[TMP8]], ...
[truncated]
``````````
</details>
https://github.com/llvm/llvm-project/pull/139235
More information about the llvm-commits
mailing list