[llvm] [LV] Increase coverage of uniformity-rewriter (PR #161219)
Ramkumar Ramachandra via llvm-commits
llvm-commits at lists.llvm.org
Mon Oct 6 02:50:39 PDT 2025
https://github.com/artagnon updated https://github.com/llvm/llvm-project/pull/161219
>From 230c66271830b0370d1c825713cac5be923072db Mon Sep 17 00:00:00 2001
From: Ramkumar Ramachandra <ramkumar.ramachandra at codasip.com>
Date: Mon, 29 Sep 2025 15:48:15 +0100
Subject: [PATCH 1/2] [LV] Increase coverage of uniformity-rewriter
Add a test with a maybe-uniform load of an argument (SCEVUnknown),
showing that SCEVAddRecForUniformityRewriter bails out when it sees that
the top-level select expression doesn't contain a UDivExpr.
---
.../LoopVectorize/uniformity-rewriter.ll | 96 +++++++++++++++++++
1 file changed, 96 insertions(+)
create mode 100644 llvm/test/Transforms/LoopVectorize/uniformity-rewriter.ll
diff --git a/llvm/test/Transforms/LoopVectorize/uniformity-rewriter.ll b/llvm/test/Transforms/LoopVectorize/uniformity-rewriter.ll
new file mode 100644
index 0000000000000..f4a1ffebe8cfb
--- /dev/null
+++ b/llvm/test/Transforms/LoopVectorize/uniformity-rewriter.ll
@@ -0,0 +1,96 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --version 6
+; RUN: opt -passes=loop-vectorize -force-vector-width=4 -S %s | FileCheck %s
+
+; Test to exercise the uniformity rewriter.
+
+define void @uniformityrew(ptr %src.a, ptr noalias %src.b, ptr noalias %dst, i64 %n) {
+; CHECK-LABEL: define void @uniformityrew(
+; CHECK-SAME: ptr [[SRC_A:%.*]], ptr noalias [[SRC_B:%.*]], ptr noalias [[DST:%.*]], i64 [[N:%.*]]) {
+; CHECK-NEXT: [[ENTRY:.*]]:
+; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[N]], 1
+; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP0]], 4
+; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
+; CHECK: [[VECTOR_PH]]:
+; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[TMP0]], 4
+; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF]]
+; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
+; CHECK: [[VECTOR_BODY]]:
+; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT: [[TMP1:%.*]] = add i64 [[INDEX]], 0
+; CHECK-NEXT: [[TMP2:%.*]] = add i64 [[INDEX]], 1
+; CHECK-NEXT: [[TMP3:%.*]] = add i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP4:%.*]] = add i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i32, ptr [[SRC_A]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP6:%.*]] = getelementptr i32, ptr [[SRC_A]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[SRC_A]], i64 [[TMP3]]
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[SRC_A]], i64 [[TMP4]]
+; CHECK-NEXT: [[TMP9:%.*]] = load i64, ptr [[TMP5]], align 4
+; CHECK-NEXT: [[TMP10:%.*]] = load i64, ptr [[TMP6]], align 4
+; CHECK-NEXT: [[TMP11:%.*]] = load i64, ptr [[TMP7]], align 4
+; CHECK-NEXT: [[TMP12:%.*]] = load i64, ptr [[TMP8]], align 4
+; CHECK-NEXT: [[TMP13:%.*]] = insertelement <4 x i64> poison, i64 [[TMP9]], i32 0
+; CHECK-NEXT: [[TMP14:%.*]] = insertelement <4 x i64> [[TMP13]], i64 [[TMP10]], i32 1
+; CHECK-NEXT: [[TMP15:%.*]] = insertelement <4 x i64> [[TMP14]], i64 [[TMP11]], i32 2
+; CHECK-NEXT: [[TMP16:%.*]] = insertelement <4 x i64> [[TMP15]], i64 [[TMP12]], i32 3
+; CHECK-NEXT: [[TMP17:%.*]] = udiv <4 x i64> [[TMP16]], splat (i64 2)
+; CHECK-NEXT: [[TMP18:%.*]] = extractelement <4 x i64> [[TMP17]], i32 0
+; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i32, ptr [[SRC_B]], i64 [[TMP18]]
+; CHECK-NEXT: [[TMP20:%.*]] = extractelement <4 x i64> [[TMP17]], i32 1
+; CHECK-NEXT: [[TMP21:%.*]] = getelementptr i32, ptr [[SRC_B]], i64 [[TMP20]]
+; CHECK-NEXT: [[TMP22:%.*]] = extractelement <4 x i64> [[TMP17]], i32 2
+; CHECK-NEXT: [[TMP23:%.*]] = getelementptr i32, ptr [[SRC_B]], i64 [[TMP22]]
+; CHECK-NEXT: [[TMP24:%.*]] = extractelement <4 x i64> [[TMP17]], i32 3
+; CHECK-NEXT: [[TMP25:%.*]] = getelementptr i32, ptr [[SRC_B]], i64 [[TMP24]]
+; CHECK-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP19]], align 4
+; CHECK-NEXT: [[TMP27:%.*]] = load i32, ptr [[TMP21]], align 4
+; CHECK-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP23]], align 4
+; CHECK-NEXT: [[TMP29:%.*]] = load i32, ptr [[TMP25]], align 4
+; CHECK-NEXT: [[TMP30:%.*]] = insertelement <4 x i32> poison, i32 [[TMP26]], i32 0
+; CHECK-NEXT: [[TMP31:%.*]] = insertelement <4 x i32> [[TMP30]], i32 [[TMP27]], i32 1
+; CHECK-NEXT: [[TMP32:%.*]] = insertelement <4 x i32> [[TMP31]], i32 [[TMP28]], i32 2
+; CHECK-NEXT: [[TMP33:%.*]] = insertelement <4 x i32> [[TMP32]], i32 [[TMP29]], i32 3
+; CHECK-NEXT: [[TMP34:%.*]] = getelementptr i32, ptr [[DST]], i64 [[TMP1]]
+; CHECK-NEXT: store <4 x i32> [[TMP33]], ptr [[TMP34]], align 4
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP35:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; CHECK-NEXT: br i1 [[TMP35]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; CHECK: [[MIDDLE_BLOCK]]:
+; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC]]
+; CHECK-NEXT: br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
+; CHECK: [[SCALAR_PH]]:
+; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ]
+; CHECK-NEXT: br label %[[LOOP:.*]]
+; CHECK: [[LOOP]]:
+; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[LOOP]] ]
+; CHECK-NEXT: [[GEP_A:%.*]] = getelementptr i32, ptr [[SRC_A]], i64 [[IV]]
+; CHECK-NEXT: [[LOAD_A:%.*]] = load i64, ptr [[GEP_A]], align 4
+; CHECK-NEXT: [[D:%.*]] = udiv i64 [[LOAD_A]], 2
+; CHECK-NEXT: [[GEP_B:%.*]] = getelementptr i32, ptr [[SRC_B]], i64 [[D]]
+; CHECK-NEXT: [[LOAD_B:%.*]] = load i32, ptr [[GEP_B]], align 4
+; CHECK-NEXT: [[GEP_DST:%.*]] = getelementptr i32, ptr [[DST]], i64 [[IV]]
+; CHECK-NEXT: store i32 [[LOAD_B]], ptr [[GEP_DST]], align 4
+; CHECK-NEXT: [[IV_NEXT]] = add i64 [[IV]], 1
+; CHECK-NEXT: [[EXIT_COND:%.*]] = icmp eq i64 [[IV]], [[N]]
+; CHECK-NEXT: br i1 [[EXIT_COND]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP3:![0-9]+]]
+; CHECK: [[EXIT]]:
+; CHECK-NEXT: ret void
+;
+entry:
+ br label %loop
+
+loop:
+ %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
+ %gep.a = getelementptr i32, ptr %src.a, i64 %iv
+ %load.a = load i64, ptr %gep.a
+ %d = udiv i64 %load.a, 2
+ %gep.b = getelementptr i32, ptr %src.b, i64 %d
+ %load.b = load i32, ptr %gep.b, align 4
+ %gep.dst = getelementptr i32, ptr %dst, i64 %iv
+ store i32 %load.b, ptr %gep.dst
+ %iv.next = add i64 %iv, 1
+ %exit.cond = icmp eq i64 %iv, %n
+ br i1 %exit.cond, label %exit, label %loop
+
+exit:
+ ret void
+}
>From a8ecec287eba1b23f46b5a42560b23a31c40b76c Mon Sep 17 00:00:00 2001
From: Ramkumar Ramachandra <ramkumar.ramachandra at codasip.com>
Date: Mon, 6 Oct 2025 10:46:51 +0100
Subject: [PATCH 2/2] [LV] Absorb test into existing file
---
.../uniform_across_vf_induction1_div_urem.ll | 103 ++++++++++++++++++
.../LoopVectorize/uniformity-rewriter.ll | 96 ----------------
2 files changed, 103 insertions(+), 96 deletions(-)
delete mode 100644 llvm/test/Transforms/LoopVectorize/uniformity-rewriter.ll
diff --git a/llvm/test/Transforms/LoopVectorize/uniform_across_vf_induction1_div_urem.ll b/llvm/test/Transforms/LoopVectorize/uniform_across_vf_induction1_div_urem.ll
index 9ed22400b7055..9357adf85b6a6 100644
--- a/llvm/test/Transforms/LoopVectorize/uniform_across_vf_induction1_div_urem.ll
+++ b/llvm/test/Transforms/LoopVectorize/uniform_across_vf_induction1_div_urem.ll
@@ -273,3 +273,106 @@ loop:
exit:
ret void
}
+
+define void @ld_div2_ld_scevunknown_nonuniform(ptr %src.a, ptr noalias %src.b, ptr noalias %dst) {
+; CHECK-LABEL: define void @ld_div2_ld_scevunknown_nonuniform
+; CHECK-SAME: (ptr [[SRC_A:%.*]], ptr noalias [[SRC_B:%.*]], ptr noalias [[DST:%.*]]) {
+; CHECK-NEXT: entry:
+; CHECK-NEXT: br 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: [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; CHECK-NEXT: [[TMP1:%.*]] = add i64 [[INDEX]], 1
+; CHECK-NEXT: [[TMP2:%.*]] = add i64 [[INDEX]], 2
+; CHECK-NEXT: [[TMP3:%.*]] = add i64 [[INDEX]], 3
+; CHECK-NEXT: [[TMP4:%.*]] = add i64 [[INDEX]], 4
+; CHECK-NEXT: [[TMP5:%.*]] = add i64 [[INDEX]], 5
+; CHECK-NEXT: [[TMP6:%.*]] = add i64 [[INDEX]], 6
+; CHECK-NEXT: [[TMP7:%.*]] = add i64 [[INDEX]], 7
+; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[SRC_A]], i64 [[TMP0]]
+; CHECK-NEXT: [[TMP9:%.*]] = getelementptr i32, ptr [[SRC_A]], i64 [[TMP1]]
+; CHECK-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[SRC_A]], i64 [[TMP2]]
+; CHECK-NEXT: [[TMP11:%.*]] = getelementptr i32, ptr [[SRC_A]], i64 [[TMP3]]
+; CHECK-NEXT: [[TMP12:%.*]] = getelementptr i32, ptr [[SRC_A]], i64 [[TMP4]]
+; CHECK-NEXT: [[TMP13:%.*]] = getelementptr i32, ptr [[SRC_A]], i64 [[TMP5]]
+; CHECK-NEXT: [[TMP14:%.*]] = getelementptr i32, ptr [[SRC_A]], i64 [[TMP6]]
+; CHECK-NEXT: [[TMP15:%.*]] = getelementptr i32, ptr [[SRC_A]], i64 [[TMP7]]
+; CHECK-NEXT: [[TMP16:%.*]] = load i64, ptr [[TMP8]], align 4
+; CHECK-NEXT: [[TMP17:%.*]] = load i64, ptr [[TMP9]], align 4
+; CHECK-NEXT: [[TMP18:%.*]] = load i64, ptr [[TMP10]], align 4
+; CHECK-NEXT: [[TMP19:%.*]] = load i64, ptr [[TMP11]], align 4
+; CHECK-NEXT: [[TMP20:%.*]] = load i64, ptr [[TMP12]], align 4
+; CHECK-NEXT: [[TMP21:%.*]] = load i64, ptr [[TMP13]], align 4
+; CHECK-NEXT: [[TMP22:%.*]] = load i64, ptr [[TMP14]], align 4
+; CHECK-NEXT: [[TMP23:%.*]] = load i64, ptr [[TMP15]], align 4
+; CHECK-NEXT: [[TMP24:%.*]] = insertelement <8 x i64> poison, i64 [[TMP16]], i32 0
+; CHECK-NEXT: [[TMP25:%.*]] = insertelement <8 x i64> [[TMP24]], i64 [[TMP17]], i32 1
+; CHECK-NEXT: [[TMP26:%.*]] = insertelement <8 x i64> [[TMP25]], i64 [[TMP18]], i32 2
+; CHECK-NEXT: [[TMP27:%.*]] = insertelement <8 x i64> [[TMP26]], i64 [[TMP19]], i32 3
+; CHECK-NEXT: [[TMP28:%.*]] = insertelement <8 x i64> [[TMP27]], i64 [[TMP20]], i32 4
+; CHECK-NEXT: [[TMP29:%.*]] = insertelement <8 x i64> [[TMP28]], i64 [[TMP21]], i32 5
+; CHECK-NEXT: [[TMP30:%.*]] = insertelement <8 x i64> [[TMP29]], i64 [[TMP22]], i32 6
+; CHECK-NEXT: [[TMP31:%.*]] = insertelement <8 x i64> [[TMP30]], i64 [[TMP23]], i32 7
+; CHECK-NEXT: [[TMP32:%.*]] = udiv <8 x i64> [[TMP31]], splat (i64 2)
+; CHECK-NEXT: [[TMP33:%.*]] = extractelement <8 x i64> [[TMP32]], i32 0
+; CHECK-NEXT: [[TMP34:%.*]] = getelementptr i32, ptr [[SRC_B]], i64 [[TMP33]]
+; CHECK-NEXT: [[TMP35:%.*]] = extractelement <8 x i64> [[TMP32]], i32 1
+; CHECK-NEXT: [[TMP36:%.*]] = getelementptr i32, ptr [[SRC_B]], i64 [[TMP35]]
+; CHECK-NEXT: [[TMP37:%.*]] = extractelement <8 x i64> [[TMP32]], i32 2
+; CHECK-NEXT: [[TMP38:%.*]] = getelementptr i32, ptr [[SRC_B]], i64 [[TMP37]]
+; CHECK-NEXT: [[TMP39:%.*]] = extractelement <8 x i64> [[TMP32]], i32 3
+; CHECK-NEXT: [[TMP40:%.*]] = getelementptr i32, ptr [[SRC_B]], i64 [[TMP39]]
+; CHECK-NEXT: [[TMP41:%.*]] = extractelement <8 x i64> [[TMP32]], i32 4
+; CHECK-NEXT: [[TMP42:%.*]] = getelementptr i32, ptr [[SRC_B]], i64 [[TMP41]]
+; CHECK-NEXT: [[TMP43:%.*]] = extractelement <8 x i64> [[TMP32]], i32 5
+; CHECK-NEXT: [[TMP44:%.*]] = getelementptr i32, ptr [[SRC_B]], i64 [[TMP43]]
+; CHECK-NEXT: [[TMP45:%.*]] = extractelement <8 x i64> [[TMP32]], i32 6
+; CHECK-NEXT: [[TMP46:%.*]] = getelementptr i32, ptr [[SRC_B]], i64 [[TMP45]]
+; CHECK-NEXT: [[TMP47:%.*]] = extractelement <8 x i64> [[TMP32]], i32 7
+; CHECK-NEXT: [[TMP48:%.*]] = getelementptr i32, ptr [[SRC_B]], i64 [[TMP47]]
+; CHECK-NEXT: [[TMP49:%.*]] = load i32, ptr [[TMP34]], align 4
+; CHECK-NEXT: [[TMP50:%.*]] = load i32, ptr [[TMP36]], align 4
+; CHECK-NEXT: [[TMP51:%.*]] = load i32, ptr [[TMP38]], align 4
+; CHECK-NEXT: [[TMP52:%.*]] = load i32, ptr [[TMP40]], align 4
+; CHECK-NEXT: [[TMP53:%.*]] = load i32, ptr [[TMP42]], align 4
+; CHECK-NEXT: [[TMP54:%.*]] = load i32, ptr [[TMP44]], align 4
+; CHECK-NEXT: [[TMP55:%.*]] = load i32, ptr [[TMP46]], align 4
+; CHECK-NEXT: [[TMP56:%.*]] = load i32, ptr [[TMP48]], align 4
+; CHECK-NEXT: [[TMP57:%.*]] = insertelement <8 x i32> poison, i32 [[TMP49]], i32 0
+; CHECK-NEXT: [[TMP58:%.*]] = insertelement <8 x i32> [[TMP57]], i32 [[TMP50]], i32 1
+; CHECK-NEXT: [[TMP59:%.*]] = insertelement <8 x i32> [[TMP58]], i32 [[TMP51]], i32 2
+; CHECK-NEXT: [[TMP60:%.*]] = insertelement <8 x i32> [[TMP59]], i32 [[TMP52]], i32 3
+; CHECK-NEXT: [[TMP61:%.*]] = insertelement <8 x i32> [[TMP60]], i32 [[TMP53]], i32 4
+; CHECK-NEXT: [[TMP62:%.*]] = insertelement <8 x i32> [[TMP61]], i32 [[TMP54]], i32 5
+; CHECK-NEXT: [[TMP63:%.*]] = insertelement <8 x i32> [[TMP62]], i32 [[TMP55]], i32 6
+; CHECK-NEXT: [[TMP64:%.*]] = insertelement <8 x i32> [[TMP63]], i32 [[TMP56]], i32 7
+; CHECK-NEXT: [[TMP65:%.*]] = getelementptr i32, ptr [[DST]], i64 [[TMP0]]
+; CHECK-NEXT: store <8 x i32> [[TMP64]], ptr [[TMP65]], align 4
+; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
+; CHECK-NEXT: [[TMP66:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
+; CHECK-NEXT: br i1 [[TMP66]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
+; CHECK: middle.block:
+; CHECK-NEXT: br label [[SCALAR_PH:%.*]]
+; CHECK: scalar.ph:
+;
+entry:
+ br label %loop
+
+loop:
+ %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
+ %gep.a = getelementptr i32, ptr %src.a, i64 %iv
+ %load.a = load i64, ptr %gep.a
+ %d = udiv i64 %load.a, 2
+ %gep.b = getelementptr i32, ptr %src.b, i64 %d
+ %load.b = load i32, ptr %gep.b
+ %gep.dst = getelementptr i32, ptr %dst, i64 %iv
+ store i32 %load.b, ptr %gep.dst
+ %iv.next = add i64 %iv, 1
+ %exit.cond = icmp eq i64 %iv, 1000
+ br i1 %exit.cond, label %exit, label %loop
+
+exit:
+ ret void
+}
diff --git a/llvm/test/Transforms/LoopVectorize/uniformity-rewriter.ll b/llvm/test/Transforms/LoopVectorize/uniformity-rewriter.ll
deleted file mode 100644
index f4a1ffebe8cfb..0000000000000
--- a/llvm/test/Transforms/LoopVectorize/uniformity-rewriter.ll
+++ /dev/null
@@ -1,96 +0,0 @@
-; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --version 6
-; RUN: opt -passes=loop-vectorize -force-vector-width=4 -S %s | FileCheck %s
-
-; Test to exercise the uniformity rewriter.
-
-define void @uniformityrew(ptr %src.a, ptr noalias %src.b, ptr noalias %dst, i64 %n) {
-; CHECK-LABEL: define void @uniformityrew(
-; CHECK-SAME: ptr [[SRC_A:%.*]], ptr noalias [[SRC_B:%.*]], ptr noalias [[DST:%.*]], i64 [[N:%.*]]) {
-; CHECK-NEXT: [[ENTRY:.*]]:
-; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[N]], 1
-; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP0]], 4
-; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]]
-; CHECK: [[VECTOR_PH]]:
-; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[TMP0]], 4
-; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF]]
-; CHECK-NEXT: br label %[[VECTOR_BODY:.*]]
-; CHECK: [[VECTOR_BODY]]:
-; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
-; CHECK-NEXT: [[TMP1:%.*]] = add i64 [[INDEX]], 0
-; CHECK-NEXT: [[TMP2:%.*]] = add i64 [[INDEX]], 1
-; CHECK-NEXT: [[TMP3:%.*]] = add i64 [[INDEX]], 2
-; CHECK-NEXT: [[TMP4:%.*]] = add i64 [[INDEX]], 3
-; CHECK-NEXT: [[TMP5:%.*]] = getelementptr i32, ptr [[SRC_A]], i64 [[TMP1]]
-; CHECK-NEXT: [[TMP6:%.*]] = getelementptr i32, ptr [[SRC_A]], i64 [[TMP2]]
-; CHECK-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[SRC_A]], i64 [[TMP3]]
-; CHECK-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[SRC_A]], i64 [[TMP4]]
-; CHECK-NEXT: [[TMP9:%.*]] = load i64, ptr [[TMP5]], align 4
-; CHECK-NEXT: [[TMP10:%.*]] = load i64, ptr [[TMP6]], align 4
-; CHECK-NEXT: [[TMP11:%.*]] = load i64, ptr [[TMP7]], align 4
-; CHECK-NEXT: [[TMP12:%.*]] = load i64, ptr [[TMP8]], align 4
-; CHECK-NEXT: [[TMP13:%.*]] = insertelement <4 x i64> poison, i64 [[TMP9]], i32 0
-; CHECK-NEXT: [[TMP14:%.*]] = insertelement <4 x i64> [[TMP13]], i64 [[TMP10]], i32 1
-; CHECK-NEXT: [[TMP15:%.*]] = insertelement <4 x i64> [[TMP14]], i64 [[TMP11]], i32 2
-; CHECK-NEXT: [[TMP16:%.*]] = insertelement <4 x i64> [[TMP15]], i64 [[TMP12]], i32 3
-; CHECK-NEXT: [[TMP17:%.*]] = udiv <4 x i64> [[TMP16]], splat (i64 2)
-; CHECK-NEXT: [[TMP18:%.*]] = extractelement <4 x i64> [[TMP17]], i32 0
-; CHECK-NEXT: [[TMP19:%.*]] = getelementptr i32, ptr [[SRC_B]], i64 [[TMP18]]
-; CHECK-NEXT: [[TMP20:%.*]] = extractelement <4 x i64> [[TMP17]], i32 1
-; CHECK-NEXT: [[TMP21:%.*]] = getelementptr i32, ptr [[SRC_B]], i64 [[TMP20]]
-; CHECK-NEXT: [[TMP22:%.*]] = extractelement <4 x i64> [[TMP17]], i32 2
-; CHECK-NEXT: [[TMP23:%.*]] = getelementptr i32, ptr [[SRC_B]], i64 [[TMP22]]
-; CHECK-NEXT: [[TMP24:%.*]] = extractelement <4 x i64> [[TMP17]], i32 3
-; CHECK-NEXT: [[TMP25:%.*]] = getelementptr i32, ptr [[SRC_B]], i64 [[TMP24]]
-; CHECK-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP19]], align 4
-; CHECK-NEXT: [[TMP27:%.*]] = load i32, ptr [[TMP21]], align 4
-; CHECK-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP23]], align 4
-; CHECK-NEXT: [[TMP29:%.*]] = load i32, ptr [[TMP25]], align 4
-; CHECK-NEXT: [[TMP30:%.*]] = insertelement <4 x i32> poison, i32 [[TMP26]], i32 0
-; CHECK-NEXT: [[TMP31:%.*]] = insertelement <4 x i32> [[TMP30]], i32 [[TMP27]], i32 1
-; CHECK-NEXT: [[TMP32:%.*]] = insertelement <4 x i32> [[TMP31]], i32 [[TMP28]], i32 2
-; CHECK-NEXT: [[TMP33:%.*]] = insertelement <4 x i32> [[TMP32]], i32 [[TMP29]], i32 3
-; CHECK-NEXT: [[TMP34:%.*]] = getelementptr i32, ptr [[DST]], i64 [[TMP1]]
-; CHECK-NEXT: store <4 x i32> [[TMP33]], ptr [[TMP34]], align 4
-; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-NEXT: [[TMP35:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[TMP35]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
-; CHECK: [[MIDDLE_BLOCK]]:
-; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC]]
-; CHECK-NEXT: br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]]
-; CHECK: [[SCALAR_PH]]:
-; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ]
-; CHECK-NEXT: br label %[[LOOP:.*]]
-; CHECK: [[LOOP]]:
-; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[LOOP]] ]
-; CHECK-NEXT: [[GEP_A:%.*]] = getelementptr i32, ptr [[SRC_A]], i64 [[IV]]
-; CHECK-NEXT: [[LOAD_A:%.*]] = load i64, ptr [[GEP_A]], align 4
-; CHECK-NEXT: [[D:%.*]] = udiv i64 [[LOAD_A]], 2
-; CHECK-NEXT: [[GEP_B:%.*]] = getelementptr i32, ptr [[SRC_B]], i64 [[D]]
-; CHECK-NEXT: [[LOAD_B:%.*]] = load i32, ptr [[GEP_B]], align 4
-; CHECK-NEXT: [[GEP_DST:%.*]] = getelementptr i32, ptr [[DST]], i64 [[IV]]
-; CHECK-NEXT: store i32 [[LOAD_B]], ptr [[GEP_DST]], align 4
-; CHECK-NEXT: [[IV_NEXT]] = add i64 [[IV]], 1
-; CHECK-NEXT: [[EXIT_COND:%.*]] = icmp eq i64 [[IV]], [[N]]
-; CHECK-NEXT: br i1 [[EXIT_COND]], label %[[EXIT]], label %[[LOOP]], !llvm.loop [[LOOP3:![0-9]+]]
-; CHECK: [[EXIT]]:
-; CHECK-NEXT: ret void
-;
-entry:
- br label %loop
-
-loop:
- %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
- %gep.a = getelementptr i32, ptr %src.a, i64 %iv
- %load.a = load i64, ptr %gep.a
- %d = udiv i64 %load.a, 2
- %gep.b = getelementptr i32, ptr %src.b, i64 %d
- %load.b = load i32, ptr %gep.b, align 4
- %gep.dst = getelementptr i32, ptr %dst, i64 %iv
- store i32 %load.b, ptr %gep.dst
- %iv.next = add i64 %iv, 1
- %exit.cond = icmp eq i64 %iv, %n
- br i1 %exit.cond, label %exit, label %loop
-
-exit:
- ret void
-}
More information about the llvm-commits
mailing list