[SLPVectorizer][NFC] A test for widening constant strided loads. #160552
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: Mikhail Gudim (mgudim) ChangesPrecommit a test. Full diff: https://github.com/llvm/llvm-project/pull/160552.diff 1 Files Affected:
diff --git a/llvm/test/Transforms/SLPVectorizer/RISCV/basic-strided-loads.ll b/llvm/test/Transforms/SLPVectorizer/RISCV/basic-strided-loads.ll
index 645dbc49269f0..b309be84f42cf 100644
--- a/llvm/test/Transforms/SLPVectorizer/RISCV/basic-strided-loads.ll
+++ b/llvm/test/Transforms/SLPVectorizer/RISCV/basic-strided-loads.ll
@@ -527,23 +527,14 @@ define void @rt_stride_1_with_reordering(ptr %pl, i64 %stride, ptr %ps) {
ret void
}
-; TODO: We want to generate this code:
-; define void @constant_stride_widen_no_reordering(ptr %pl, i64 %stride, ptr %ps) {
-; %gep_l0 = getelementptr inbounds i8, ptr %pl, i64 %offset0
-; %gep_s0 = getelementptr inbounds i8, ptr %ps, i64 0
-; %strided_load = call <4 x i32> @llvm.experimental.vp.strided.load.v4i32.p0.i64(ptr align 16 %gep_l0, i64 8, <4 x i1> splat (i1 true), i32 4)
-; %bitcast_ = bitcast <4 x i32> %strided_load to <16 x i8>
-; store <16 x i8> %bitcast_, ptr %gep_s0, align 16
-; ret void
-; }
-define void @constant_stride_widen_no_reordering(ptr %pl, i64 %stride, ptr %ps) {
-; CHECK-LABEL: define void @constant_stride_widen_no_reordering(
+define void @constant_stride_masked_no_reordering(ptr %pl, i64 %stride, ptr %ps) {
+; CHECK-LABEL: define void @constant_stride_masked_no_reordering(
; CHECK-SAME: ptr [[PL:%.*]], i64 [[STRIDE:%.*]], ptr [[PS:%.*]]) #[[ATTR0]] {
; CHECK-NEXT: [[GEP_L0:%.*]] = getelementptr inbounds i8, ptr [[PL]], i64 0
; CHECK-NEXT: [[GEP_S0:%.*]] = getelementptr inbounds i8, ptr [[PS]], i64 0
; CHECK-NEXT: [[TMP1:%.*]] = call <28 x i8> @llvm.masked.load.v28i8.p0(ptr [[GEP_L0]], i32 16, <28 x i1> <i1 true, i1 true, i1 true, i1 true, i1 false, i1 false, i1 false, i1 false, i1 true, i1 true, i1 true, i1 true, i1 false, i1 false, i1 false, i1 false, i1 true, i1 true, i1 true, i1 true, i1 false, i1 false, i1 false, i1 false, i1 true, i1 true, i1 true, i1 true>, <28 x i8> poison)
-; CHECK-NEXT: [[TMP8:%.*]] = shufflevector <28 x i8> [[TMP1]], <28 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 18, i32 19, i32 24, i32 25, i32 26, i32 27>
-; CHECK-NEXT: store <16 x i8> [[TMP8]], ptr [[GEP_S0]], align 16
+; CHECK-NEXT: [[TMP2:%.*]] = shufflevector <28 x i8> [[TMP1]], <28 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 18, i32 19, i32 24, i32 25, i32 26, i32 27>
+; CHECK-NEXT: store <16 x i8> [[TMP2]], ptr [[GEP_S0]], align 16
; CHECK-NEXT: ret void
;
%gep_l0 = getelementptr inbounds i8, ptr %pl, i64 0
@@ -617,6 +608,107 @@ define void @constant_stride_widen_no_reordering(ptr %pl, i64 %stride, ptr %ps)
ret void
}
+; TODO: We want to generate this code:
+; define void @constant_stride_widen_no_reordering(ptr %pl, i64 %stride, ptr %ps) #0 {
+; %gep_l0 = getelementptr inbounds i8, ptr %pl, i64 0
+; %gep_s0 = getelementptr inbounds i8, ptr %ps, i64 0
+; %1 = call <4 x i32> @llvm.experimental.vp.strided.load.v4i32.p0.i64(ptr align 16 %gep_l0, i64 100, <4 x i1> splat (i1 true), i32 4)
+; %2 = bitcast <4 x i32> %1 to <16 x i8>
+; store <16 x i8> %2, ptr %gep_s0, align 16
+; ret void
+; }
+define void @constant_stride_widen_no_reordering(ptr %pl, i64 %stride, ptr %ps) {
+; CHECK-LABEL: define void @constant_stride_widen_no_reordering(
+; CHECK-SAME: ptr [[PL:%.*]], i64 [[STRIDE:%.*]], ptr [[PS:%.*]]) #[[ATTR0]] {
+; CHECK-NEXT: [[GEP_L0:%.*]] = getelementptr inbounds i8, ptr [[PL]], i64 0
+; CHECK-NEXT: [[GEP_L4:%.*]] = getelementptr inbounds i8, ptr [[PL]], i64 100
+; CHECK-NEXT: [[GEP_L8:%.*]] = getelementptr inbounds i8, ptr [[PL]], i64 200
+; CHECK-NEXT: [[GEP_L12:%.*]] = getelementptr inbounds i8, ptr [[PL]], i64 300
+; CHECK-NEXT: [[GEP_S0:%.*]] = getelementptr inbounds i8, ptr [[PS]], i64 0
+; CHECK-NEXT: [[TMP1:%.*]] = load <4 x i8>, ptr [[GEP_L0]], align 16
+; CHECK-NEXT: [[TMP2:%.*]] = load <4 x i8>, ptr [[GEP_L4]], align 16
+; CHECK-NEXT: [[TMP3:%.*]] = load <4 x i8>, ptr [[GEP_L8]], align 16
+; CHECK-NEXT: [[TMP4:%.*]] = load <4 x i8>, ptr [[GEP_L12]], align 16
+; CHECK-NEXT: [[TMP5:%.*]] = shufflevector <4 x i8> [[TMP1]], <4 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
+; CHECK-NEXT: [[TMP6:%.*]] = shufflevector <4 x i8> [[TMP2]], <4 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
+; CHECK-NEXT: [[TMP7:%.*]] = shufflevector <4 x i8> [[TMP1]], <4 x i8> [[TMP2]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
+; CHECK-NEXT: [[TMP11:%.*]] = shufflevector <4 x i8> [[TMP3]], <4 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
+; CHECK-NEXT: [[TMP9:%.*]] = shufflevector <16 x i8> [[TMP7]], <16 x i8> [[TMP11]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 poison, i32 poison, i32 poison, i32 poison>
+; CHECK-NEXT: [[TMP10:%.*]] = shufflevector <4 x i8> [[TMP4]], <4 x i8> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
+; CHECK-NEXT: [[TMP8:%.*]] = shufflevector <16 x i8> [[TMP9]], <16 x i8> [[TMP10]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 16, i32 17, i32 18, i32 19>
+; CHECK-NEXT: store <16 x i8> [[TMP8]], ptr [[GEP_S0]], align 16
+; CHECK-NEXT: ret void
+;
+ %gep_l0 = getelementptr inbounds i8, ptr %pl, i64 0
+ %gep_l1 = getelementptr inbounds i8, ptr %pl, i64 1
+ %gep_l2 = getelementptr inbounds i8, ptr %pl, i64 2
+ %gep_l3 = getelementptr inbounds i8, ptr %pl, i64 3
+ %gep_l4 = getelementptr inbounds i8, ptr %pl, i64 100
+ %gep_l5 = getelementptr inbounds i8, ptr %pl, i64 101
+ %gep_l6 = getelementptr inbounds i8, ptr %pl, i64 102
+ %gep_l7 = getelementptr inbounds i8, ptr %pl, i64 103
+ %gep_l8 = getelementptr inbounds i8, ptr %pl, i64 200
+ %gep_l9 = getelementptr inbounds i8, ptr %pl, i64 201
+ %gep_l10 = getelementptr inbounds i8, ptr %pl, i64 202
+ %gep_l11 = getelementptr inbounds i8, ptr %pl, i64 203
+ %gep_l12 = getelementptr inbounds i8, ptr %pl, i64 300
+ %gep_l13 = getelementptr inbounds i8, ptr %pl, i64 301
+ %gep_l14 = getelementptr inbounds i8, ptr %pl, i64 302
+ %gep_l15 = getelementptr inbounds i8, ptr %pl, i64 303
+
+ %load0 = load i8, ptr %gep_l0 , align 16
+ %load1 = load i8, ptr %gep_l1 , align 16
+ %load2 = load i8, ptr %gep_l2 , align 16
+ %load3 = load i8, ptr %gep_l3 , align 16
+ %load4 = load i8, ptr %gep_l4 , align 16
+ %load5 = load i8, ptr %gep_l5 , align 16
+ %load6 = load i8, ptr %gep_l6 , align 16
+ %load7 = load i8, ptr %gep_l7 , align 16
+ %load8 = load i8, ptr %gep_l8 , align 16
+ %load9 = load i8, ptr %gep_l9 , align 16
+ %load10 = load i8, ptr %gep_l10, align 16
+ %load11 = load i8, ptr %gep_l11, align 16
+ %load12 = load i8, ptr %gep_l12, align 16
+ %load13 = load i8, ptr %gep_l13, align 16
+ %load14 = load i8, ptr %gep_l14, align 16
+ %load15 = load i8, ptr %gep_l15, align 16
+
+ %gep_s0 = getelementptr inbounds i8, ptr %ps, i64 0
+ %gep_s1 = getelementptr inbounds i8, ptr %ps, i64 1
+ %gep_s2 = getelementptr inbounds i8, ptr %ps, i64 2
+ %gep_s3 = getelementptr inbounds i8, ptr %ps, i64 3
+ %gep_s4 = getelementptr inbounds i8, ptr %ps, i64 4
+ %gep_s5 = getelementptr inbounds i8, ptr %ps, i64 5
+ %gep_s6 = getelementptr inbounds i8, ptr %ps, i64 6
+ %gep_s7 = getelementptr inbounds i8, ptr %ps, i64 7
+ %gep_s8 = getelementptr inbounds i8, ptr %ps, i64 8
+ %gep_s9 = getelementptr inbounds i8, ptr %ps, i64 9
+ %gep_s10 = getelementptr inbounds i8, ptr %ps, i64 10
+ %gep_s11 = getelementptr inbounds i8, ptr %ps, i64 11
+ %gep_s12 = getelementptr inbounds i8, ptr %ps, i64 12
+ %gep_s13 = getelementptr inbounds i8, ptr %ps, i64 13
+ %gep_s14 = getelementptr inbounds i8, ptr %ps, i64 14
+ %gep_s15 = getelementptr inbounds i8, ptr %ps, i64 15
+
+ store i8 %load0, ptr %gep_s0, align 16
+ store i8 %load1, ptr %gep_s1, align 16
+ store i8 %load2, ptr %gep_s2, align 16
+ store i8 %load3, ptr %gep_s3, align 16
+ store i8 %load4, ptr %gep_s4, align 16
+ store i8 %load5, ptr %gep_s5, align 16
+ store i8 %load6, ptr %gep_s6, align 16
+ store i8 %load7, ptr %gep_s7, align 16
+ store i8 %load8, ptr %gep_s8, align 16
+ store i8 %load9, ptr %gep_s9, align 16
+ store i8 %load10, ptr %gep_s10, align 16
+ store i8 %load11, ptr %gep_s11, align 16
+ store i8 %load12, ptr %gep_s12, align 16
+ store i8 %load13, ptr %gep_s13, align 16
+ store i8 %load14, ptr %gep_s14, align 16
+ store i8 %load15, ptr %gep_s15, align 16
+
+ ret void
+}
; TODO: We want to generate this code:
; define void @rt_stride_widen_no_reordering(ptr %pl, i64 %stride, ptr %ps) {
; %gep_l0 = getelementptr inbounds i8, ptr %pl, i64 %offset0
|
| %gep_s14 = getelementptr inbounds i8, ptr %ps, i64 14 | ||
| %gep_s15 = getelementptr inbounds i8, ptr %ps, i64 15 | ||
|
|
||
| store i8 %load0, ptr %gep_s0, align 16 |
There was a problem hiding this comment.
I copy-pasted from other functions. Should I remove alignment everywhere in basic-strided-loads.ll?
There was a problem hiding this comment.
Just it is too much for i8, they usually have align 1
There was a problem hiding this comment.
I updated the alignment (force-pushed by mistake)
Precommit a test.
mgudim
force-pushed
the
test_const_strided_load
branch
from
2ece844 to
e6bb662
Compare
constant_stride_masked_no_reordering
alexey-bataev
changed the title
alexey-bataev
approved these changes
Oct 1, 2025
mahesh-attarde
pushed a commit
to mahesh-attarde/llvm-project
that referenced
this pull request
Oct 3, 2025
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.
Precommit a test.