Skip to content

Commit 94d13aa

Browse files
alexey-bataevalexey-bataev
committed
[𝘀𝗽𝗿] initial version
Created using spr 1.3.5
1 parent 7451e4c commit 94d13aa

11 files changed

+1369
-229
lines changed

‎llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp

Lines changed: 1327 additions & 169 deletions
Large diffs are not rendered by default.

‎llvm/test/Transforms/SLPVectorizer/X86/buildvector-schedule-for-subvector.ll

Lines changed: 1 addition & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -4,20 +4,14 @@
44
define void @test() {
55
; CHECK-LABEL: define void @test() {
66
; CHECK-NEXT: [[BB:.*:]]
7-
; CHECK-NEXT: [[ADD:%.*]] = add i32 1, 0
8-
; CHECK-NEXT: [[TMP0:%.*]] = insertelement <4 x i32> <i32 0, i32 0, i32 0, i32 poison>, i32 [[ADD]], i32 3
9-
; CHECK-NEXT: [[TMP1:%.*]] = icmp ult <4 x i32> [[TMP0]], zeroinitializer
10-
; CHECK-NEXT: [[ICMP:%.*]] = icmp samesign ult i32 0, 0
11-
; CHECK-NEXT: [[SELECT:%.*]] = select i1 [[ICMP]], i32 0, i32 0
7+
; CHECK-NEXT: [[SELECT:%.*]] = select i1 false, i32 0, i32 0
128
; CHECK-NEXT: [[ZEXT:%.*]] = zext i32 [[SELECT]] to i64
139
; CHECK-NEXT: [[GETELEMENTPTR:%.*]] = getelementptr ptr addrspace(1), ptr addrspace(1) null, i64 [[ZEXT]]
1410
; CHECK-NEXT: store ptr addrspace(1) null, ptr addrspace(1) [[GETELEMENTPTR]], align 8
1511
; CHECK-NEXT: store volatile i32 0, ptr addrspace(1) null, align 4
1612
; CHECK-NEXT: [[CALL:%.*]] = call i32 null(<2 x double> zeroinitializer)
1713
; CHECK-NEXT: [[TMP2:%.*]] = insertelement <4 x i32> <i32 0, i32 0, i32 0, i32 poison>, i32 [[CALL]], i32 3
1814
; CHECK-NEXT: [[TMP3:%.*]] = icmp eq <4 x i32> [[TMP2]], zeroinitializer
19-
; CHECK-NEXT: [[TMP4:%.*]] = shufflevector <4 x i1> [[TMP3]], <4 x i1> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 poison, i32 poison, i32 poison, i32 poison>
20-
; CHECK-NEXT: [[TMP5:%.*]] = call <8 x i1> @llvm.vector.insert.v8i1.v4i1(<8 x i1> [[TMP4]], <4 x i1> [[TMP1]], i64 4)
2115
; CHECK-NEXT: ret void
2216
;
2317
bb:

‎llvm/test/Transforms/SLPVectorizer/X86/cast-operand-extracted.ll

Lines changed: 4 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -19,13 +19,13 @@ define void @test(ptr %0, i32 %add651) {
1919
; CHECK-NEXT: [[ARRAYIDX660:%.*]] = getelementptr i8, ptr [[TMP4]], i64 7800
2020
; CHECK-NEXT: [[ARRAYIDX689:%.*]] = getelementptr i8, ptr [[TMP4]], i64 7816
2121
; CHECK-NEXT: [[TMP6:%.*]] = add <2 x i32> [[TMP3]], splat (i32 1)
22-
; CHECK-NEXT: [[TMP8:%.*]] = add <2 x i32> [[TMP6]], [[TMP7]]
23-
; CHECK-NEXT: [[TMP9:%.*]] = insertelement <2 x i32> <i32 1, i32 poison>, i32 [[TMP5]], i32 1
24-
; CHECK-NEXT: [[TMP10:%.*]] = add <2 x i32> [[TMP8]], [[TMP9]]
22+
; CHECK-NEXT: [[TMP10:%.*]] = add <2 x i32> [[TMP6]], [[TMP7]]
2523
; CHECK-NEXT: [[TMP11:%.*]] = insertelement <4 x i32> poison, i32 [[ADD651]], i32 0
2624
; CHECK-NEXT: [[TMP13:%.*]] = insertelement <4 x i32> [[TMP11]], i32 [[TMP2]], i32 1
2725
; CHECK-NEXT: [[TMP14:%.*]] = call <4 x i32> @llvm.vector.insert.v4i32.v2i32(<4 x i32> [[TMP13]], <2 x i32> [[TMP10]], i64 2)
28-
; CHECK-NEXT: [[TMP15:%.*]] = lshr <4 x i32> [[TMP14]], splat (i32 1)
26+
; CHECK-NEXT: [[TMP12:%.*]] = insertelement <4 x i32> <i32 0, i32 0, i32 1, i32 poison>, i32 [[TMP5]], i32 3
27+
; CHECK-NEXT: [[TMP19:%.*]] = add <4 x i32> [[TMP14]], [[TMP12]]
28+
; CHECK-NEXT: [[TMP15:%.*]] = lshr <4 x i32> [[TMP19]], splat (i32 1)
2929
; CHECK-NEXT: [[SHR685:%.*]] = lshr i32 [[TMP2]], 1
3030
; CHECK-NEXT: [[TMP16:%.*]] = trunc <4 x i32> [[TMP15]] to <4 x i16>
3131
; CHECK-NEXT: [[CONV686:%.*]] = trunc i32 [[SHR685]] to i16

‎llvm/test/Transforms/SLPVectorizer/X86/full-match-with-poison-scalar.ll

Lines changed: 4 additions & 10 deletions
Original file line numberDiff line numberDiff line change
@@ -7,16 +7,10 @@ define i32 @test() {
77
; CHECK-NEXT: br label %[[FUNC_135_EXIT_I:.*]]
88
; CHECK: [[FUNC_135_EXIT_I]]:
99
; CHECK-NEXT: [[G_228_PROMOTED166_I1105_I:%.*]] = phi i32 [ 0, %[[ENTRY]] ]
10-
; CHECK-NEXT: [[TMP0:%.*]] = insertelement <4 x i32> <i32 poison, i32 0, i32 poison, i32 poison>, i32 [[G_228_PROMOTED166_I1105_I]], i32 0
11-
; CHECK-NEXT: [[TMP1:%.*]] = shufflevector <4 x i32> [[TMP0]], <4 x i32> poison, <4 x i32> <i32 0, i32 0, i32 0, i32 1>
12-
; CHECK-NEXT: [[TMP2:%.*]] = add <4 x i32> zeroinitializer, [[TMP1]]
13-
; CHECK-NEXT: [[TMP3:%.*]] = shufflevector <4 x i32> [[TMP2]], <4 x i32> poison, <12 x i32> <i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1, i32 2, i32 2, i32 2, i32 2, i32 3>
14-
; CHECK-NEXT: [[TMP7:%.*]] = shufflevector <4 x i32> [[TMP0]], <4 x i32> poison, <16 x i32> <i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 0, i32 0, i32 0, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
15-
; CHECK-NEXT: [[TMP5:%.*]] = insertelement <16 x i32> poison, i32 [[G_228_PROMOTED166_I1105_I]], i32 0
16-
; CHECK-NEXT: [[TMP9:%.*]] = shufflevector <16 x i32> [[TMP5]], <16 x i32> poison, <16 x i32> <i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 0, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
17-
; CHECK-NEXT: [[TMP8:%.*]] = shufflevector <16 x i32> [[TMP7]], <16 x i32> [[TMP9]], <16 x i32> <i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 23, i32 8, i32 9, i32 10, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
18-
; CHECK-NEXT: [[TMP6:%.*]] = call <16 x i32> @llvm.vector.insert.v16i32.v12i32(<16 x i32> poison, <12 x i32> [[TMP3]], i64 0)
19-
; CHECK-NEXT: [[TMP11:%.*]] = shufflevector <16 x i32> [[TMP6]], <16 x i32> [[TMP8]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 23, i32 24, i32 25, i32 26, i32 7, i32 8, i32 9, i32 10, i32 11>
10+
; CHECK-NEXT: [[TMP0:%.*]] = insertelement <8 x i32> <i32 poison, i32 poison, i32 poison, i32 poison, i32 0, i32 poison, i32 poison, i32 poison>, i32 [[G_228_PROMOTED166_I1105_I]], i32 0
11+
; CHECK-NEXT: [[TMP1:%.*]] = shufflevector <8 x i32> [[TMP0]], <8 x i32> poison, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 4, i32 poison, i32 poison, i32 poison>
12+
; CHECK-NEXT: [[TMP2:%.*]] = add <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 poison, i32 poison, i32 poison>, [[TMP1]]
13+
; CHECK-NEXT: [[TMP11:%.*]] = shufflevector <8 x i32> [[TMP2]], <8 x i32> poison, <16 x i32> <i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1, i32 2, i32 2, i32 2, i32 2, i32 3, i32 3, i32 3, i32 3, i32 4>
2014
; CHECK-NEXT: [[TMP12:%.*]] = icmp ugt <16 x i32> [[TMP11]], zeroinitializer
2115
; CHECK-NEXT: [[TMP13:%.*]] = icmp ult <16 x i32> [[TMP11]], zeroinitializer
2216
; CHECK-NEXT: [[TMP14:%.*]] = shufflevector <16 x i1> [[TMP12]], <16 x i1> [[TMP13]], <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 12, i32 13, i32 14, i32 31>

‎llvm/test/Transforms/SLPVectorizer/X86/node-outside-used-only.ll

Lines changed: 1 addition & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -4,11 +4,10 @@
44
define i64 @test() {
55
; CHECK-LABEL: define i64 @test() {
66
; CHECK-NEXT: [[BB:.*]]:
7-
; CHECK-NEXT: [[TMP0:%.*]] = insertelement <2 x i32> <i32 0, i32 poison>, i32 0, i32 1
87
; CHECK-NEXT: br label %[[BB1:.*]]
98
; CHECK: [[BB1]]:
109
; CHECK-NEXT: [[TMP1:%.*]] = phi <2 x i32> [ zeroinitializer, %[[BB]] ], [ [[TMP4:%.*]], %[[BB5:.*]] ]
11-
; CHECK-NEXT: [[TMP2:%.*]] = or <2 x i32> [[TMP0]], [[TMP1]]
10+
; CHECK-NEXT: [[TMP2:%.*]] = or <2 x i32> zeroinitializer, [[TMP1]]
1211
; CHECK-NEXT: [[TMP3:%.*]] = shufflevector <2 x i32> [[TMP1]], <2 x i32> [[TMP2]], <2 x i32> <i32 0, i32 3>
1312
; CHECK-NEXT: [[TMP4]] = or <2 x i32> [[TMP3]], zeroinitializer
1413
; CHECK-NEXT: br label %[[BB5]]

‎llvm/test/Transforms/SLPVectorizer/X86/non-schedulable-instructions-become-schedulable.ll

Lines changed: 7 additions & 9 deletions
Original file line numberDiff line numberDiff line change
@@ -7,19 +7,17 @@ define void @test() {
77
; CHECK-NEXT: [[ENTRY:.*:]]
88
; CHECK-NEXT: br label %[[BB1:.*]]
99
; CHECK: [[IF_THEN_I_I:.*]]:
10-
; CHECK-NEXT: br label %[[BB5:.*]]
10+
; CHECK-NEXT: br label %[[BB3:.*]]
1111
; CHECK: [[BB1]]:
1212
; CHECK-NEXT: [[TMP0:%.*]] = zext i1 false to i64
13-
; CHECK-NEXT: [[TMP1:%.*]] = insertelement <2 x i64> <i64 poison, i64 0>, i64 [[TMP0]], i32 0
14-
; CHECK-NEXT: [[TMP2:%.*]] = add <2 x i64> zeroinitializer, [[TMP1]]
15-
; CHECK-NEXT: [[TMP3:%.*]] = call <4 x i64> @llvm.vector.insert.v4i64.v2i64(<4 x i64> <i64 0, i64 0, i64 poison, i64 poison>, <2 x i64> [[TMP2]], i64 2)
16-
; CHECK-NEXT: [[TMP4:%.*]] = call <4 x i64> @llvm.vector.insert.v4i64.v2i64(<4 x i64> <i64 0, i64 0, i64 poison, i64 poison>, <2 x i64> [[TMP2]], i64 2)
17-
; CHECK-NEXT: br i1 false, label %[[BB5]], label %[[BB2:.*]]
18-
; CHECK: [[BB5]]:
19-
; CHECK-NEXT: [[TMP6:%.*]] = phi <4 x i64> [ [[TMP3]], %[[BB1]] ], [ poison, %[[IF_THEN_I_I]] ]
13+
; CHECK-NEXT: [[TMP1:%.*]] = insertelement <4 x i64> <i64 0, i64 0, i64 poison, i64 0>, i64 [[TMP0]], i32 2
14+
; CHECK-NEXT: [[TMP2:%.*]] = add <4 x i64> zeroinitializer, [[TMP1]]
15+
; CHECK-NEXT: br i1 false, label %[[BB3]], label %[[BB2:.*]]
16+
; CHECK: [[BB3]]:
17+
; CHECK-NEXT: [[TMP4:%.*]] = phi <4 x i64> [ [[TMP2]], %[[BB1]] ], [ poison, %[[IF_THEN_I_I]] ]
2018
; CHECK-NEXT: br label %[[BB2]]
2119
; CHECK: [[BB2]]:
22-
; CHECK-NEXT: [[TMP7:%.*]] = phi <4 x i64> [ [[TMP6]], %[[BB5]] ], [ [[TMP4]], %[[BB1]] ]
20+
; CHECK-NEXT: [[TMP7:%.*]] = phi <4 x i64> [ [[TMP4]], %[[BB3]] ], [ [[TMP2]], %[[BB1]] ]
2321
; CHECK-NEXT: store <4 x i64> [[TMP7]], ptr getelementptr inbounds nuw (i8, ptr null, i64 40), align 8
2422
; CHECK-NEXT: ret void
2523
;

‎llvm/test/Transforms/SLPVectorizer/X86/pr35497.ll

Lines changed: 10 additions & 10 deletions
Original file line numberDiff line numberDiff line change
@@ -59,41 +59,41 @@ define void @pr35497() local_unnamed_addr #0 {
5959
; SSE-LABEL: @pr35497(
6060
; SSE-NEXT: entry:
6161
; SSE-NEXT: [[TMP0:%.*]] = load i64, ptr undef, align 1
62-
; SSE-NEXT: [[ADD:%.*]] = add i64 undef, undef
63-
; SSE-NEXT: store i64 [[ADD]], ptr undef, align 1
6462
; SSE-NEXT: [[ARRAYIDX2_2:%.*]] = getelementptr inbounds [0 x i64], ptr undef, i64 0, i64 4
6563
; SSE-NEXT: [[TMP1:%.*]] = insertelement <2 x i64> <i64 poison, i64 undef>, i64 [[TMP0]], i32 0
6664
; SSE-NEXT: [[TMP2:%.*]] = shl <2 x i64> [[TMP1]], splat (i64 2)
6765
; SSE-NEXT: [[TMP3:%.*]] = and <2 x i64> [[TMP2]], splat (i64 20)
6866
; SSE-NEXT: [[TMP4:%.*]] = shufflevector <2 x i64> [[TMP3]], <2 x i64> poison, <2 x i32> <i32 1, i32 0>
6967
; SSE-NEXT: [[TMP5:%.*]] = add nuw nsw <2 x i64> [[TMP4]], zeroinitializer
7068
; SSE-NEXT: store <2 x i64> [[TMP5]], ptr undef, align 1
71-
; SSE-NEXT: [[TMP6:%.*]] = shufflevector <2 x i64> [[TMP5]], <2 x i64> poison, <2 x i32> <i32 1, i32 poison>
72-
; SSE-NEXT: [[TMP7:%.*]] = insertelement <2 x i64> [[TMP6]], i64 [[ADD]], i32 1
69+
; SSE-NEXT: [[ADD:%.*]] = add i64 undef, undef
70+
; SSE-NEXT: [[TMP7:%.*]] = add <2 x i64> [[TMP5]], <i64 undef, i64 0>
71+
; SSE-NEXT: store i64 [[ADD]], ptr undef, align 1
7372
; SSE-NEXT: [[TMP8:%.*]] = shl <2 x i64> [[TMP7]], splat (i64 2)
7473
; SSE-NEXT: [[TMP9:%.*]] = and <2 x i64> [[TMP8]], splat (i64 20)
74+
; SSE-NEXT: [[TMP12:%.*]] = shufflevector <2 x i64> [[TMP9]], <2 x i64> poison, <2 x i32> <i32 1, i32 0>
7575
; SSE-NEXT: [[TMP10:%.*]] = lshr <2 x i64> [[TMP5]], splat (i64 6)
76-
; SSE-NEXT: [[TMP11:%.*]] = add nuw nsw <2 x i64> [[TMP9]], [[TMP10]]
76+
; SSE-NEXT: [[TMP11:%.*]] = add nuw nsw <2 x i64> [[TMP12]], [[TMP10]]
7777
; SSE-NEXT: store <2 x i64> [[TMP11]], ptr [[ARRAYIDX2_2]], align 1
7878
; SSE-NEXT: ret void
7979
;
8080
; AVX-LABEL: @pr35497(
8181
; AVX-NEXT: entry:
8282
; AVX-NEXT: [[TMP0:%.*]] = load i64, ptr undef, align 1
83-
; AVX-NEXT: [[ADD:%.*]] = add i64 undef, undef
84-
; AVX-NEXT: store i64 [[ADD]], ptr undef, align 1
8583
; AVX-NEXT: [[ARRAYIDX2_2:%.*]] = getelementptr inbounds [0 x i64], ptr undef, i64 0, i64 4
8684
; AVX-NEXT: [[TMP1:%.*]] = insertelement <2 x i64> <i64 undef, i64 poison>, i64 [[TMP0]], i32 1
8785
; AVX-NEXT: [[TMP2:%.*]] = shl <2 x i64> [[TMP1]], splat (i64 2)
8886
; AVX-NEXT: [[TMP3:%.*]] = and <2 x i64> [[TMP2]], splat (i64 20)
8987
; AVX-NEXT: [[TMP4:%.*]] = add nuw nsw <2 x i64> [[TMP3]], zeroinitializer
9088
; AVX-NEXT: store <2 x i64> [[TMP4]], ptr undef, align 1
91-
; AVX-NEXT: [[TMP5:%.*]] = shufflevector <2 x i64> [[TMP4]], <2 x i64> poison, <2 x i32> <i32 1, i32 poison>
92-
; AVX-NEXT: [[TMP6:%.*]] = insertelement <2 x i64> [[TMP5]], i64 [[ADD]], i32 1
89+
; AVX-NEXT: [[ADD:%.*]] = add i64 undef, undef
90+
; AVX-NEXT: [[TMP6:%.*]] = add <2 x i64> [[TMP4]], <i64 undef, i64 0>
91+
; AVX-NEXT: store i64 [[ADD]], ptr undef, align 1
9392
; AVX-NEXT: [[TMP7:%.*]] = shl <2 x i64> [[TMP6]], splat (i64 2)
9493
; AVX-NEXT: [[TMP8:%.*]] = and <2 x i64> [[TMP7]], splat (i64 20)
94+
; AVX-NEXT: [[TMP11:%.*]] = shufflevector <2 x i64> [[TMP8]], <2 x i64> poison, <2 x i32> <i32 1, i32 0>
9595
; AVX-NEXT: [[TMP9:%.*]] = lshr <2 x i64> [[TMP4]], splat (i64 6)
96-
; AVX-NEXT: [[TMP10:%.*]] = add nuw nsw <2 x i64> [[TMP8]], [[TMP9]]
96+
; AVX-NEXT: [[TMP10:%.*]] = add nuw nsw <2 x i64> [[TMP11]], [[TMP9]]
9797
; AVX-NEXT: store <2 x i64> [[TMP10]], ptr [[ARRAYIDX2_2]], align 1
9898
; AVX-NEXT: ret void
9999
;

‎llvm/test/Transforms/SLPVectorizer/X86/pr47642.ll

Lines changed: 3 additions & 8 deletions
Original file line numberDiff line numberDiff line change
@@ -6,14 +6,9 @@ target triple = "x86_64-unknown-linux-gnu"
66

77
define <4 x i32> @foo(<4 x i32> %x, i32 %f) {
88
; CHECK-LABEL: @foo(
9-
; CHECK-NEXT: [[VECINIT:%.*]] = insertelement <4 x i32> undef, i32 [[F:%.*]], i32 0
10-
; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[F]], 1
11-
; CHECK-NEXT: [[VECINIT1:%.*]] = insertelement <4 x i32> [[VECINIT]], i32 [[ADD]], i32 1
12-
; CHECK-NEXT: [[TMP1:%.*]] = insertelement <2 x i32> poison, i32 [[F]], i32 0
13-
; CHECK-NEXT: [[TMP2:%.*]] = shufflevector <2 x i32> [[TMP1]], <2 x i32> poison, <2 x i32> zeroinitializer
14-
; CHECK-NEXT: [[TMP3:%.*]] = add nsw <2 x i32> [[TMP2]], <i32 2, i32 3>
15-
; CHECK-NEXT: [[TMP4:%.*]] = shufflevector <2 x i32> [[TMP3]], <2 x i32> poison, <4 x i32> <i32 0, i32 1, i32 poison, i32 poison>
16-
; CHECK-NEXT: [[VECINIT51:%.*]] = shufflevector <4 x i32> [[VECINIT1]], <4 x i32> [[TMP4]], <4 x i32> <i32 0, i32 1, i32 4, i32 5>
9+
; CHECK-NEXT: [[VECINIT:%.*]] = insertelement <4 x i32> poison, i32 [[F:%.*]], i32 0
10+
; CHECK-NEXT: [[TMP2:%.*]] = shufflevector <4 x i32> [[VECINIT]], <4 x i32> poison, <4 x i32> zeroinitializer
11+
; CHECK-NEXT: [[VECINIT51:%.*]] = add <4 x i32> [[TMP2]], <i32 0, i32 1, i32 2, i32 3>
1712
; CHECK-NEXT: ret <4 x i32> [[VECINIT51]]
1813
;
1914
%vecinit = insertelement <4 x i32> undef, i32 %f, i32 0

‎llvm/test/Transforms/SLPVectorizer/X86/shuffle-mask-emission.ll

Lines changed: 3 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -5,9 +5,11 @@ define i1 @test() {
55
; CHECK-LABEL: define i1 @test() {
66
; CHECK-NEXT: [[ENTRY:.*:]]
77
; CHECK-NEXT: [[H_PROMOTED118_I_FR:%.*]] = freeze i32 1
8+
; CHECK-NEXT: [[TMP3:%.*]] = insertelement <2 x i32> <i32 poison, i32 0>, i32 [[H_PROMOTED118_I_FR]], i32 0
9+
; CHECK-NEXT: [[TMP4:%.*]] = add <2 x i32> zeroinitializer, [[TMP3]]
10+
; CHECK-NEXT: [[TMP2:%.*]] = shufflevector <2 x i32> [[TMP4]], <2 x i32> poison, <4 x i32> <i32 0, i32 0, i32 1, i32 0>
811
; CHECK-NEXT: [[TMP0:%.*]] = insertelement <4 x i32> <i32 0, i32 0, i32 poison, i32 0>, i32 [[H_PROMOTED118_I_FR]], i32 2
912
; CHECK-NEXT: [[TMP1:%.*]] = add <4 x i32> zeroinitializer, [[TMP0]]
10-
; CHECK-NEXT: [[TMP2:%.*]] = shufflevector <4 x i32> [[TMP0]], <4 x i32> [[TMP1]], <4 x i32> <i32 2, i32 2, i32 7, i32 2>
1113
; CHECK-NEXT: [[TMP5:%.*]] = add <4 x i32> [[TMP1]], [[TMP2]]
1214
; CHECK-NEXT: [[TMP6:%.*]] = and <4 x i32> [[TMP5]], <i32 0, i32 1, i32 1, i32 1>
1315
; CHECK-NEXT: [[TMP7:%.*]] = icmp eq <4 x i32> [[TMP6]], <i32 1, i32 0, i32 0, i32 0>

‎llvm/test/Transforms/SLPVectorizer/alternate-non-profitable.ll

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -150,9 +150,9 @@ define <2 x i32> @replace_through_int_casts_ele0_only(i16 %inp, <2 x i16> %dead)
150150
define <2 x i8> @replace_through_binop_fail_cant_speculate(i8 %inp, <2 x i8> %d, <2 x i8> %any) {
151151
; CHECK-LABEL: define <2 x i8> @replace_through_binop_fail_cant_speculate(
152152
; CHECK-SAME: i8 [[INP:%.*]], <2 x i8> [[D:%.*]], <2 x i8> [[ANY:%.*]]) {
153-
; CHECK-NEXT: [[ADD:%.*]] = add i8 [[INP]], 5
154-
; CHECK-NEXT: [[V0:%.*]] = insertelement <2 x i8> poison, i8 [[INP]], i64 0
155-
; CHECK-NEXT: [[V:%.*]] = insertelement <2 x i8> [[V0]], i8 [[ADD]], i64 1
153+
; CHECK-NEXT: [[TMP3:%.*]] = insertelement <2 x i8> poison, i8 [[INP]], i32 0
154+
; CHECK-NEXT: [[TMP2:%.*]] = shufflevector <2 x i8> [[TMP3]], <2 x i8> poison, <2 x i32> zeroinitializer
155+
; CHECK-NEXT: [[V:%.*]] = add <2 x i8> [[TMP2]], <i8 0, i8 5>
156156
; CHECK-NEXT: [[DIV0:%.*]] = sdiv <2 x i8> splat (i8 -128), [[V]]
157157
; CHECK-NEXT: [[TMP1:%.*]] = xor i8 [[INP]], 123
158158
; CHECK-NEXT: [[R:%.*]] = insertelement <2 x i8> [[DIV0]], i8 [[TMP1]], i64 0

0 commit comments

Comments
 (0)