Skip to content

Commit 214f99b

Browse files
alexey-bataevalexey-bataev
committed
Revert "[SLP]Do not emit extra shuffle for insertelements vectorization."
This reverts commit 871ea69 to fix the problem if the first vector is not just undef.
1 parent 0ebb756 commit 214f99b

11 files changed

+49
-32
lines changed

‎llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -5251,7 +5251,7 @@ Value *BoUpSLP::vectorizeTree(TreeEntry *E) {
52515251
if (!IsIdentity || NumElts != NumScalars)
52525252
V = Builder.CreateShuffleVector(V, Mask);
52535253

5254-
if ((!IsIdentity || Offset != 0) && NumElts != NumScalars) {
5254+
if (NumElts != NumScalars) {
52555255
SmallVector<int> InsertMask(NumElts);
52565256
std::iota(InsertMask.begin(), InsertMask.end(), 0);
52575257
for (unsigned I = 0; I < NumElts; I++) {

‎llvm/test/Transforms/SLPVectorizer/AArch64/vectorize-free-extracts-inserts.ll

Lines changed: 25 additions & 16 deletions
Original file line numberDiff line numberDiff line change
@@ -105,9 +105,10 @@ define void @noop_extract_second_2_lanes(<4 x double>* %ptr.1, <4 x double>* %pt
105105
; CHECK-NEXT: [[TMP3:%.*]] = insertelement <2 x double> [[TMP2]], double [[V2_LANE_2]], i32 1
106106
; CHECK-NEXT: [[TMP4:%.*]] = fmul <2 x double> [[TMP1]], [[TMP3]]
107107
; CHECK-NEXT: [[TMP5:%.*]] = shufflevector <2 x double> [[TMP4]], <2 x double> poison, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
108+
; CHECK-NEXT: [[A_INS_11:%.*]] = shufflevector <4 x double> undef, <4 x double> [[TMP5]], <4 x i32> <i32 4, i32 5, i32 2, i32 3>
108109
; CHECK-NEXT: call void @use(double [[V1_LANE_2]])
109110
; CHECK-NEXT: call void @use(double [[V1_LANE_3]])
110-
; CHECK-NEXT: store <4 x double> [[TMP5]], <4 x double>* [[PTR_1]], align 8
111+
; CHECK-NEXT: store <4 x double> [[A_INS_11]], <4 x double>* [[PTR_1]], align 8
111112
; CHECK-NEXT: ret void
112113
;
113114
bb:
@@ -185,9 +186,10 @@ define void @extract_lanes_1_and_2(<4 x double>* %ptr.1, <4 x double>* %ptr.2) {
185186
; CHECK-NEXT: [[TMP3:%.*]] = insertelement <2 x double> [[TMP2]], double [[V2_LANE_2]], i32 1
186187
; CHECK-NEXT: [[TMP4:%.*]] = fmul <2 x double> [[TMP1]], [[TMP3]]
187188
; CHECK-NEXT: [[TMP5:%.*]] = shufflevector <2 x double> [[TMP4]], <2 x double> poison, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
189+
; CHECK-NEXT: [[A_INS_11:%.*]] = shufflevector <4 x double> undef, <4 x double> [[TMP5]], <4 x i32> <i32 4, i32 5, i32 2, i32 3>
188190
; CHECK-NEXT: call void @use(double [[V1_LANE_1]])
189191
; CHECK-NEXT: call void @use(double [[V1_LANE_2]])
190-
; CHECK-NEXT: store <4 x double> [[TMP5]], <4 x double>* [[PTR_1]], align 8
192+
; CHECK-NEXT: store <4 x double> [[A_INS_11]], <4 x double>* [[PTR_1]], align 8
191193
; CHECK-NEXT: ret void
192194
;
193195
bb:
@@ -235,11 +237,12 @@ define void @noop_extracts_existing_vector_4_lanes(<9 x double>* %ptr.1, <4 x do
235237
; CHECK-NEXT: [[SHUFFLE:%.*]] = shufflevector <4 x double> [[TMP5]], <4 x double> poison, <4 x i32> <i32 0, i32 0, i32 0, i32 1>
236238
; CHECK-NEXT: [[TMP6:%.*]] = fmul <4 x double> [[TMP3]], [[SHUFFLE]]
237239
; CHECK-NEXT: [[TMP7:%.*]] = shufflevector <4 x double> [[TMP6]], <4 x double> poison, <9 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
240+
; CHECK-NEXT: [[A_INS_31:%.*]] = shufflevector <9 x double> undef, <9 x double> [[TMP7]], <9 x i32> <i32 9, i32 10, i32 11, i32 12, i32 4, i32 5, i32 6, i32 7, i32 8>
238241
; CHECK-NEXT: call void @use(double [[V1_LANE_0]])
239242
; CHECK-NEXT: call void @use(double [[V1_LANE_1]])
240243
; CHECK-NEXT: call void @use(double [[V1_LANE_2]])
241244
; CHECK-NEXT: call void @use(double [[V1_LANE_3]])
242-
; CHECK-NEXT: store <9 x double> [[TMP7]], <9 x double>* [[PTR_1]], align 8
245+
; CHECK-NEXT: store <9 x double> [[A_INS_31]], <9 x double>* [[PTR_1]], align 8
243246
; CHECK-NEXT: ret void
244247
;
245248
bb:
@@ -355,11 +358,12 @@ define void @noop_extracts_9_lanes(<9 x double>* %ptr.1, <4 x double>* %ptr.2) {
355358
; CHECK-NEXT: [[TMP8:%.*]] = insertelement <8 x double> poison, double [[V2_LANE_0]], i32 0
356359
; CHECK-NEXT: [[TMP9:%.*]] = insertelement <8 x double> [[TMP8]], double [[V2_LANE_2]], i32 1
357360
; CHECK-NEXT: [[TMP10:%.*]] = insertelement <8 x double> [[TMP9]], double [[V2_LANE_1]], i32 2
358-
; CHECK-NEXT: [[SHUFFLE1:%.*]] = shufflevector <8 x double> [[TMP10]], <8 x double> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 0, i32 1, i32 0, i32 1, i32 2>
359-
; CHECK-NEXT: [[TMP11:%.*]] = fmul <8 x double> [[TMP7]], [[SHUFFLE1]]
361+
; CHECK-NEXT: [[SHUFFLE2:%.*]] = shufflevector <8 x double> [[TMP10]], <8 x double> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 0, i32 1, i32 0, i32 1, i32 2>
362+
; CHECK-NEXT: [[TMP11:%.*]] = fmul <8 x double> [[TMP7]], [[SHUFFLE2]]
360363
; CHECK-NEXT: [[A_LANE_8:%.*]] = fmul double [[V1_LANE_2]], [[V2_LANE_0]]
361364
; CHECK-NEXT: [[TMP12:%.*]] = shufflevector <8 x double> [[TMP11]], <8 x double> poison, <9 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef>
362-
; CHECK-NEXT: [[A_INS_8:%.*]] = insertelement <9 x double> [[TMP12]], double [[A_LANE_8]], i32 8
365+
; CHECK-NEXT: [[A_INS_73:%.*]] = shufflevector <9 x double> undef, <9 x double> [[TMP12]], <9 x i32> <i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 8>
366+
; CHECK-NEXT: [[A_INS_8:%.*]] = insertelement <9 x double> [[A_INS_73]], double [[A_LANE_8]], i32 8
363367
; CHECK-NEXT: [[TMP13:%.*]] = insertelement <8 x double> poison, double [[V1_LANE_6]], i32 0
364368
; CHECK-NEXT: [[TMP14:%.*]] = insertelement <8 x double> [[TMP13]], double [[V1_LANE_7]], i32 1
365369
; CHECK-NEXT: [[TMP15:%.*]] = insertelement <8 x double> [[TMP14]], double [[V1_LANE_8]], i32 2
@@ -375,7 +379,8 @@ define void @noop_extracts_9_lanes(<9 x double>* %ptr.1, <4 x double>* %ptr.2) {
375379
; CHECK-NEXT: [[TMP24:%.*]] = fmul <8 x double> [[TMP20]], [[SHUFFLE]]
376380
; CHECK-NEXT: [[B_LANE_8:%.*]] = fmul double [[V1_LANE_5]], [[V2_LANE_0]]
377381
; CHECK-NEXT: [[TMP25:%.*]] = shufflevector <8 x double> [[TMP24]], <8 x double> poison, <9 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef>
378-
; CHECK-NEXT: [[B_INS_8:%.*]] = insertelement <9 x double> [[TMP25]], double [[B_LANE_8]], i32 8
382+
; CHECK-NEXT: [[B_INS_71:%.*]] = shufflevector <9 x double> undef, <9 x double> [[TMP25]], <9 x i32> <i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 8>
383+
; CHECK-NEXT: [[B_INS_8:%.*]] = insertelement <9 x double> [[B_INS_71]], double [[B_LANE_8]], i32 8
379384
; CHECK-NEXT: [[RES:%.*]] = fsub <9 x double> [[A_INS_8]], [[B_INS_8]]
380385
; CHECK-NEXT: store <9 x double> [[RES]], <9 x double>* [[PTR_1]], align 8
381386
; CHECK-NEXT: ret void
@@ -472,11 +477,12 @@ define void @first_mul_chain_jumbled(<9 x double>* %ptr.1, <4 x double>* %ptr.2)
472477
; CHECK-NEXT: [[TMP8:%.*]] = insertelement <8 x double> poison, double [[V2_LANE_1]], i32 0
473478
; CHECK-NEXT: [[TMP9:%.*]] = insertelement <8 x double> [[TMP8]], double [[V2_LANE_0]], i32 1
474479
; CHECK-NEXT: [[TMP10:%.*]] = insertelement <8 x double> [[TMP9]], double [[V2_LANE_2]], i32 2
475-
; CHECK-NEXT: [[SHUFFLE1:%.*]] = shufflevector <8 x double> [[TMP10]], <8 x double> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 1, i32 2, i32 0, i32 1, i32 2>
476-
; CHECK-NEXT: [[TMP11:%.*]] = fmul <8 x double> [[TMP7]], [[SHUFFLE1]]
480+
; CHECK-NEXT: [[SHUFFLE2:%.*]] = shufflevector <8 x double> [[TMP10]], <8 x double> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 1, i32 2, i32 0, i32 1, i32 2>
481+
; CHECK-NEXT: [[TMP11:%.*]] = fmul <8 x double> [[TMP7]], [[SHUFFLE2]]
477482
; CHECK-NEXT: [[A_LANE_8:%.*]] = fmul double [[V1_LANE_2]], [[V2_LANE_1]]
478483
; CHECK-NEXT: [[TMP12:%.*]] = shufflevector <8 x double> [[TMP11]], <8 x double> poison, <9 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef>
479-
; CHECK-NEXT: [[A_INS_8:%.*]] = insertelement <9 x double> [[TMP12]], double [[A_LANE_8]], i32 8
484+
; CHECK-NEXT: [[A_INS_73:%.*]] = shufflevector <9 x double> undef, <9 x double> [[TMP12]], <9 x i32> <i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 8>
485+
; CHECK-NEXT: [[A_INS_8:%.*]] = insertelement <9 x double> [[A_INS_73]], double [[A_LANE_8]], i32 8
480486
; CHECK-NEXT: [[TMP13:%.*]] = insertelement <8 x double> poison, double [[V1_LANE_6]], i32 0
481487
; CHECK-NEXT: [[TMP14:%.*]] = insertelement <8 x double> [[TMP13]], double [[V1_LANE_7]], i32 1
482488
; CHECK-NEXT: [[TMP15:%.*]] = insertelement <8 x double> [[TMP14]], double [[V1_LANE_8]], i32 2
@@ -485,10 +491,11 @@ define void @first_mul_chain_jumbled(<9 x double>* %ptr.1, <4 x double>* %ptr.2)
485491
; CHECK-NEXT: [[TMP18:%.*]] = insertelement <8 x double> [[TMP17]], double [[V1_LANE_2]], i32 5
486492
; CHECK-NEXT: [[TMP19:%.*]] = insertelement <8 x double> [[TMP18]], double [[V1_LANE_3]], i32 6
487493
; CHECK-NEXT: [[TMP20:%.*]] = insertelement <8 x double> [[TMP19]], double [[V1_LANE_4]], i32 7
488-
; CHECK-NEXT: [[TMP21:%.*]] = fmul <8 x double> [[TMP20]], [[SHUFFLE1]]
494+
; CHECK-NEXT: [[TMP21:%.*]] = fmul <8 x double> [[TMP20]], [[SHUFFLE2]]
489495
; CHECK-NEXT: [[B_LANE_8:%.*]] = fmul double [[V1_LANE_5]], [[V2_LANE_0]]
490496
; CHECK-NEXT: [[TMP22:%.*]] = shufflevector <8 x double> [[TMP21]], <8 x double> poison, <9 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef>
491-
; CHECK-NEXT: [[B_INS_8:%.*]] = insertelement <9 x double> [[TMP22]], double [[B_LANE_8]], i32 8
497+
; CHECK-NEXT: [[B_INS_71:%.*]] = shufflevector <9 x double> undef, <9 x double> [[TMP22]], <9 x i32> <i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 8>
498+
; CHECK-NEXT: [[B_INS_8:%.*]] = insertelement <9 x double> [[B_INS_71]], double [[B_LANE_8]], i32 8
492499
; CHECK-NEXT: [[RES:%.*]] = fsub <9 x double> [[A_INS_8]], [[B_INS_8]]
493500
; CHECK-NEXT: store <9 x double> [[RES]], <9 x double>* [[PTR_1]], align 8
494501
; CHECK-NEXT: ret void
@@ -585,11 +592,12 @@ define void @first_and_second_mul_chain_jumbled(<9 x double>* %ptr.1, <4 x doubl
585592
; CHECK-NEXT: [[TMP8:%.*]] = insertelement <8 x double> poison, double [[V2_LANE_0]], i32 0
586593
; CHECK-NEXT: [[TMP9:%.*]] = insertelement <8 x double> [[TMP8]], double [[V2_LANE_2]], i32 1
587594
; CHECK-NEXT: [[TMP10:%.*]] = insertelement <8 x double> [[TMP9]], double [[V2_LANE_1]], i32 2
588-
; CHECK-NEXT: [[SHUFFLE1:%.*]] = shufflevector <8 x double> [[TMP10]], <8 x double> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 1, i32 2, i32 0, i32 1, i32 2>
589-
; CHECK-NEXT: [[TMP11:%.*]] = fmul <8 x double> [[TMP7]], [[SHUFFLE1]]
595+
; CHECK-NEXT: [[SHUFFLE2:%.*]] = shufflevector <8 x double> [[TMP10]], <8 x double> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 1, i32 2, i32 0, i32 1, i32 2>
596+
; CHECK-NEXT: [[TMP11:%.*]] = fmul <8 x double> [[TMP7]], [[SHUFFLE2]]
590597
; CHECK-NEXT: [[A_LANE_8:%.*]] = fmul double [[V1_LANE_2]], [[V2_LANE_0]]
591598
; CHECK-NEXT: [[TMP12:%.*]] = shufflevector <8 x double> [[TMP11]], <8 x double> poison, <9 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef>
592-
; CHECK-NEXT: [[A_INS_8:%.*]] = insertelement <9 x double> [[TMP12]], double [[A_LANE_8]], i32 8
599+
; CHECK-NEXT: [[A_INS_73:%.*]] = shufflevector <9 x double> undef, <9 x double> [[TMP12]], <9 x i32> <i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 8>
600+
; CHECK-NEXT: [[A_INS_8:%.*]] = insertelement <9 x double> [[A_INS_73]], double [[A_LANE_8]], i32 8
593601
; CHECK-NEXT: [[TMP13:%.*]] = insertelement <8 x double> poison, double [[V1_LANE_7]], i32 0
594602
; CHECK-NEXT: [[TMP14:%.*]] = insertelement <8 x double> [[TMP13]], double [[V1_LANE_6]], i32 1
595603
; CHECK-NEXT: [[TMP15:%.*]] = insertelement <8 x double> [[TMP14]], double [[V1_LANE_8]], i32 2
@@ -605,7 +613,8 @@ define void @first_and_second_mul_chain_jumbled(<9 x double>* %ptr.1, <4 x doubl
605613
; CHECK-NEXT: [[TMP24:%.*]] = fmul <8 x double> [[TMP20]], [[SHUFFLE]]
606614
; CHECK-NEXT: [[B_LANE_8:%.*]] = fmul double [[V1_LANE_4]], [[V2_LANE_2]]
607615
; CHECK-NEXT: [[TMP25:%.*]] = shufflevector <8 x double> [[TMP24]], <8 x double> poison, <9 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 undef>
608-
; CHECK-NEXT: [[B_INS_8:%.*]] = insertelement <9 x double> [[TMP25]], double [[B_LANE_8]], i32 8
616+
; CHECK-NEXT: [[B_INS_71:%.*]] = shufflevector <9 x double> undef, <9 x double> [[TMP25]], <9 x i32> <i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 8>
617+
; CHECK-NEXT: [[B_INS_8:%.*]] = insertelement <9 x double> [[B_INS_71]], double [[B_LANE_8]], i32 8
609618
; CHECK-NEXT: [[RES:%.*]] = fsub <9 x double> [[A_INS_8]], [[B_INS_8]]
610619
; CHECK-NEXT: store <9 x double> [[RES]], <9 x double>* [[PTR_1]], align 8
611620
; CHECK-NEXT: ret void

‎llvm/test/Transforms/SLPVectorizer/X86/arith-fp-inseltpoison.ll

Lines changed: 5 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -644,13 +644,14 @@ define <8 x double> @buildvector_div_8f64(<8 x double> %a, <8 x double> %b) {
644644
; SLM-NEXT: [[TMP19:%.*]] = insertelement <2 x double> [[TMP18]], double [[B7]], i32 1
645645
; SLM-NEXT: [[TMP20:%.*]] = fdiv <2 x double> [[TMP17]], [[TMP19]]
646646
; SLM-NEXT: [[TMP21:%.*]] = shufflevector <2 x double> [[TMP5]], <2 x double> poison, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
647+
; SLM-NEXT: [[R11:%.*]] = shufflevector <8 x double> poison, <8 x double> [[TMP21]], <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
647648
; SLM-NEXT: [[TMP22:%.*]] = shufflevector <2 x double> [[TMP10]], <2 x double> poison, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
648-
; SLM-NEXT: [[R31:%.*]] = shufflevector <8 x double> [[TMP21]], <8 x double> [[TMP22]], <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
649+
; SLM-NEXT: [[R32:%.*]] = shufflevector <8 x double> [[R11]], <8 x double> [[TMP22]], <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
649650
; SLM-NEXT: [[TMP23:%.*]] = shufflevector <2 x double> [[TMP15]], <2 x double> poison, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
650-
; SLM-NEXT: [[R52:%.*]] = shufflevector <8 x double> [[R31]], <8 x double> [[TMP23]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
651+
; SLM-NEXT: [[R53:%.*]] = shufflevector <8 x double> [[R32]], <8 x double> [[TMP23]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
651652
; SLM-NEXT: [[TMP24:%.*]] = shufflevector <2 x double> [[TMP20]], <2 x double> poison, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
652-
; SLM-NEXT: [[R73:%.*]] = shufflevector <8 x double> [[R52]], <8 x double> [[TMP24]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
653-
; SLM-NEXT: ret <8 x double> [[R73]]
653+
; SLM-NEXT: [[R74:%.*]] = shufflevector <8 x double> [[R53]], <8 x double> [[TMP24]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
654+
; SLM-NEXT: ret <8 x double> [[R74]]
654655
;
655656
; AVX-LABEL: @buildvector_div_8f64(
656657
; AVX-NEXT: [[TMP1:%.*]] = fdiv <8 x double> [[A:%.*]], [[B:%.*]]

‎llvm/test/Transforms/SLPVectorizer/X86/arith-fp.ll

Lines changed: 5 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -644,13 +644,14 @@ define <8 x double> @buildvector_div_8f64(<8 x double> %a, <8 x double> %b) {
644644
; SLM-NEXT: [[TMP19:%.*]] = insertelement <2 x double> [[TMP18]], double [[B7]], i32 1
645645
; SLM-NEXT: [[TMP20:%.*]] = fdiv <2 x double> [[TMP17]], [[TMP19]]
646646
; SLM-NEXT: [[TMP21:%.*]] = shufflevector <2 x double> [[TMP5]], <2 x double> poison, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
647+
; SLM-NEXT: [[R11:%.*]] = shufflevector <8 x double> undef, <8 x double> [[TMP21]], <8 x i32> <i32 8, i32 9, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
647648
; SLM-NEXT: [[TMP22:%.*]] = shufflevector <2 x double> [[TMP10]], <2 x double> poison, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
648-
; SLM-NEXT: [[R31:%.*]] = shufflevector <8 x double> [[TMP21]], <8 x double> [[TMP22]], <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
649+
; SLM-NEXT: [[R32:%.*]] = shufflevector <8 x double> [[R11]], <8 x double> [[TMP22]], <8 x i32> <i32 0, i32 1, i32 8, i32 9, i32 4, i32 5, i32 6, i32 7>
649650
; SLM-NEXT: [[TMP23:%.*]] = shufflevector <2 x double> [[TMP15]], <2 x double> poison, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
650-
; SLM-NEXT: [[R52:%.*]] = shufflevector <8 x double> [[R31]], <8 x double> [[TMP23]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
651+
; SLM-NEXT: [[R53:%.*]] = shufflevector <8 x double> [[R32]], <8 x double> [[TMP23]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 8, i32 9, i32 6, i32 7>
651652
; SLM-NEXT: [[TMP24:%.*]] = shufflevector <2 x double> [[TMP20]], <2 x double> poison, <8 x i32> <i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
652-
; SLM-NEXT: [[R73:%.*]] = shufflevector <8 x double> [[R52]], <8 x double> [[TMP24]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
653-
; SLM-NEXT: ret <8 x double> [[R73]]
653+
; SLM-NEXT: [[R74:%.*]] = shufflevector <8 x double> [[R53]], <8 x double> [[TMP24]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 8, i32 9>
654+
; SLM-NEXT: ret <8 x double> [[R74]]
654655
;
655656
; AVX-LABEL: @buildvector_div_8f64(
656657
; AVX-NEXT: [[TMP1:%.*]] = fdiv <8 x double> [[A:%.*]], [[B:%.*]]

‎llvm/test/Transforms/SLPVectorizer/X86/insert-element-build-vector-inseltpoison.ll

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -300,6 +300,7 @@ define <4 x float> @simple_select_no_users(<4 x float> %a, <4 x float> %b, <4 x
300300
; CHECK-NEXT: [[TMP15:%.*]] = insertelement <2 x float> [[TMP14]], float [[B3]], i32 1
301301
; CHECK-NEXT: [[TMP16:%.*]] = select <2 x i1> [[TMP6]], <2 x float> [[TMP13]], <2 x float> [[TMP15]]
302302
; CHECK-NEXT: [[TMP17:%.*]] = shufflevector <2 x float> [[TMP11]], <2 x float> poison, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
303+
; CHECK-NEXT: [[RB2:%.*]] = shufflevector <4 x float> poison, <4 x float> [[TMP17]], <4 x i32> <i32 4, i32 5, i32 2, i32 3>
303304
; CHECK-NEXT: [[TMP18:%.*]] = shufflevector <2 x float> [[TMP16]], <2 x float> poison, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
304305
; CHECK-NEXT: [[RD1:%.*]] = shufflevector <4 x float> poison, <4 x float> [[TMP18]], <4 x i32> <i32 0, i32 1, i32 4, i32 5>
305306
; CHECK-NEXT: ret <4 x float> [[RD1]]

‎llvm/test/Transforms/SLPVectorizer/X86/insert-element-build-vector.ll

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -335,6 +335,7 @@ define <4 x float> @simple_select_no_users(<4 x float> %a, <4 x float> %b, <4 x
335335
; CHECK-NEXT: [[TMP15:%.*]] = insertelement <2 x float> [[TMP14]], float [[B3]], i32 1
336336
; CHECK-NEXT: [[TMP16:%.*]] = select <2 x i1> [[TMP6]], <2 x float> [[TMP13]], <2 x float> [[TMP15]]
337337
; CHECK-NEXT: [[TMP17:%.*]] = shufflevector <2 x float> [[TMP11]], <2 x float> poison, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
338+
; CHECK-NEXT: [[RB2:%.*]] = shufflevector <4 x float> undef, <4 x float> [[TMP17]], <4 x i32> <i32 4, i32 5, i32 2, i32 3>
338339
; CHECK-NEXT: [[TMP18:%.*]] = shufflevector <2 x float> [[TMP16]], <2 x float> poison, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
339340
; CHECK-NEXT: [[RD1:%.*]] = shufflevector <4 x float> undef, <4 x float> [[TMP18]], <4 x i32> <i32 0, i32 1, i32 4, i32 5>
340341
; CHECK-NEXT: ret <4 x float> [[RD1]]

‎llvm/test/Transforms/SLPVectorizer/X86/load-merge-inseltpoison.ll

Lines changed: 2 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -58,7 +58,8 @@ define <4 x float> @PR16739_byref(<4 x float>* nocapture readonly dereferenceabl
5858
; CHECK-NEXT: [[TMP2:%.*]] = load <2 x float>, <2 x float>* [[TMP1]], align 4
5959
; CHECK-NEXT: [[X2:%.*]] = load float, float* [[GEP2]], align 4
6060
; CHECK-NEXT: [[TMP3:%.*]] = shufflevector <2 x float> [[TMP2]], <2 x float> poison, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
61-
; CHECK-NEXT: [[I2:%.*]] = insertelement <4 x float> [[TMP3]], float [[X2]], i32 2
61+
; CHECK-NEXT: [[I11:%.*]] = shufflevector <4 x float> poison, <4 x float> [[TMP3]], <4 x i32> <i32 4, i32 5, i32 2, i32 3>
62+
; CHECK-NEXT: [[I2:%.*]] = insertelement <4 x float> [[I11]], float [[X2]], i32 2
6263
; CHECK-NEXT: [[I3:%.*]] = insertelement <4 x float> [[I2]], float [[X2]], i32 3
6364
; CHECK-NEXT: ret <4 x float> [[I3]]
6465
;

‎llvm/test/Transforms/SLPVectorizer/X86/load-merge.ll

Lines changed: 2 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -58,7 +58,8 @@ define <4 x float> @PR16739_byref(<4 x float>* nocapture readonly dereferenceabl
5858
; CHECK-NEXT: [[TMP2:%.*]] = load <2 x float>, <2 x float>* [[TMP1]], align 4
5959
; CHECK-NEXT: [[X2:%.*]] = load float, float* [[GEP2]], align 4
6060
; CHECK-NEXT: [[TMP3:%.*]] = shufflevector <2 x float> [[TMP2]], <2 x float> poison, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
61-
; CHECK-NEXT: [[I2:%.*]] = insertelement <4 x float> [[TMP3]], float [[X2]], i32 2
61+
; CHECK-NEXT: [[I11:%.*]] = shufflevector <4 x float> undef, <4 x float> [[TMP3]], <4 x i32> <i32 4, i32 5, i32 2, i32 3>
62+
; CHECK-NEXT: [[I2:%.*]] = insertelement <4 x float> [[I11]], float [[X2]], i32 2
6263
; CHECK-NEXT: [[I3:%.*]] = insertelement <4 x float> [[I2]], float [[X2]], i32 3
6364
; CHECK-NEXT: ret <4 x float> [[I3]]
6465
;

0 commit comments

Comments
 (0)