Skip to content

Commit 3ab1864

Browse files
lukel97lukel97
committed
Update SLP tests
1 parent 4c47e2b commit 3ab1864

File tree

4 files changed

+56
-76
lines changed

4 files changed

+56
-76
lines changed

‎llvm/test/Transforms/SLPVectorizer/AMDGPU/add_sub_sat-inseltpoison.ll

Lines changed: 10 additions & 20 deletions
Original file line numberDiff line numberDiff line change
@@ -264,23 +264,17 @@ define <3 x i16> @uadd_sat_v3i16(<3 x i16> %arg0, <3 x i16> %arg1) {
264264
; GFX8-NEXT: bb:
265265
; GFX8-NEXT: [[ARG0_2:%.*]] = extractelement <3 x i16> [[ARG0:%.*]], i64 2
266266
; GFX8-NEXT: [[ARG1_2:%.*]] = extractelement <3 x i16> [[ARG1:%.*]], i64 2
267-
; GFX8-NEXT: [[TMP0:%.*]] = shufflevector <3 x i16> [[ARG0]], <3 x i16> poison, <2 x i32> <i32 0, i32 1>
268-
; GFX8-NEXT: [[TMP1:%.*]] = shufflevector <3 x i16> [[ARG1]], <3 x i16> poison, <2 x i32> <i32 0, i32 1>
269-
; GFX8-NEXT: [[TMP2:%.*]] = call <2 x i16> @llvm.uadd.sat.v2i16(<2 x i16> [[TMP0]], <2 x i16> [[TMP1]])
267+
; GFX8-NEXT: [[TMP3:%.*]] = call <3 x i16> @llvm.uadd.sat.v3i16(<3 x i16> [[ARG0]], <3 x i16> [[ARG1]])
270268
; GFX8-NEXT: [[ADD_2:%.*]] = call i16 @llvm.uadd.sat.i16(i16 [[ARG0_2]], i16 [[ARG1_2]])
271-
; GFX8-NEXT: [[TMP3:%.*]] = shufflevector <2 x i16> [[TMP2]], <2 x i16> poison, <3 x i32> <i32 0, i32 1, i32 poison>
272269
; GFX8-NEXT: [[INS_2:%.*]] = insertelement <3 x i16> [[TMP3]], i16 [[ADD_2]], i64 2
273270
; GFX8-NEXT: ret <3 x i16> [[INS_2]]
274271
;
275272
; GFX9-LABEL: @uadd_sat_v3i16(
276273
; GFX9-NEXT: bb:
277274
; GFX9-NEXT: [[ARG0_2:%.*]] = extractelement <3 x i16> [[ARG0:%.*]], i64 2
278275
; GFX9-NEXT: [[ARG1_2:%.*]] = extractelement <3 x i16> [[ARG1:%.*]], i64 2
279-
; GFX9-NEXT: [[TMP0:%.*]] = shufflevector <3 x i16> [[ARG0]], <3 x i16> poison, <2 x i32> <i32 0, i32 1>
280-
; GFX9-NEXT: [[TMP1:%.*]] = shufflevector <3 x i16> [[ARG1]], <3 x i16> poison, <2 x i32> <i32 0, i32 1>
281-
; GFX9-NEXT: [[TMP2:%.*]] = call <2 x i16> @llvm.uadd.sat.v2i16(<2 x i16> [[TMP0]], <2 x i16> [[TMP1]])
276+
; GFX9-NEXT: [[TMP3:%.*]] = call <3 x i16> @llvm.uadd.sat.v3i16(<3 x i16> [[ARG0]], <3 x i16> [[ARG1]])
282277
; GFX9-NEXT: [[ADD_2:%.*]] = call i16 @llvm.uadd.sat.i16(i16 [[ARG0_2]], i16 [[ARG1_2]])
283-
; GFX9-NEXT: [[TMP3:%.*]] = shufflevector <2 x i16> [[TMP2]], <2 x i16> poison, <3 x i32> <i32 0, i32 1, i32 poison>
284278
; GFX9-NEXT: [[INS_2:%.*]] = insertelement <3 x i16> [[TMP3]], i16 [[ADD_2]], i64 2
285279
; GFX9-NEXT: ret <3 x i16> [[INS_2]]
286280
;
@@ -323,24 +317,20 @@ define <4 x i16> @uadd_sat_v4i16(<4 x i16> %arg0, <4 x i16> %arg1) {
323317
;
324318
; GFX8-LABEL: @uadd_sat_v4i16(
325319
; GFX8-NEXT: bb:
326-
; GFX8-NEXT: [[TMP0:%.*]] = shufflevector <4 x i16> [[ARG0:%.*]], <4 x i16> poison, <2 x i32> <i32 0, i32 1>
327-
; GFX8-NEXT: [[TMP1:%.*]] = shufflevector <4 x i16> [[ARG1:%.*]], <4 x i16> poison, <2 x i32> <i32 0, i32 1>
328-
; GFX8-NEXT: [[TMP2:%.*]] = call <2 x i16> @llvm.uadd.sat.v2i16(<2 x i16> [[TMP0]], <2 x i16> [[TMP1]])
329-
; GFX8-NEXT: [[TMP3:%.*]] = shufflevector <4 x i16> [[ARG0]], <4 x i16> poison, <2 x i32> <i32 2, i32 3>
320+
; GFX8-NEXT: [[TMP0:%.*]] = call <4 x i16> @llvm.uadd.sat.v4i16(<4 x i16> [[ARG0:%.*]], <4 x i16> [[ARG2:%.*]])
321+
; GFX8-NEXT: [[ARG1:%.*]] = call <4 x i16> @llvm.uadd.sat.v4i16(<4 x i16> [[ARG0]], <4 x i16> [[ARG2]])
330322
; GFX8-NEXT: [[TMP4:%.*]] = shufflevector <4 x i16> [[ARG1]], <4 x i16> poison, <2 x i32> <i32 2, i32 3>
331-
; GFX8-NEXT: [[TMP5:%.*]] = call <2 x i16> @llvm.uadd.sat.v2i16(<2 x i16> [[TMP3]], <2 x i16> [[TMP4]])
332-
; GFX8-NEXT: [[INS_31:%.*]] = shufflevector <2 x i16> [[TMP2]], <2 x i16> [[TMP5]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
323+
; GFX8-NEXT: [[TMP3:%.*]] = shufflevector <2 x i16> [[TMP4]], <2 x i16> poison, <4 x i32> <i32 0, i32 1, i32 poison, i32 poison>
324+
; GFX8-NEXT: [[INS_31:%.*]] = shufflevector <4 x i16> [[TMP0]], <4 x i16> [[TMP3]], <4 x i32> <i32 0, i32 1, i32 4, i32 5>
333325
; GFX8-NEXT: ret <4 x i16> [[INS_31]]
334326
;
335327
; GFX9-LABEL: @uadd_sat_v4i16(
336328
; GFX9-NEXT: bb:
337-
; GFX9-NEXT: [[TMP0:%.*]] = shufflevector <4 x i16> [[ARG0:%.*]], <4 x i16> poison, <2 x i32> <i32 0, i32 1>
338-
; GFX9-NEXT: [[TMP1:%.*]] = shufflevector <4 x i16> [[ARG1:%.*]], <4 x i16> poison, <2 x i32> <i32 0, i32 1>
339-
; GFX9-NEXT: [[TMP2:%.*]] = call <2 x i16> @llvm.uadd.sat.v2i16(<2 x i16> [[TMP0]], <2 x i16> [[TMP1]])
340-
; GFX9-NEXT: [[TMP3:%.*]] = shufflevector <4 x i16> [[ARG0]], <4 x i16> poison, <2 x i32> <i32 2, i32 3>
329+
; GFX9-NEXT: [[TMP0:%.*]] = call <4 x i16> @llvm.uadd.sat.v4i16(<4 x i16> [[ARG0:%.*]], <4 x i16> [[ARG2:%.*]])
330+
; GFX9-NEXT: [[ARG1:%.*]] = call <4 x i16> @llvm.uadd.sat.v4i16(<4 x i16> [[ARG0]], <4 x i16> [[ARG2]])
341331
; GFX9-NEXT: [[TMP4:%.*]] = shufflevector <4 x i16> [[ARG1]], <4 x i16> poison, <2 x i32> <i32 2, i32 3>
342-
; GFX9-NEXT: [[TMP5:%.*]] = call <2 x i16> @llvm.uadd.sat.v2i16(<2 x i16> [[TMP3]], <2 x i16> [[TMP4]])
343-
; GFX9-NEXT: [[INS_31:%.*]] = shufflevector <2 x i16> [[TMP2]], <2 x i16> [[TMP5]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
332+
; GFX9-NEXT: [[TMP3:%.*]] = shufflevector <2 x i16> [[TMP4]], <2 x i16> poison, <4 x i32> <i32 0, i32 1, i32 poison, i32 poison>
333+
; GFX9-NEXT: [[INS_31:%.*]] = shufflevector <4 x i16> [[TMP0]], <4 x i16> [[TMP3]], <4 x i32> <i32 0, i32 1, i32 4, i32 5>
344334
; GFX9-NEXT: ret <4 x i16> [[INS_31]]
345335
;
346336
bb:

‎llvm/test/Transforms/SLPVectorizer/AMDGPU/add_sub_sat.ll

Lines changed: 10 additions & 20 deletions
Original file line numberDiff line numberDiff line change
@@ -264,23 +264,17 @@ define <3 x i16> @uadd_sat_v3i16(<3 x i16> %arg0, <3 x i16> %arg1) {
264264
; GFX8-NEXT: bb:
265265
; GFX8-NEXT: [[ARG0_2:%.*]] = extractelement <3 x i16> [[ARG0:%.*]], i64 2
266266
; GFX8-NEXT: [[ARG1_2:%.*]] = extractelement <3 x i16> [[ARG1:%.*]], i64 2
267-
; GFX8-NEXT: [[TMP0:%.*]] = shufflevector <3 x i16> [[ARG0]], <3 x i16> poison, <2 x i32> <i32 0, i32 1>
268-
; GFX8-NEXT: [[TMP1:%.*]] = shufflevector <3 x i16> [[ARG1]], <3 x i16> poison, <2 x i32> <i32 0, i32 1>
269-
; GFX8-NEXT: [[TMP2:%.*]] = call <2 x i16> @llvm.uadd.sat.v2i16(<2 x i16> [[TMP0]], <2 x i16> [[TMP1]])
267+
; GFX8-NEXT: [[TMP3:%.*]] = call <3 x i16> @llvm.uadd.sat.v3i16(<3 x i16> [[ARG0]], <3 x i16> [[ARG1]])
270268
; GFX8-NEXT: [[ADD_2:%.*]] = call i16 @llvm.uadd.sat.i16(i16 [[ARG0_2]], i16 [[ARG1_2]])
271-
; GFX8-NEXT: [[TMP3:%.*]] = shufflevector <2 x i16> [[TMP2]], <2 x i16> poison, <3 x i32> <i32 0, i32 1, i32 poison>
272269
; GFX8-NEXT: [[INS_2:%.*]] = insertelement <3 x i16> [[TMP3]], i16 [[ADD_2]], i64 2
273270
; GFX8-NEXT: ret <3 x i16> [[INS_2]]
274271
;
275272
; GFX9-LABEL: @uadd_sat_v3i16(
276273
; GFX9-NEXT: bb:
277274
; GFX9-NEXT: [[ARG0_2:%.*]] = extractelement <3 x i16> [[ARG0:%.*]], i64 2
278275
; GFX9-NEXT: [[ARG1_2:%.*]] = extractelement <3 x i16> [[ARG1:%.*]], i64 2
279-
; GFX9-NEXT: [[TMP0:%.*]] = shufflevector <3 x i16> [[ARG0]], <3 x i16> poison, <2 x i32> <i32 0, i32 1>
280-
; GFX9-NEXT: [[TMP1:%.*]] = shufflevector <3 x i16> [[ARG1]], <3 x i16> poison, <2 x i32> <i32 0, i32 1>
281-
; GFX9-NEXT: [[TMP2:%.*]] = call <2 x i16> @llvm.uadd.sat.v2i16(<2 x i16> [[TMP0]], <2 x i16> [[TMP1]])
276+
; GFX9-NEXT: [[TMP3:%.*]] = call <3 x i16> @llvm.uadd.sat.v3i16(<3 x i16> [[ARG0]], <3 x i16> [[ARG1]])
282277
; GFX9-NEXT: [[ADD_2:%.*]] = call i16 @llvm.uadd.sat.i16(i16 [[ARG0_2]], i16 [[ARG1_2]])
283-
; GFX9-NEXT: [[TMP3:%.*]] = shufflevector <2 x i16> [[TMP2]], <2 x i16> poison, <3 x i32> <i32 0, i32 1, i32 poison>
284278
; GFX9-NEXT: [[INS_2:%.*]] = insertelement <3 x i16> [[TMP3]], i16 [[ADD_2]], i64 2
285279
; GFX9-NEXT: ret <3 x i16> [[INS_2]]
286280
;
@@ -323,24 +317,20 @@ define <4 x i16> @uadd_sat_v4i16(<4 x i16> %arg0, <4 x i16> %arg1) {
323317
;
324318
; GFX8-LABEL: @uadd_sat_v4i16(
325319
; GFX8-NEXT: bb:
326-
; GFX8-NEXT: [[TMP0:%.*]] = shufflevector <4 x i16> [[ARG0:%.*]], <4 x i16> poison, <2 x i32> <i32 0, i32 1>
327-
; GFX8-NEXT: [[TMP1:%.*]] = shufflevector <4 x i16> [[ARG1:%.*]], <4 x i16> poison, <2 x i32> <i32 0, i32 1>
328-
; GFX8-NEXT: [[TMP2:%.*]] = call <2 x i16> @llvm.uadd.sat.v2i16(<2 x i16> [[TMP0]], <2 x i16> [[TMP1]])
329-
; GFX8-NEXT: [[TMP3:%.*]] = shufflevector <4 x i16> [[ARG0]], <4 x i16> poison, <2 x i32> <i32 2, i32 3>
320+
; GFX8-NEXT: [[TMP0:%.*]] = call <4 x i16> @llvm.uadd.sat.v4i16(<4 x i16> [[ARG0:%.*]], <4 x i16> [[ARG2:%.*]])
321+
; GFX8-NEXT: [[ARG1:%.*]] = call <4 x i16> @llvm.uadd.sat.v4i16(<4 x i16> [[ARG0]], <4 x i16> [[ARG2]])
330322
; GFX8-NEXT: [[TMP4:%.*]] = shufflevector <4 x i16> [[ARG1]], <4 x i16> poison, <2 x i32> <i32 2, i32 3>
331-
; GFX8-NEXT: [[TMP5:%.*]] = call <2 x i16> @llvm.uadd.sat.v2i16(<2 x i16> [[TMP3]], <2 x i16> [[TMP4]])
332-
; GFX8-NEXT: [[INS_31:%.*]] = shufflevector <2 x i16> [[TMP2]], <2 x i16> [[TMP5]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
323+
; GFX8-NEXT: [[TMP3:%.*]] = shufflevector <2 x i16> [[TMP4]], <2 x i16> poison, <4 x i32> <i32 0, i32 1, i32 poison, i32 poison>
324+
; GFX8-NEXT: [[INS_31:%.*]] = shufflevector <4 x i16> [[TMP0]], <4 x i16> [[TMP3]], <4 x i32> <i32 0, i32 1, i32 4, i32 5>
333325
; GFX8-NEXT: ret <4 x i16> [[INS_31]]
334326
;
335327
; GFX9-LABEL: @uadd_sat_v4i16(
336328
; GFX9-NEXT: bb:
337-
; GFX9-NEXT: [[TMP0:%.*]] = shufflevector <4 x i16> [[ARG0:%.*]], <4 x i16> poison, <2 x i32> <i32 0, i32 1>
338-
; GFX9-NEXT: [[TMP1:%.*]] = shufflevector <4 x i16> [[ARG1:%.*]], <4 x i16> poison, <2 x i32> <i32 0, i32 1>
339-
; GFX9-NEXT: [[TMP2:%.*]] = call <2 x i16> @llvm.uadd.sat.v2i16(<2 x i16> [[TMP0]], <2 x i16> [[TMP1]])
340-
; GFX9-NEXT: [[TMP3:%.*]] = shufflevector <4 x i16> [[ARG0]], <4 x i16> poison, <2 x i32> <i32 2, i32 3>
329+
; GFX9-NEXT: [[TMP0:%.*]] = call <4 x i16> @llvm.uadd.sat.v4i16(<4 x i16> [[ARG0:%.*]], <4 x i16> [[ARG2:%.*]])
330+
; GFX9-NEXT: [[ARG1:%.*]] = call <4 x i16> @llvm.uadd.sat.v4i16(<4 x i16> [[ARG0]], <4 x i16> [[ARG2]])
341331
; GFX9-NEXT: [[TMP4:%.*]] = shufflevector <4 x i16> [[ARG1]], <4 x i16> poison, <2 x i32> <i32 2, i32 3>
342-
; GFX9-NEXT: [[TMP5:%.*]] = call <2 x i16> @llvm.uadd.sat.v2i16(<2 x i16> [[TMP3]], <2 x i16> [[TMP4]])
343-
; GFX9-NEXT: [[INS_31:%.*]] = shufflevector <2 x i16> [[TMP2]], <2 x i16> [[TMP5]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
332+
; GFX9-NEXT: [[TMP3:%.*]] = shufflevector <2 x i16> [[TMP4]], <2 x i16> poison, <4 x i32> <i32 0, i32 1, i32 poison, i32 poison>
333+
; GFX9-NEXT: [[INS_31:%.*]] = shufflevector <4 x i16> [[TMP0]], <4 x i16> [[TMP3]], <4 x i32> <i32 0, i32 1, i32 4, i32 5>
344334
; GFX9-NEXT: ret <4 x i16> [[INS_31]]
345335
;
346336
bb:

‎llvm/test/Transforms/SLPVectorizer/X86/alternate-calls-inseltpoison.ll

Lines changed: 18 additions & 18 deletions
Original file line numberDiff line numberDiff line change
@@ -38,13 +38,13 @@ define <8 x float> @ceil_floor(<8 x float> %a) {
3838
; SLM-NEXT: [[A0:%.*]] = extractelement <8 x float> [[A:%.*]], i64 0
3939
; SLM-NEXT: [[A3:%.*]] = extractelement <8 x float> [[A]], i64 3
4040
; SLM-NEXT: [[AB0:%.*]] = call float @llvm.ceil.f32(float [[A0]])
41-
; SLM-NEXT: [[TMP1:%.*]] = shufflevector <8 x float> [[A]], <8 x float> poison, <2 x i32> <i32 1, i32 2>
42-
; SLM-NEXT: [[TMP2:%.*]] = call <2 x float> @llvm.floor.v2f32(<2 x float> [[TMP1]])
41+
; SLM-NEXT: [[TMP1:%.*]] = call <8 x float> @llvm.floor.v8f32(<8 x float> [[A]])
42+
; SLM-NEXT: [[TMP2:%.*]] = shufflevector <8 x float> [[TMP1]], <8 x float> poison, <2 x i32> <i32 1, i32 2>
4343
; SLM-NEXT: [[AB3:%.*]] = call float @llvm.ceil.f32(float [[A3]])
44-
; SLM-NEXT: [[TMP3:%.*]] = shufflevector <8 x float> [[A]], <8 x float> poison, <2 x i32> <i32 4, i32 5>
45-
; SLM-NEXT: [[TMP4:%.*]] = call <2 x float> @llvm.ceil.v2f32(<2 x float> [[TMP3]])
46-
; SLM-NEXT: [[TMP5:%.*]] = shufflevector <8 x float> [[A]], <8 x float> poison, <2 x i32> <i32 6, i32 7>
47-
; SLM-NEXT: [[TMP6:%.*]] = call <2 x float> @llvm.floor.v2f32(<2 x float> [[TMP5]])
44+
; SLM-NEXT: [[TMP3:%.*]] = call <8 x float> @llvm.ceil.v8f32(<8 x float> [[A]])
45+
; SLM-NEXT: [[TMP4:%.*]] = shufflevector <8 x float> [[TMP3]], <8 x float> poison, <2 x i32> <i32 4, i32 5>
46+
; SLM-NEXT: [[TMP5:%.*]] = call <8 x float> @llvm.floor.v8f32(<8 x float> [[A]])
47+
; SLM-NEXT: [[TMP6:%.*]] = shufflevector <8 x float> [[TMP5]], <8 x float> poison, <2 x i32> <i32 6, i32 7>
4848
; SLM-NEXT: [[R0:%.*]] = insertelement <8 x float> poison, float [[AB0]], i64 0
4949
; SLM-NEXT: [[TMP7:%.*]] = shufflevector <2 x float> [[TMP2]], <2 x float> poison, <8 x i32> <i32 0, i32 1, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
5050
; SLM-NEXT: [[R23:%.*]] = shufflevector <8 x float> [[R0]], <8 x float> [[TMP7]], <8 x i32> <i32 0, i32 8, i32 9, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
@@ -59,13 +59,13 @@ define <8 x float> @ceil_floor(<8 x float> %a) {
5959
; AVX-NEXT: [[A0:%.*]] = extractelement <8 x float> [[A:%.*]], i64 0
6060
; AVX-NEXT: [[A3:%.*]] = extractelement <8 x float> [[A]], i64 3
6161
; AVX-NEXT: [[AB0:%.*]] = call float @llvm.ceil.f32(float [[A0]])
62-
; AVX-NEXT: [[TMP8:%.*]] = shufflevector <8 x float> [[A]], <8 x float> poison, <2 x i32> <i32 1, i32 2>
63-
; AVX-NEXT: [[TMP9:%.*]] = call <2 x float> @llvm.floor.v2f32(<2 x float> [[TMP8]])
62+
; AVX-NEXT: [[TMP1:%.*]] = call <8 x float> @llvm.floor.v8f32(<8 x float> [[A]])
63+
; AVX-NEXT: [[TMP9:%.*]] = shufflevector <8 x float> [[TMP1]], <8 x float> poison, <2 x i32> <i32 1, i32 2>
6464
; AVX-NEXT: [[AB3:%.*]] = call float @llvm.ceil.f32(float [[A3]])
65-
; AVX-NEXT: [[TMP1:%.*]] = shufflevector <8 x float> [[A]], <8 x float> poison, <2 x i32> <i32 4, i32 5>
66-
; AVX-NEXT: [[TMP2:%.*]] = call <2 x float> @llvm.ceil.v2f32(<2 x float> [[TMP1]])
67-
; AVX-NEXT: [[TMP3:%.*]] = shufflevector <8 x float> [[A]], <8 x float> poison, <2 x i32> <i32 6, i32 7>
68-
; AVX-NEXT: [[TMP4:%.*]] = call <2 x float> @llvm.floor.v2f32(<2 x float> [[TMP3]])
65+
; AVX-NEXT: [[TMP3:%.*]] = call <8 x float> @llvm.ceil.v8f32(<8 x float> [[A]])
66+
; AVX-NEXT: [[TMP2:%.*]] = shufflevector <8 x float> [[TMP3]], <8 x float> poison, <2 x i32> <i32 4, i32 5>
67+
; AVX-NEXT: [[TMP8:%.*]] = call <8 x float> @llvm.floor.v8f32(<8 x float> [[A]])
68+
; AVX-NEXT: [[TMP4:%.*]] = shufflevector <8 x float> [[TMP8]], <8 x float> poison, <2 x i32> <i32 6, i32 7>
6969
; AVX-NEXT: [[R0:%.*]] = insertelement <8 x float> poison, float [[AB0]], i64 0
7070
; AVX-NEXT: [[TMP7:%.*]] = shufflevector <2 x float> [[TMP9]], <2 x float> poison, <8 x i32> <i32 0, i32 1, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
7171
; AVX-NEXT: [[R2:%.*]] = shufflevector <8 x float> [[R0]], <8 x float> [[TMP7]], <8 x i32> <i32 0, i32 8, i32 9, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
@@ -80,13 +80,13 @@ define <8 x float> @ceil_floor(<8 x float> %a) {
8080
; AVX2-NEXT: [[A0:%.*]] = extractelement <8 x float> [[A:%.*]], i64 0
8181
; AVX2-NEXT: [[A3:%.*]] = extractelement <8 x float> [[A]], i64 3
8282
; AVX2-NEXT: [[AB0:%.*]] = call float @llvm.ceil.f32(float [[A0]])
83-
; AVX2-NEXT: [[TMP1:%.*]] = shufflevector <8 x float> [[A]], <8 x float> poison, <2 x i32> <i32 1, i32 2>
84-
; AVX2-NEXT: [[TMP2:%.*]] = call <2 x float> @llvm.floor.v2f32(<2 x float> [[TMP1]])
83+
; AVX2-NEXT: [[TMP1:%.*]] = call <8 x float> @llvm.floor.v8f32(<8 x float> [[A]])
84+
; AVX2-NEXT: [[TMP2:%.*]] = shufflevector <8 x float> [[TMP1]], <8 x float> poison, <2 x i32> <i32 1, i32 2>
8585
; AVX2-NEXT: [[AB3:%.*]] = call float @llvm.ceil.f32(float [[A3]])
86-
; AVX2-NEXT: [[TMP3:%.*]] = shufflevector <8 x float> [[A]], <8 x float> poison, <2 x i32> <i32 4, i32 5>
87-
; AVX2-NEXT: [[TMP4:%.*]] = call <2 x float> @llvm.ceil.v2f32(<2 x float> [[TMP3]])
88-
; AVX2-NEXT: [[TMP5:%.*]] = shufflevector <8 x float> [[A]], <8 x float> poison, <2 x i32> <i32 6, i32 7>
89-
; AVX2-NEXT: [[TMP6:%.*]] = call <2 x float> @llvm.floor.v2f32(<2 x float> [[TMP5]])
86+
; AVX2-NEXT: [[TMP3:%.*]] = call <8 x float> @llvm.ceil.v8f32(<8 x float> [[A]])
87+
; AVX2-NEXT: [[TMP4:%.*]] = shufflevector <8 x float> [[TMP3]], <8 x float> poison, <2 x i32> <i32 4, i32 5>
88+
; AVX2-NEXT: [[TMP5:%.*]] = call <8 x float> @llvm.floor.v8f32(<8 x float> [[A]])
89+
; AVX2-NEXT: [[TMP6:%.*]] = shufflevector <8 x float> [[TMP5]], <8 x float> poison, <2 x i32> <i32 6, i32 7>
9090
; AVX2-NEXT: [[R0:%.*]] = insertelement <8 x float> poison, float [[AB0]], i64 0
9191
; AVX2-NEXT: [[TMP7:%.*]] = shufflevector <2 x float> [[TMP2]], <2 x float> poison, <8 x i32> <i32 0, i32 1, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
9292
; AVX2-NEXT: [[R23:%.*]] = shufflevector <8 x float> [[R0]], <8 x float> [[TMP7]], <8 x i32> <i32 0, i32 8, i32 9, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>

0 commit comments

Comments
 (0)