Skip to content

Commit b16ce8f

Browse files
RKSimonRKSimon
authored
[X86] getFauxShuffleMask - match 256-bit CONCAT(SUB0, SUB1) 64-bit elt patterns as well as 512-bit (#127392)
The 512-bit filter was to prevent AVX1/2 regressions, but most of that is now handled by canonicalizeShuffleWithOp Ideally we need to support smaller element widths as well. Noticed while triaging #116931
1 parent 153dd19 commit b16ce8f

15 files changed

+676
-827
lines changed

‎llvm/lib/Target/X86/X86ISelLowering.cpp‎

Lines changed: 2 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -6130,11 +6130,9 @@ static bool getFauxShuffleMask(SDValue N, const APInt &DemandedElts,
61306130
return true;
61316131
}
61326132
// Handle CONCAT(SUB0, SUB1).
6133-
// Limit this to vXi64 512-bit vector cases to make the most of AVX512
6134-
// cross lane shuffles.
6133+
// Limit this to vXi64 vector cases to make the most of cross lane shuffles.
61356134
if (Depth > 0 && InsertIdx == NumSubElts && NumElts == (2 * NumSubElts) &&
6136-
NumBitsPerElt == 64 && NumSizeInBits == 512 &&
6137-
Src.getOpcode() == ISD::INSERT_SUBVECTOR &&
6135+
NumBitsPerElt == 64 && Src.getOpcode() == ISD::INSERT_SUBVECTOR &&
61386136
Src.getOperand(0).isUndef() &&
61396137
Src.getOperand(1).getValueType() == SubVT &&
61406138
Src.getConstantOperandVal(2) == 0) {

‎llvm/test/CodeGen/X86/vector-interleaved-store-i16-stride-3.ll‎

Lines changed: 41 additions & 45 deletions
Original file line numberDiff line numberDiff line change
@@ -211,10 +211,10 @@ define void @store_i16_stride3_vf4(ptr %in.vecptr0, ptr %in.vecptr1, ptr %in.vec
211211
; AVX2-NEXT: vmovq {{.*#+}} xmm1 = mem[0],zero
212212
; AVX2-NEXT: vmovq {{.*#+}} xmm2 = mem[0],zero
213213
; AVX2-NEXT: vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
214-
; AVX2-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm0
215-
; AVX2-NEXT: vpshufb {{.*#+}} ymm1 = ymm0[0,1,8,9],zero,zero,ymm0[2,3,10,11],zero,zero,ymm0[4,5,12,13,20,21],zero,zero,zero,zero,ymm0[22,23],zero,zero,zero,zero,zero,zero,zero,zero
216-
; AVX2-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
214+
; AVX2-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm2
215+
; AVX2-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
217216
; AVX2-NEXT: vpshufb {{.*#+}} ymm0 = zero,zero,zero,zero,ymm0[0,1],zero,zero,zero,zero,ymm0[2,3],zero,zero,zero,zero,zero,zero,ymm0[22,23,30,31],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
217+
; AVX2-NEXT: vpshufb {{.*#+}} ymm1 = ymm2[0,1,8,9],zero,zero,ymm2[2,3,10,11],zero,zero,ymm2[4,5,12,13,20,21],zero,zero,zero,zero,ymm2[22,23],zero,zero,zero,zero,zero,zero,zero,zero
218218
; AVX2-NEXT: vpor %ymm0, %ymm1, %ymm0
219219
; AVX2-NEXT: vextracti128 $1, %ymm0, %xmm1
220220
; AVX2-NEXT: vmovq %xmm1, 16(%rcx)
@@ -228,10 +228,10 @@ define void @store_i16_stride3_vf4(ptr %in.vecptr0, ptr %in.vecptr1, ptr %in.vec
228228
; AVX2-FP-NEXT: vmovq {{.*#+}} xmm1 = mem[0],zero
229229
; AVX2-FP-NEXT: vmovq {{.*#+}} xmm2 = mem[0],zero
230230
; AVX2-FP-NEXT: vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
231-
; AVX2-FP-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm0
232-
; AVX2-FP-NEXT: vpshufb {{.*#+}} ymm1 = ymm0[0,1,8,9],zero,zero,ymm0[2,3,10,11],zero,zero,ymm0[4,5,12,13,20,21],zero,zero,zero,zero,ymm0[22,23],zero,zero,zero,zero,zero,zero,zero,zero
233-
; AVX2-FP-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
231+
; AVX2-FP-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm2
232+
; AVX2-FP-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
234233
; AVX2-FP-NEXT: vpshufb {{.*#+}} ymm0 = zero,zero,zero,zero,ymm0[0,1],zero,zero,zero,zero,ymm0[2,3],zero,zero,zero,zero,zero,zero,ymm0[22,23,30,31],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
234+
; AVX2-FP-NEXT: vpshufb {{.*#+}} ymm1 = ymm2[0,1,8,9],zero,zero,ymm2[2,3,10,11],zero,zero,ymm2[4,5,12,13,20,21],zero,zero,zero,zero,ymm2[22,23],zero,zero,zero,zero,zero,zero,zero,zero
235235
; AVX2-FP-NEXT: vpor %ymm0, %ymm1, %ymm0
236236
; AVX2-FP-NEXT: vextracti128 $1, %ymm0, %xmm1
237237
; AVX2-FP-NEXT: vmovq %xmm1, 16(%rcx)
@@ -245,10 +245,10 @@ define void @store_i16_stride3_vf4(ptr %in.vecptr0, ptr %in.vecptr1, ptr %in.vec
245245
; AVX2-FCP-NEXT: vmovq {{.*#+}} xmm1 = mem[0],zero
246246
; AVX2-FCP-NEXT: vmovq {{.*#+}} xmm2 = mem[0],zero
247247
; AVX2-FCP-NEXT: vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
248-
; AVX2-FCP-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm0
249-
; AVX2-FCP-NEXT: vpshufb {{.*#+}} ymm1 = ymm0[0,1,8,9],zero,zero,ymm0[2,3,10,11],zero,zero,ymm0[4,5,12,13,20,21],zero,zero,zero,zero,ymm0[22,23],zero,zero,zero,zero,zero,zero,zero,zero
250-
; AVX2-FCP-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
248+
; AVX2-FCP-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm2
249+
; AVX2-FCP-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
251250
; AVX2-FCP-NEXT: vpshufb {{.*#+}} ymm0 = zero,zero,zero,zero,ymm0[0,1],zero,zero,zero,zero,ymm0[2,3],zero,zero,zero,zero,zero,zero,ymm0[22,23,30,31],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
251+
; AVX2-FCP-NEXT: vpshufb {{.*#+}} ymm1 = ymm2[0,1,8,9],zero,zero,ymm2[2,3,10,11],zero,zero,ymm2[4,5,12,13,20,21],zero,zero,zero,zero,ymm2[22,23],zero,zero,zero,zero,zero,zero,zero,zero
252252
; AVX2-FCP-NEXT: vpor %ymm0, %ymm1, %ymm0
253253
; AVX2-FCP-NEXT: vextracti128 $1, %ymm0, %xmm1
254254
; AVX2-FCP-NEXT: vmovq %xmm1, 16(%rcx)
@@ -262,10 +262,10 @@ define void @store_i16_stride3_vf4(ptr %in.vecptr0, ptr %in.vecptr1, ptr %in.vec
262262
; AVX512-NEXT: vmovq {{.*#+}} xmm1 = mem[0],zero
263263
; AVX512-NEXT: vmovq {{.*#+}} xmm2 = mem[0],zero
264264
; AVX512-NEXT: vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
265-
; AVX512-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm0
266-
; AVX512-NEXT: vpshufb {{.*#+}} ymm1 = ymm0[0,1,8,9],zero,zero,ymm0[2,3,10,11],zero,zero,ymm0[4,5,12,13,20,21],zero,zero,zero,zero,ymm0[22,23,u,u,u,u,u,u,u,u]
267-
; AVX512-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
265+
; AVX512-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm2
266+
; AVX512-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
268267
; AVX512-NEXT: vpshufb {{.*#+}} ymm0 = zero,zero,zero,zero,ymm0[0,1],zero,zero,zero,zero,ymm0[2,3],zero,zero,zero,zero,zero,zero,ymm0[22,23,30,31],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
268+
; AVX512-NEXT: vpshufb {{.*#+}} ymm1 = ymm2[0,1,8,9],zero,zero,ymm2[2,3,10,11],zero,zero,ymm2[4,5,12,13,20,21],zero,zero,zero,zero,ymm2[22,23,u,u,u,u,u,u,u,u]
269269
; AVX512-NEXT: vpor %ymm0, %ymm1, %ymm0
270270
; AVX512-NEXT: vextracti128 $1, %ymm0, %xmm1
271271
; AVX512-NEXT: vmovq %xmm1, 16(%rcx)
@@ -279,10 +279,10 @@ define void @store_i16_stride3_vf4(ptr %in.vecptr0, ptr %in.vecptr1, ptr %in.vec
279279
; AVX512-FCP-NEXT: vmovq {{.*#+}} xmm1 = mem[0],zero
280280
; AVX512-FCP-NEXT: vmovq {{.*#+}} xmm2 = mem[0],zero
281281
; AVX512-FCP-NEXT: vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
282-
; AVX512-FCP-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm0
283-
; AVX512-FCP-NEXT: vpshufb {{.*#+}} ymm1 = ymm0[0,1,8,9],zero,zero,ymm0[2,3,10,11],zero,zero,ymm0[4,5,12,13,20,21],zero,zero,zero,zero,ymm0[22,23,u,u,u,u,u,u,u,u]
284-
; AVX512-FCP-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
282+
; AVX512-FCP-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm2
283+
; AVX512-FCP-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
285284
; AVX512-FCP-NEXT: vpshufb {{.*#+}} ymm0 = zero,zero,zero,zero,ymm0[0,1],zero,zero,zero,zero,ymm0[2,3],zero,zero,zero,zero,zero,zero,ymm0[22,23,30,31],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
285+
; AVX512-FCP-NEXT: vpshufb {{.*#+}} ymm1 = ymm2[0,1,8,9],zero,zero,ymm2[2,3,10,11],zero,zero,ymm2[4,5,12,13,20,21],zero,zero,zero,zero,ymm2[22,23,u,u,u,u,u,u,u,u]
286286
; AVX512-FCP-NEXT: vpor %ymm0, %ymm1, %ymm0
287287
; AVX512-FCP-NEXT: vextracti128 $1, %ymm0, %xmm1
288288
; AVX512-FCP-NEXT: vmovq %xmm1, 16(%rcx)
@@ -296,10 +296,10 @@ define void @store_i16_stride3_vf4(ptr %in.vecptr0, ptr %in.vecptr1, ptr %in.vec
296296
; AVX512DQ-NEXT: vmovq {{.*#+}} xmm1 = mem[0],zero
297297
; AVX512DQ-NEXT: vmovq {{.*#+}} xmm2 = mem[0],zero
298298
; AVX512DQ-NEXT: vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
299-
; AVX512DQ-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm0
300-
; AVX512DQ-NEXT: vpshufb {{.*#+}} ymm1 = ymm0[0,1,8,9],zero,zero,ymm0[2,3,10,11],zero,zero,ymm0[4,5,12,13,20,21],zero,zero,zero,zero,ymm0[22,23,u,u,u,u,u,u,u,u]
301-
; AVX512DQ-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
299+
; AVX512DQ-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm2
300+
; AVX512DQ-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
302301
; AVX512DQ-NEXT: vpshufb {{.*#+}} ymm0 = zero,zero,zero,zero,ymm0[0,1],zero,zero,zero,zero,ymm0[2,3],zero,zero,zero,zero,zero,zero,ymm0[22,23,30,31],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
302+
; AVX512DQ-NEXT: vpshufb {{.*#+}} ymm1 = ymm2[0,1,8,9],zero,zero,ymm2[2,3,10,11],zero,zero,ymm2[4,5,12,13,20,21],zero,zero,zero,zero,ymm2[22,23,u,u,u,u,u,u,u,u]
303303
; AVX512DQ-NEXT: vpor %ymm0, %ymm1, %ymm0
304304
; AVX512DQ-NEXT: vextracti128 $1, %ymm0, %xmm1
305305
; AVX512DQ-NEXT: vmovq %xmm1, 16(%rcx)
@@ -313,10 +313,10 @@ define void @store_i16_stride3_vf4(ptr %in.vecptr0, ptr %in.vecptr1, ptr %in.vec
313313
; AVX512DQ-FCP-NEXT: vmovq {{.*#+}} xmm1 = mem[0],zero
314314
; AVX512DQ-FCP-NEXT: vmovq {{.*#+}} xmm2 = mem[0],zero
315315
; AVX512DQ-FCP-NEXT: vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
316-
; AVX512DQ-FCP-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm0
317-
; AVX512DQ-FCP-NEXT: vpshufb {{.*#+}} ymm1 = ymm0[0,1,8,9],zero,zero,ymm0[2,3,10,11],zero,zero,ymm0[4,5,12,13,20,21],zero,zero,zero,zero,ymm0[22,23,u,u,u,u,u,u,u,u]
318-
; AVX512DQ-FCP-NEXT: vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
316+
; AVX512DQ-FCP-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm2
317+
; AVX512DQ-FCP-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
319318
; AVX512DQ-FCP-NEXT: vpshufb {{.*#+}} ymm0 = zero,zero,zero,zero,ymm0[0,1],zero,zero,zero,zero,ymm0[2,3],zero,zero,zero,zero,zero,zero,ymm0[22,23,30,31],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
319+
; AVX512DQ-FCP-NEXT: vpshufb {{.*#+}} ymm1 = ymm2[0,1,8,9],zero,zero,ymm2[2,3,10,11],zero,zero,ymm2[4,5,12,13,20,21],zero,zero,zero,zero,ymm2[22,23,u,u,u,u,u,u,u,u]
320320
; AVX512DQ-FCP-NEXT: vpor %ymm0, %ymm1, %ymm0
321321
; AVX512DQ-FCP-NEXT: vextracti128 $1, %ymm0, %xmm1
322322
; AVX512DQ-FCP-NEXT: vmovq %xmm1, 16(%rcx)
@@ -330,12 +330,11 @@ define void @store_i16_stride3_vf4(ptr %in.vecptr0, ptr %in.vecptr1, ptr %in.vec
330330
; AVX512BW-NEXT: vmovq {{.*#+}} xmm1 = mem[0],zero
331331
; AVX512BW-NEXT: vmovq {{.*#+}} xmm2 = mem[0],zero
332332
; AVX512BW-NEXT: vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
333-
; AVX512BW-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm0
334-
; AVX512BW-NEXT: vpmovsxbw {{.*#+}} ymm1 = [0,4,8,1,5,9,2,6,10,3,7,11,0,0,0,0]
335-
; AVX512BW-NEXT: vpermw %ymm0, %ymm1, %ymm0
336-
; AVX512BW-NEXT: vextracti128 $1, %ymm0, %xmm1
337-
; AVX512BW-NEXT: vmovq %xmm1, 16(%rcx)
338-
; AVX512BW-NEXT: vmovdqa %xmm0, (%rcx)
333+
; AVX512BW-NEXT: vpmovsxbw {{.*#+}} ymm2 = [0,4,16,1,5,17,2,6,18,3,7,19,0,0,0,0]
334+
; AVX512BW-NEXT: vpermi2w %ymm0, %ymm1, %ymm2
335+
; AVX512BW-NEXT: vextracti128 $1, %ymm2, %xmm0
336+
; AVX512BW-NEXT: vmovq %xmm0, 16(%rcx)
337+
; AVX512BW-NEXT: vmovdqa %xmm2, (%rcx)
339338
; AVX512BW-NEXT: vzeroupper
340339
; AVX512BW-NEXT: retq
341340
;
@@ -345,12 +344,11 @@ define void @store_i16_stride3_vf4(ptr %in.vecptr0, ptr %in.vecptr1, ptr %in.vec
345344
; AVX512BW-FCP-NEXT: vmovq {{.*#+}} xmm1 = mem[0],zero
346345
; AVX512BW-FCP-NEXT: vmovq {{.*#+}} xmm2 = mem[0],zero
347346
; AVX512BW-FCP-NEXT: vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
348-
; AVX512BW-FCP-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm0
349-
; AVX512BW-FCP-NEXT: vpmovsxbw {{.*#+}} ymm1 = [0,4,8,1,5,9,2,6,10,3,7,11,0,0,0,0]
350-
; AVX512BW-FCP-NEXT: vpermw %ymm0, %ymm1, %ymm0
351-
; AVX512BW-FCP-NEXT: vextracti128 $1, %ymm0, %xmm1
352-
; AVX512BW-FCP-NEXT: vmovq %xmm1, 16(%rcx)
353-
; AVX512BW-FCP-NEXT: vmovdqa %xmm0, (%rcx)
347+
; AVX512BW-FCP-NEXT: vpmovsxbw {{.*#+}} ymm2 = [0,4,16,1,5,17,2,6,18,3,7,19,0,0,0,0]
348+
; AVX512BW-FCP-NEXT: vpermi2w %ymm0, %ymm1, %ymm2
349+
; AVX512BW-FCP-NEXT: vextracti128 $1, %ymm2, %xmm0
350+
; AVX512BW-FCP-NEXT: vmovq %xmm0, 16(%rcx)
351+
; AVX512BW-FCP-NEXT: vmovdqa %xmm2, (%rcx)
354352
; AVX512BW-FCP-NEXT: vzeroupper
355353
; AVX512BW-FCP-NEXT: retq
356354
;
@@ -360,12 +358,11 @@ define void @store_i16_stride3_vf4(ptr %in.vecptr0, ptr %in.vecptr1, ptr %in.vec
360358
; AVX512DQ-BW-NEXT: vmovq {{.*#+}} xmm1 = mem[0],zero
361359
; AVX512DQ-BW-NEXT: vmovq {{.*#+}} xmm2 = mem[0],zero
362360
; AVX512DQ-BW-NEXT: vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
363-
; AVX512DQ-BW-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm0
364-
; AVX512DQ-BW-NEXT: vpmovsxbw {{.*#+}} ymm1 = [0,4,8,1,5,9,2,6,10,3,7,11,0,0,0,0]
365-
; AVX512DQ-BW-NEXT: vpermw %ymm0, %ymm1, %ymm0
366-
; AVX512DQ-BW-NEXT: vextracti128 $1, %ymm0, %xmm1
367-
; AVX512DQ-BW-NEXT: vmovq %xmm1, 16(%rcx)
368-
; AVX512DQ-BW-NEXT: vmovdqa %xmm0, (%rcx)
361+
; AVX512DQ-BW-NEXT: vpmovsxbw {{.*#+}} ymm2 = [0,4,16,1,5,17,2,6,18,3,7,19,0,0,0,0]
362+
; AVX512DQ-BW-NEXT: vpermi2w %ymm0, %ymm1, %ymm2
363+
; AVX512DQ-BW-NEXT: vextracti128 $1, %ymm2, %xmm0
364+
; AVX512DQ-BW-NEXT: vmovq %xmm0, 16(%rcx)
365+
; AVX512DQ-BW-NEXT: vmovdqa %xmm2, (%rcx)
369366
; AVX512DQ-BW-NEXT: vzeroupper
370367
; AVX512DQ-BW-NEXT: retq
371368
;
@@ -375,12 +372,11 @@ define void @store_i16_stride3_vf4(ptr %in.vecptr0, ptr %in.vecptr1, ptr %in.vec
375372
; AVX512DQ-BW-FCP-NEXT: vmovq {{.*#+}} xmm1 = mem[0],zero
376373
; AVX512DQ-BW-FCP-NEXT: vmovq {{.*#+}} xmm2 = mem[0],zero
377374
; AVX512DQ-BW-FCP-NEXT: vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
378-
; AVX512DQ-BW-FCP-NEXT: vinserti128 $1, %xmm0, %ymm1, %ymm0
379-
; AVX512DQ-BW-FCP-NEXT: vpmovsxbw {{.*#+}} ymm1 = [0,4,8,1,5,9,2,6,10,3,7,11,0,0,0,0]
380-
; AVX512DQ-BW-FCP-NEXT: vpermw %ymm0, %ymm1, %ymm0
381-
; AVX512DQ-BW-FCP-NEXT: vextracti128 $1, %ymm0, %xmm1
382-
; AVX512DQ-BW-FCP-NEXT: vmovq %xmm1, 16(%rcx)
383-
; AVX512DQ-BW-FCP-NEXT: vmovdqa %xmm0, (%rcx)
375+
; AVX512DQ-BW-FCP-NEXT: vpmovsxbw {{.*#+}} ymm2 = [0,4,16,1,5,17,2,6,18,3,7,19,0,0,0,0]
376+
; AVX512DQ-BW-FCP-NEXT: vpermi2w %ymm0, %ymm1, %ymm2
377+
; AVX512DQ-BW-FCP-NEXT: vextracti128 $1, %ymm2, %xmm0
378+
; AVX512DQ-BW-FCP-NEXT: vmovq %xmm0, 16(%rcx)
379+
; AVX512DQ-BW-FCP-NEXT: vmovdqa %xmm2, (%rcx)
384380
; AVX512DQ-BW-FCP-NEXT: vzeroupper
385381
; AVX512DQ-BW-FCP-NEXT: retq
386382
%in.vec0 = load <4 x i16>, ptr %in.vecptr0, align 64

0 commit comments

Comments
 (0)