Skip to content

Commit b020391

Browse files
gbossugbossu
committed
[AArch64] Update cost model for extracting halves from 128+ bit vectors
Previously, only 128-bit "NEON" vectors were given sensible costs. Cores with vscale>1 can use SVE's EXT instruction to perform a fixed-length subvector extract.
1 parent dc66670 commit b020391

File tree

2 files changed

+20
-18
lines changed

2 files changed

+20
-18
lines changed

‎llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp‎

Lines changed: 6 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -5750,11 +5750,13 @@ AArch64TTIImpl::getShuffleCost(TTI::ShuffleKind Kind, VectorType *DstTy,
57505750

57515751
Kind = improveShuffleKindFromMask(Kind, Mask, SrcTy, Index, SubTp);
57525752
bool IsExtractSubvector = Kind == TTI::SK_ExtractSubvector;
5753-
// A subvector extract can be implemented with an ext (or trivial extract, if
5754-
// from lane 0). This currently only handles low or high extracts to prevent
5755-
// SLP vectorizer regressions.
5753+
// A subvector extract can be implemented with a NEON/SVE ext (or trivial
5754+
// extract, if from lane 0). This currently only handles low or high extracts
5755+
// to prevent SLP vectorizer regressions.
5756+
// Note that SVE's ext instruciton is destructive, but it can be fused with
5757+
// a movprfx to act like a constructive instruction.
57565758
if (IsExtractSubvector && LT.second.isFixedLengthVector()) {
5757-
if (LT.second.is128BitVector() &&
5759+
if (LT.second.getFixedSizeInBits() >= AArch64::SVEBitsPerBlock &&
57585760
cast<FixedVectorType>(SubTp)->getNumElements() ==
57595761
LT.second.getVectorNumElements() / 2) {
57605762
if (Index == 0)

‎llvm/test/Analysis/CostModel/AArch64/shuffle-extract.ll‎

Lines changed: 14 additions & 14 deletions
Original file line numberDiff line numberDiff line change
@@ -108,25 +108,25 @@ define void @extract_half_vscale2() vscale_range(2,2) {
108108
; CHECK-LABEL: 'extract_half_vscale2'
109109
; CHECK-NEXT: Cost Model: Found costs of 0 for: %v32i8_lo = shufflevector <32 x i8> poison, <32 x i8> poison, <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 15>
110110
; CHECK-NEXT: Cost Model: Found costs of RThru:64 CodeSize:32 Lat:64 SizeLat:64 for: %v32i8_mi = shufflevector <32 x i8> poison, <32 x i8> poison, <16 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23>
111-
; CHECK-NEXT: Cost Model: Found costs of RThru:64 CodeSize:32 Lat:64 SizeLat:64 for: %v32i8_hi = shufflevector <32 x i8> poison, <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
111+
; CHECK-NEXT: Cost Model: Found costs of 1 for: %v32i8_hi = shufflevector <32 x i8> poison, <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
112112
; CHECK-NEXT: Cost Model: Found costs of 0 for: %v64i8_lo = shufflevector <64 x i8> poison, <64 x i8> poison, <32 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 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
113113
; CHECK-NEXT: Cost Model: Found costs of RThru:128 CodeSize:64 Lat:128 SizeLat:128 for: %v64i8_mi = shufflevector <64 x i8> poison, <64 x i8> poison, <32 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39>
114114
; CHECK-NEXT: Cost Model: Found costs of RThru:128 CodeSize:64 Lat:128 SizeLat:128 for: %v64i8_hi = shufflevector <64 x i8> poison, <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
115115
; CHECK-NEXT: Cost Model: Found costs of 0 for: %v16i16_lo = shufflevector <16 x i16> poison, <16 x i16> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
116116
; CHECK-NEXT: Cost Model: Found costs of RThru:32 CodeSize:16 Lat:32 SizeLat:32 for: %v16i16_mi = shufflevector <16 x i16> poison, <16 x i16> poison, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
117-
; CHECK-NEXT: Cost Model: Found costs of RThru:32 CodeSize:16 Lat:32 SizeLat:32 for: %v16i16_hi = shufflevector <16 x i16> poison, <16 x i16> poison, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
117+
; CHECK-NEXT: Cost Model: Found costs of 1 for: %v16i16_hi = shufflevector <16 x i16> poison, <16 x i16> poison, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
118118
; CHECK-NEXT: Cost Model: Found costs of 0 for: %v32i16_lo = shufflevector <32 x i16> poison, <32 x i16> poison, <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 15>
119119
; CHECK-NEXT: Cost Model: Found costs of RThru:64 CodeSize:32 Lat:64 SizeLat:64 for: %v32i16_mi = shufflevector <32 x i16> poison, <32 x i16> poison, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19>
120120
; CHECK-NEXT: Cost Model: Found costs of RThru:64 CodeSize:32 Lat:64 SizeLat:64 for: %v32i16_hi = shufflevector <32 x i16> poison, <32 x i16> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
121121
; CHECK-NEXT: Cost Model: Found costs of 0 for: %v8i32_lo = shufflevector <8 x i32> poison, <8 x i32> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
122122
; CHECK-NEXT: Cost Model: Found costs of RThru:16 CodeSize:8 Lat:16 SizeLat:16 for: %v8i32_mi = shufflevector <8 x i32> poison, <8 x i32> poison, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
123-
; CHECK-NEXT: Cost Model: Found costs of RThru:16 CodeSize:8 Lat:16 SizeLat:16 for: %v8i32_hi = shufflevector <8 x i32> poison, <8 x i32> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
123+
; CHECK-NEXT: Cost Model: Found costs of 1 for: %v8i32_hi = shufflevector <8 x i32> poison, <8 x i32> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
124124
; CHECK-NEXT: Cost Model: Found costs of 0 for: %v16i32_lo = shufflevector <16 x i32> poison, <16 x i32> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
125125
; CHECK-NEXT: Cost Model: Found costs of RThru:32 CodeSize:16 Lat:32 SizeLat:32 for: %v16i32_mi = shufflevector <16 x i32> poison, <16 x i32> poison, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
126126
; CHECK-NEXT: Cost Model: Found costs of RThru:32 CodeSize:16 Lat:32 SizeLat:32 for: %v16i32_hi = shufflevector <16 x i32> poison, <16 x i32> poison, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
127127
; CHECK-NEXT: Cost Model: Found costs of 0 for: %v4i64_lo = shufflevector <4 x i64> poison, <4 x i64> poison, <2 x i32> <i32 0, i32 1>
128128
; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %v4i64_mi = shufflevector <4 x i64> poison, <4 x i64> poison, <2 x i32> <i32 1, i32 2>
129-
; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %v4i64_hi = shufflevector <4 x i64> poison, <4 x i64> poison, <2 x i32> <i32 2, i32 3>
129+
; CHECK-NEXT: Cost Model: Found costs of 1 for: %v4i64_hi = shufflevector <4 x i64> poison, <4 x i64> poison, <2 x i32> <i32 2, i32 3>
130130
; CHECK-NEXT: Cost Model: Found costs of 0 for: %v8i64_lo = shufflevector <8 x i64> poison, <8 x i64> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
131131
; CHECK-NEXT: Cost Model: Found costs of RThru:16 CodeSize:8 Lat:16 SizeLat:16 for: %v8i64_mi = shufflevector <8 x i64> poison, <8 x i64> poison, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
132132
; CHECK-NEXT: Cost Model: Found costs of RThru:16 CodeSize:8 Lat:16 SizeLat:16 for: %v8i64_hi = shufflevector <8 x i64> poison, <8 x i64> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
@@ -168,28 +168,28 @@ define void @extract_half_vscale4() vscale_range(4,4) {
168168
; CHECK-LABEL: 'extract_half_vscale4'
169169
; CHECK-NEXT: Cost Model: Found costs of 0 for: %v32i8_lo = shufflevector <32 x i8> poison, <32 x i8> poison, <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 15>
170170
; CHECK-NEXT: Cost Model: Found costs of RThru:64 CodeSize:32 Lat:64 SizeLat:64 for: %v32i8_mi = shufflevector <32 x i8> poison, <32 x i8> poison, <16 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23>
171-
; CHECK-NEXT: Cost Model: Found costs of RThru:64 CodeSize:32 Lat:64 SizeLat:64 for: %v32i8_hi = shufflevector <32 x i8> poison, <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
171+
; CHECK-NEXT: Cost Model: Found costs of 1 for: %v32i8_hi = shufflevector <32 x i8> poison, <32 x i8> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
172172
; CHECK-NEXT: Cost Model: Found costs of 0 for: %v64i8_lo = shufflevector <64 x i8> poison, <64 x i8> poison, <32 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 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
173173
; CHECK-NEXT: Cost Model: Found costs of RThru:128 CodeSize:64 Lat:128 SizeLat:128 for: %v64i8_mi = shufflevector <64 x i8> poison, <64 x i8> poison, <32 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39>
174-
; CHECK-NEXT: Cost Model: Found costs of RThru:128 CodeSize:64 Lat:128 SizeLat:128 for: %v64i8_hi = shufflevector <64 x i8> poison, <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
174+
; CHECK-NEXT: Cost Model: Found costs of 1 for: %v64i8_hi = shufflevector <64 x i8> poison, <64 x i8> poison, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
175175
; CHECK-NEXT: Cost Model: Found costs of 0 for: %v16i16_lo = shufflevector <16 x i16> poison, <16 x i16> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
176176
; CHECK-NEXT: Cost Model: Found costs of RThru:32 CodeSize:16 Lat:32 SizeLat:32 for: %v16i16_mi = shufflevector <16 x i16> poison, <16 x i16> poison, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
177-
; CHECK-NEXT: Cost Model: Found costs of RThru:32 CodeSize:16 Lat:32 SizeLat:32 for: %v16i16_hi = shufflevector <16 x i16> poison, <16 x i16> poison, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
177+
; CHECK-NEXT: Cost Model: Found costs of 1 for: %v16i16_hi = shufflevector <16 x i16> poison, <16 x i16> poison, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
178178
; CHECK-NEXT: Cost Model: Found costs of 0 for: %v32i16_lo = shufflevector <32 x i16> poison, <32 x i16> poison, <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 15>
179179
; CHECK-NEXT: Cost Model: Found costs of RThru:64 CodeSize:32 Lat:64 SizeLat:64 for: %v32i16_mi = shufflevector <32 x i16> poison, <32 x i16> poison, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19>
180-
; CHECK-NEXT: Cost Model: Found costs of RThru:64 CodeSize:32 Lat:64 SizeLat:64 for: %v32i16_hi = shufflevector <32 x i16> poison, <32 x i16> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
180+
; CHECK-NEXT: Cost Model: Found costs of 1 for: %v32i16_hi = shufflevector <32 x i16> poison, <32 x i16> poison, <16 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
181181
; CHECK-NEXT: Cost Model: Found costs of 0 for: %v8i32_lo = shufflevector <8 x i32> poison, <8 x i32> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
182182
; CHECK-NEXT: Cost Model: Found costs of RThru:16 CodeSize:8 Lat:16 SizeLat:16 for: %v8i32_mi = shufflevector <8 x i32> poison, <8 x i32> poison, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
183-
; CHECK-NEXT: Cost Model: Found costs of RThru:16 CodeSize:8 Lat:16 SizeLat:16 for: %v8i32_hi = shufflevector <8 x i32> poison, <8 x i32> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
183+
; CHECK-NEXT: Cost Model: Found costs of 1 for: %v8i32_hi = shufflevector <8 x i32> poison, <8 x i32> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
184184
; CHECK-NEXT: Cost Model: Found costs of 0 for: %v16i32_lo = shufflevector <16 x i32> poison, <16 x i32> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
185185
; CHECK-NEXT: Cost Model: Found costs of RThru:32 CodeSize:16 Lat:32 SizeLat:32 for: %v16i32_mi = shufflevector <16 x i32> poison, <16 x i32> poison, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
186-
; CHECK-NEXT: Cost Model: Found costs of RThru:32 CodeSize:16 Lat:32 SizeLat:32 for: %v16i32_hi = shufflevector <16 x i32> poison, <16 x i32> poison, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
186+
; CHECK-NEXT: Cost Model: Found costs of 1 for: %v16i32_hi = shufflevector <16 x i32> poison, <16 x i32> poison, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
187187
; CHECK-NEXT: Cost Model: Found costs of 0 for: %v4i64_lo = shufflevector <4 x i64> poison, <4 x i64> poison, <2 x i32> <i32 0, i32 1>
188188
; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %v4i64_mi = shufflevector <4 x i64> poison, <4 x i64> poison, <2 x i32> <i32 1, i32 2>
189-
; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %v4i64_hi = shufflevector <4 x i64> poison, <4 x i64> poison, <2 x i32> <i32 2, i32 3>
189+
; CHECK-NEXT: Cost Model: Found costs of 1 for: %v4i64_hi = shufflevector <4 x i64> poison, <4 x i64> poison, <2 x i32> <i32 2, i32 3>
190190
; CHECK-NEXT: Cost Model: Found costs of 0 for: %v8i64_lo = shufflevector <8 x i64> poison, <8 x i64> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
191191
; CHECK-NEXT: Cost Model: Found costs of RThru:16 CodeSize:8 Lat:16 SizeLat:16 for: %v8i64_mi = shufflevector <8 x i64> poison, <8 x i64> poison, <4 x i32> <i32 2, i32 3, i32 4, i32 5>
192-
; CHECK-NEXT: Cost Model: Found costs of RThru:16 CodeSize:8 Lat:16 SizeLat:16 for: %v8i64_hi = shufflevector <8 x i64> poison, <8 x i64> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
192+
; CHECK-NEXT: Cost Model: Found costs of 1 for: %v8i64_hi = shufflevector <8 x i64> poison, <8 x i64> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
193193
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret void
194194
;
195195
%v32i8_lo = shufflevector <32 x i8> poison, <32 x i8> poison, <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 15>
@@ -320,13 +320,13 @@ define void @extract_qtr_vscale2() vscale_range(2,2) {
320320
; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %v8i32_mi = shufflevector <8 x i32> poison, <8 x i32> poison, <2 x i32> <i32 2, i32 3>
321321
; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %v8i32_hi = shufflevector <8 x i32> poison, <8 x i32> poison, <2 x i32> <i32 4, i32 5>
322322
; CHECK-NEXT: Cost Model: Found costs of 0 for: %v16i32_lo = shufflevector <16 x i32> poison, <16 x i32> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
323-
; CHECK-NEXT: Cost Model: Found costs of RThru:16 CodeSize:8 Lat:16 SizeLat:16 for: %v16i32_mi = shufflevector <16 x i32> poison, <16 x i32> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
323+
; CHECK-NEXT: Cost Model: Found costs of 1 for: %v16i32_mi = shufflevector <16 x i32> poison, <16 x i32> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
324324
; CHECK-NEXT: Cost Model: Found costs of RThru:16 CodeSize:8 Lat:16 SizeLat:16 for: %v16i32_hi = shufflevector <16 x i32> poison, <16 x i32> poison, <4 x i32> <i32 8, i32 9, i32 10, i32 11>
325325
; CHECK-NEXT: Cost Model: Found costs of 0 for: %v4i64_lo = shufflevector <4 x i64> poison, <4 x i64> poison, <1 x i32> zeroinitializer
326326
; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %v4i64_mi = shufflevector <4 x i64> poison, <4 x i64> poison, <1 x i32> <i32 1>
327327
; CHECK-NEXT: Cost Model: Found costs of RThru:4 CodeSize:2 Lat:4 SizeLat:4 for: %v4i64_hi = shufflevector <4 x i64> poison, <4 x i64> poison, <1 x i32> <i32 2>
328328
; CHECK-NEXT: Cost Model: Found costs of 0 for: %v8i64_lo = shufflevector <8 x i64> poison, <8 x i64> poison, <2 x i32> <i32 0, i32 1>
329-
; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %v8i64_mi = shufflevector <8 x i64> poison, <8 x i64> poison, <2 x i32> <i32 2, i32 3>
329+
; CHECK-NEXT: Cost Model: Found costs of 1 for: %v8i64_mi = shufflevector <8 x i64> poison, <8 x i64> poison, <2 x i32> <i32 2, i32 3>
330330
; CHECK-NEXT: Cost Model: Found costs of RThru:8 CodeSize:4 Lat:8 SizeLat:8 for: %v8i64_hi = shufflevector <8 x i64> poison, <8 x i64> poison, <2 x i32> <i32 4, i32 5>
331331
; CHECK-NEXT: Cost Model: Found costs of RThru:0 CodeSize:1 Lat:1 SizeLat:1 for: ret void
332332
;

0 commit comments

Comments
 (0)