Skip to content

Commit 0cf9e47

Browse files
davemgreendavemgreen
committed
[AArch64] Add SK_Splice fixed-width costs
A fixed length SK_Splice shuffle vector is lowered to a Ext under AArch64, which should have a cost of 1. Differential Revision: https://reviews.llvm.org/D132299
1 parent d0397db commit 0cf9e47

File tree

2 files changed

+47
-29
lines changed

2 files changed

+47
-29
lines changed

‎llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp

Lines changed: 16 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -2909,8 +2909,7 @@ InstructionCost AArch64TTIImpl::getShuffleCost(TTI::ShuffleKind Kind,
29092909

29102910
if (Kind == TTI::SK_Broadcast || Kind == TTI::SK_Transpose ||
29112911
Kind == TTI::SK_Select || Kind == TTI::SK_PermuteSingleSrc ||
2912-
Kind == TTI::SK_Reverse) {
2913-
2912+
Kind == TTI::SK_Reverse || Kind == TTI::SK_Splice) {
29142913
static const CostTblEntry ShuffleTbl[] = {
29152914
// Broadcast shuffle kinds can be performed with 'dup'.
29162915
{ TTI::SK_Broadcast, MVT::v8i8, 1 },
@@ -2971,6 +2970,21 @@ InstructionCost AArch64TTIImpl::getShuffleCost(TTI::ShuffleKind Kind,
29712970
{ TTI::SK_Reverse, MVT::v4f16, 1 }, // REV64
29722971
{ TTI::SK_Reverse, MVT::v4i16, 1 }, // REV64
29732972
{ TTI::SK_Reverse, MVT::v8i8, 1 }, // REV64
2973+
// Splice can all be lowered as `ext`.
2974+
{ TTI::SK_Splice, MVT::v2i32, 1 },
2975+
{ TTI::SK_Splice, MVT::v4i32, 1 },
2976+
{ TTI::SK_Splice, MVT::v2i64, 1 },
2977+
{ TTI::SK_Splice, MVT::v2f32, 1 },
2978+
{ TTI::SK_Splice, MVT::v4f32, 1 },
2979+
{ TTI::SK_Splice, MVT::v2f64, 1 },
2980+
{ TTI::SK_Splice, MVT::v8f16, 1 },
2981+
{ TTI::SK_Splice, MVT::v8bf16, 1 },
2982+
{ TTI::SK_Splice, MVT::v8i16, 1 },
2983+
{ TTI::SK_Splice, MVT::v16i8, 1 },
2984+
{ TTI::SK_Splice, MVT::v4bf16, 1 },
2985+
{ TTI::SK_Splice, MVT::v4f16, 1 },
2986+
{ TTI::SK_Splice, MVT::v4i16, 1 },
2987+
{ TTI::SK_Splice, MVT::v8i8, 1 },
29742988
// Broadcast shuffle kinds for scalable vectors
29752989
{ TTI::SK_Broadcast, MVT::nxv16i8, 1 },
29762990
{ TTI::SK_Broadcast, MVT::nxv8i16, 1 },

‎llvm/test/Analysis/CostModel/AArch64/splice.ll

Lines changed: 31 additions & 27 deletions
Original file line numberDiff line numberDiff line change
@@ -3,33 +3,35 @@
33

44
define void @vector_splice() #0 {
55
; CHECK-LABEL: 'vector_splice'
6-
; CHECK-NEXT: Cost Model: Found an estimated cost of 90 for instruction: %splice.v16i8 = call <16 x i8> @llvm.experimental.vector.splice.v16i8(<16 x i8> zeroinitializer, <16 x i8> zeroinitializer, i32 1)
7-
; CHECK-NEXT: Cost Model: Found an estimated cost of 180 for instruction: %splice.v32i8 = call <32 x i8> @llvm.experimental.vector.splice.v32i8(<32 x i8> zeroinitializer, <32 x i8> zeroinitializer, i32 1)
8-
; CHECK-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %splice.v2i16 = call <2 x i16> @llvm.experimental.vector.splice.v2i16(<2 x i16> zeroinitializer, <2 x i16> zeroinitializer, i32 1)
9-
; CHECK-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %splice.v4i16 = call <4 x i16> @llvm.experimental.vector.splice.v4i16(<4 x i16> zeroinitializer, <4 x i16> zeroinitializer, i32 1)
10-
; CHECK-NEXT: Cost Model: Found an estimated cost of 42 for instruction: %splice.v8i16 = call <8 x i16> @llvm.experimental.vector.splice.v8i16(<8 x i16> zeroinitializer, <8 x i16> zeroinitializer, i32 1)
11-
; CHECK-NEXT: Cost Model: Found an estimated cost of 84 for instruction: %splice.v16i16 = call <16 x i16> @llvm.experimental.vector.splice.v16i16(<16 x i16> zeroinitializer, <16 x i16> zeroinitializer, i32 1)
12-
; CHECK-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %splice.v4i32 = call <4 x i32> @llvm.experimental.vector.splice.v4i32(<4 x i32> zeroinitializer, <4 x i32> zeroinitializer, i32 1)
13-
; CHECK-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %splice.v8i32 = call <8 x i32> @llvm.experimental.vector.splice.v8i32(<8 x i32> zeroinitializer, <8 x i32> zeroinitializer, i32 1)
14-
; CHECK-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %splice.v2i64 = call <2 x i64> @llvm.experimental.vector.splice.v2i64(<2 x i64> zeroinitializer, <2 x i64> zeroinitializer, i32 1)
15-
; CHECK-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %splice.v4i64 = call <4 x i64> @llvm.experimental.vector.splice.v4i64(<4 x i64> zeroinitializer, <4 x i64> zeroinitializer, i32 1)
16-
; CHECK-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %splice.v2f16 = call <2 x half> @llvm.experimental.vector.splice.v2f16(<2 x half> zeroinitializer, <2 x half> zeroinitializer, i32 1)
17-
; CHECK-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %splice.v4f16 = call <4 x half> @llvm.experimental.vector.splice.v4f16(<4 x half> zeroinitializer, <4 x half> zeroinitializer, i32 1)
18-
; CHECK-NEXT: Cost Model: Found an estimated cost of 42 for instruction: %splice.v8f16 = call <8 x half> @llvm.experimental.vector.splice.v8f16(<8 x half> zeroinitializer, <8 x half> zeroinitializer, i32 1)
19-
; CHECK-NEXT: Cost Model: Found an estimated cost of 84 for instruction: %splice.v16f16 = call <16 x half> @llvm.experimental.vector.splice.v16f16(<16 x half> zeroinitializer, <16 x half> zeroinitializer, i32 1)
20-
; CHECK-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %splice.v2f32 = call <2 x float> @llvm.experimental.vector.splice.v2f32(<2 x float> zeroinitializer, <2 x float> zeroinitializer, i32 1)
21-
; CHECK-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %splice.v4f32 = call <4 x float> @llvm.experimental.vector.splice.v4f32(<4 x float> zeroinitializer, <4 x float> zeroinitializer, i32 1)
22-
; CHECK-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %splice.v8f32 = call <8 x float> @llvm.experimental.vector.splice.v8f32(<8 x float> zeroinitializer, <8 x float> zeroinitializer, i32 1)
23-
; CHECK-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %splice.v2f64 = call <2 x double> @llvm.experimental.vector.splice.v2f64(<2 x double> zeroinitializer, <2 x double> zeroinitializer, i32 1)
24-
; CHECK-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %splice.v4f64 = call <4 x double> @llvm.experimental.vector.splice.v4f64(<4 x double> zeroinitializer, <4 x double> zeroinitializer, i32 1)
25-
; CHECK-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %splice.v2bf16 = call <2 x bfloat> @llvm.experimental.vector.splice.v2bf16(<2 x bfloat> zeroinitializer, <2 x bfloat> zeroinitializer, i32 1)
26-
; CHECK-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %splice.v4bf16 = call <4 x bfloat> @llvm.experimental.vector.splice.v4bf16(<4 x bfloat> zeroinitializer, <4 x bfloat> zeroinitializer, i32 1)
27-
; CHECK-NEXT: Cost Model: Found an estimated cost of 42 for instruction: %splice.v8bf16 = call <8 x bfloat> @llvm.experimental.vector.splice.v8bf16(<8 x bfloat> zeroinitializer, <8 x bfloat> zeroinitializer, i32 1)
28-
; CHECK-NEXT: Cost Model: Found an estimated cost of 84 for instruction: %splice.v16bf16 = call <16 x bfloat> @llvm.experimental.vector.splice.v16bf16(<16 x bfloat> zeroinitializer, <16 x bfloat> zeroinitializer, i32 1)
29-
; CHECK-NEXT: Cost Model: Found an estimated cost of 90 for instruction: %splice.v16i1 = call <16 x i1> @llvm.experimental.vector.splice.v16i1(<16 x i1> zeroinitializer, <16 x i1> zeroinitializer, i32 1)
30-
; CHECK-NEXT: Cost Model: Found an estimated cost of 42 for instruction: %splice.v8i1 = call <8 x i1> @llvm.experimental.vector.splice.v8i1(<8 x i1> zeroinitializer, <8 x i1> zeroinitializer, i32 1)
31-
; CHECK-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %splice.v4i1 = call <4 x i1> @llvm.experimental.vector.splice.v4i1(<4 x i1> zeroinitializer, <4 x i1> zeroinitializer, i32 1)
32-
; CHECK-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %splice.v2i1 = call <2 x i1> @llvm.experimental.vector.splice.v2i1(<2 x i1> zeroinitializer, <2 x i1> zeroinitializer, i32 1)
6+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splice.v16i8 = call <16 x i8> @llvm.experimental.vector.splice.v16i8(<16 x i8> zeroinitializer, <16 x i8> zeroinitializer, i32 1)
7+
; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %splice.v32i8 = call <32 x i8> @llvm.experimental.vector.splice.v32i8(<32 x i8> zeroinitializer, <32 x i8> zeroinitializer, i32 1)
8+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splice.v2i16 = call <2 x i16> @llvm.experimental.vector.splice.v2i16(<2 x i16> zeroinitializer, <2 x i16> zeroinitializer, i32 1)
9+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splice.v4i16 = call <4 x i16> @llvm.experimental.vector.splice.v4i16(<4 x i16> zeroinitializer, <4 x i16> zeroinitializer, i32 1)
10+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splice.v8i16 = call <8 x i16> @llvm.experimental.vector.splice.v8i16(<8 x i16> zeroinitializer, <8 x i16> zeroinitializer, i32 1)
11+
; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %splice.v16i16 = call <16 x i16> @llvm.experimental.vector.splice.v16i16(<16 x i16> zeroinitializer, <16 x i16> zeroinitializer, i32 1)
12+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splice.v4i32 = call <4 x i32> @llvm.experimental.vector.splice.v4i32(<4 x i32> zeroinitializer, <4 x i32> zeroinitializer, i32 1)
13+
; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %splice.v8i32 = call <8 x i32> @llvm.experimental.vector.splice.v8i32(<8 x i32> zeroinitializer, <8 x i32> zeroinitializer, i32 1)
14+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splice.v2i64 = call <2 x i64> @llvm.experimental.vector.splice.v2i64(<2 x i64> zeroinitializer, <2 x i64> zeroinitializer, i32 1)
15+
; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %splice.v4i64 = call <4 x i64> @llvm.experimental.vector.splice.v4i64(<4 x i64> zeroinitializer, <4 x i64> zeroinitializer, i32 1)
16+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splice.v2f16 = call <2 x half> @llvm.experimental.vector.splice.v2f16(<2 x half> zeroinitializer, <2 x half> zeroinitializer, i32 1)
17+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splice.v4f16 = call <4 x half> @llvm.experimental.vector.splice.v4f16(<4 x half> zeroinitializer, <4 x half> zeroinitializer, i32 1)
18+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splice.v8f16 = call <8 x half> @llvm.experimental.vector.splice.v8f16(<8 x half> zeroinitializer, <8 x half> zeroinitializer, i32 1)
19+
; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %splice.v16f16 = call <16 x half> @llvm.experimental.vector.splice.v16f16(<16 x half> zeroinitializer, <16 x half> zeroinitializer, i32 1)
20+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splice.v2f32 = call <2 x float> @llvm.experimental.vector.splice.v2f32(<2 x float> zeroinitializer, <2 x float> zeroinitializer, i32 1)
21+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splice.v4f32 = call <4 x float> @llvm.experimental.vector.splice.v4f32(<4 x float> zeroinitializer, <4 x float> zeroinitializer, i32 1)
22+
; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %splice.v8f32 = call <8 x float> @llvm.experimental.vector.splice.v8f32(<8 x float> zeroinitializer, <8 x float> zeroinitializer, i32 1)
23+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splice.v2f64 = call <2 x double> @llvm.experimental.vector.splice.v2f64(<2 x double> zeroinitializer, <2 x double> zeroinitializer, i32 1)
24+
; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %splice.v4f64 = call <4 x double> @llvm.experimental.vector.splice.v4f64(<4 x double> zeroinitializer, <4 x double> zeroinitializer, i32 1)
25+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splice.v2bf16 = call <2 x bfloat> @llvm.experimental.vector.splice.v2bf16(<2 x bfloat> zeroinitializer, <2 x bfloat> zeroinitializer, i32 1)
26+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splice.v4bf16 = call <4 x bfloat> @llvm.experimental.vector.splice.v4bf16(<4 x bfloat> zeroinitializer, <4 x bfloat> zeroinitializer, i32 1)
27+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splice.v8bf16 = call <8 x bfloat> @llvm.experimental.vector.splice.v8bf16(<8 x bfloat> zeroinitializer, <8 x bfloat> zeroinitializer, i32 1)
28+
; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %splice.v16bf16 = call <16 x bfloat> @llvm.experimental.vector.splice.v16bf16(<16 x bfloat> zeroinitializer, <16 x bfloat> zeroinitializer, i32 1)
29+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splice.v16i1 = call <16 x i1> @llvm.experimental.vector.splice.v16i1(<16 x i1> zeroinitializer, <16 x i1> zeroinitializer, i32 1)
30+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splice.v8i1 = call <8 x i1> @llvm.experimental.vector.splice.v8i1(<8 x i1> zeroinitializer, <8 x i1> zeroinitializer, i32 1)
31+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splice.v4i1 = call <4 x i1> @llvm.experimental.vector.splice.v4i1(<4 x i1> zeroinitializer, <4 x i1> zeroinitializer, i32 1)
32+
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %splice.v2i1 = call <2 x i1> @llvm.experimental.vector.splice.v2i1(<2 x i1> zeroinitializer, <2 x i1> zeroinitializer, i32 1)
33+
; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %splice.v2i128 = call <2 x i128> @llvm.experimental.vector.splice.v2i128(<2 x i128> zeroinitializer, <2 x i128> zeroinitializer, i32 1)
34+
; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
3335
;
3436
%splice.v16i8 = call <16 x i8> @llvm.experimental.vector.splice.v16i8(<16 x i8> zeroinitializer, <16 x i8> zeroinitializer, i32 1)
3537
%splice.v32i8 = call <32 x i8> @llvm.experimental.vector.splice.v32i8(<32 x i8> zeroinitializer, <32 x i8> zeroinitializer, i32 1)
@@ -58,6 +60,7 @@ define void @vector_splice() #0 {
5860
%splice.v8i1 = call <8 x i1> @llvm.experimental.vector.splice.v8i1(<8 x i1> zeroinitializer, <8 x i1> zeroinitializer, i32 1)
5961
%splice.v4i1 = call <4 x i1> @llvm.experimental.vector.splice.v4i1(<4 x i1> zeroinitializer, <4 x i1> zeroinitializer, i32 1)
6062
%splice.v2i1 = call <2 x i1> @llvm.experimental.vector.splice.v2i1(<2 x i1> zeroinitializer, <2 x i1> zeroinitializer, i32 1)
63+
%splice.v2i128 = call <2 x i128> @llvm.experimental.vector.splice.v2i128(<2 x i128> zeroinitializer, <2 x i128> zeroinitializer, i32 1)
6164
ret void
6265
}
6366

@@ -90,5 +93,6 @@ declare <8 x float> @llvm.experimental.vector.splice.v8f32(<8 x float>, <8 x flo
9093
declare <16 x float> @llvm.experimental.vector.splice.v16f32(<16 x float>, <16 x float>, i32)
9194
declare <2 x double> @llvm.experimental.vector.splice.v2f64(<2 x double>, <2 x double>, i32)
9295
declare <4 x double> @llvm.experimental.vector.splice.v4f64(<4 x double>, <4 x double>, i32)
96+
declare <2 x i128> @llvm.experimental.vector.splice.v2i128(<2 x i128>, <2 x i128>, i32)
9397

9498
attributes #0 = { "target-features"="+bf16" }

0 commit comments

Comments
 (0)