1 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
2 ; RUN: opt -lower-matrix-intrinsics -S < %s | FileCheck %s
3 ; RUN: opt -passes='lower-matrix-intrinsics' -S < %s | FileCheck %s
6 define <4 x float> @multiply_2x2(<4 x float> %a, <4 x float> %b) {
7 ; CHECK-LABEL: @multiply_2x2(
9 ; CHECK-NEXT: [[SPLIT:%.*]] = shufflevector <4 x float> [[A:%.*]], <4 x float> poison, <2 x i32> <i32 0, i32 1>
10 ; CHECK-NEXT: [[SPLIT1:%.*]] = shufflevector <4 x float> [[A]], <4 x float> poison, <2 x i32> <i32 2, i32 3>
11 ; CHECK-NEXT: [[SPLIT2:%.*]] = shufflevector <4 x float> [[B:%.*]], <4 x float> poison, <2 x i32> <i32 0, i32 1>
12 ; CHECK-NEXT: [[SPLIT3:%.*]] = shufflevector <4 x float> [[B]], <4 x float> poison, <2 x i32> <i32 2, i32 3>
13 ; CHECK-NEXT: [[BLOCK:%.*]] = shufflevector <2 x float> [[SPLIT]], <2 x float> poison, <1 x i32> zeroinitializer
14 ; CHECK-NEXT: [[TMP0:%.*]] = extractelement <2 x float> [[SPLIT2]], i64 0
15 ; CHECK-NEXT: [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x float> poison, float [[TMP0]], i32 0
16 ; CHECK-NEXT: [[SPLAT_SPLAT:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT]], <1 x float> poison, <1 x i32> zeroinitializer
17 ; CHECK-NEXT: [[TMP1:%.*]] = fmul contract <1 x float> [[BLOCK]], [[SPLAT_SPLAT]]
18 ; CHECK-NEXT: [[BLOCK4:%.*]] = shufflevector <2 x float> [[SPLIT1]], <2 x float> poison, <1 x i32> zeroinitializer
19 ; CHECK-NEXT: [[TMP2:%.*]] = extractelement <2 x float> [[SPLIT2]], i64 1
20 ; CHECK-NEXT: [[SPLAT_SPLATINSERT5:%.*]] = insertelement <1 x float> poison, float [[TMP2]], i32 0
21 ; CHECK-NEXT: [[SPLAT_SPLAT6:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT5]], <1 x float> poison, <1 x i32> zeroinitializer
22 ; CHECK-NEXT: [[TMP3:%.*]] = call contract <1 x float> @llvm.fmuladd.v1f32(<1 x float> [[BLOCK4]], <1 x float> [[SPLAT_SPLAT6]], <1 x float> [[TMP1]])
23 ; CHECK-NEXT: [[TMP4:%.*]] = shufflevector <1 x float> [[TMP3]], <1 x float> poison, <2 x i32> <i32 0, i32 undef>
24 ; CHECK-NEXT: [[TMP5:%.*]] = shufflevector <2 x float> undef, <2 x float> [[TMP4]], <2 x i32> <i32 2, i32 1>
25 ; CHECK-NEXT: [[BLOCK7:%.*]] = shufflevector <2 x float> [[SPLIT]], <2 x float> poison, <1 x i32> <i32 1>
26 ; CHECK-NEXT: [[TMP6:%.*]] = extractelement <2 x float> [[SPLIT2]], i64 0
27 ; CHECK-NEXT: [[SPLAT_SPLATINSERT8:%.*]] = insertelement <1 x float> poison, float [[TMP6]], i32 0
28 ; CHECK-NEXT: [[SPLAT_SPLAT9:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT8]], <1 x float> poison, <1 x i32> zeroinitializer
29 ; CHECK-NEXT: [[TMP7:%.*]] = fmul contract <1 x float> [[BLOCK7]], [[SPLAT_SPLAT9]]
30 ; CHECK-NEXT: [[BLOCK10:%.*]] = shufflevector <2 x float> [[SPLIT1]], <2 x float> poison, <1 x i32> <i32 1>
31 ; CHECK-NEXT: [[TMP8:%.*]] = extractelement <2 x float> [[SPLIT2]], i64 1
32 ; CHECK-NEXT: [[SPLAT_SPLATINSERT11:%.*]] = insertelement <1 x float> poison, float [[TMP8]], i32 0
33 ; CHECK-NEXT: [[SPLAT_SPLAT12:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT11]], <1 x float> poison, <1 x i32> zeroinitializer
34 ; CHECK-NEXT: [[TMP9:%.*]] = call contract <1 x float> @llvm.fmuladd.v1f32(<1 x float> [[BLOCK10]], <1 x float> [[SPLAT_SPLAT12]], <1 x float> [[TMP7]])
35 ; CHECK-NEXT: [[TMP10:%.*]] = shufflevector <1 x float> [[TMP9]], <1 x float> poison, <2 x i32> <i32 0, i32 undef>
36 ; CHECK-NEXT: [[TMP11:%.*]] = shufflevector <2 x float> [[TMP5]], <2 x float> [[TMP10]], <2 x i32> <i32 0, i32 2>
37 ; CHECK-NEXT: [[BLOCK13:%.*]] = shufflevector <2 x float> [[SPLIT]], <2 x float> poison, <1 x i32> zeroinitializer
38 ; CHECK-NEXT: [[TMP12:%.*]] = extractelement <2 x float> [[SPLIT3]], i64 0
39 ; CHECK-NEXT: [[SPLAT_SPLATINSERT14:%.*]] = insertelement <1 x float> poison, float [[TMP12]], i32 0
40 ; CHECK-NEXT: [[SPLAT_SPLAT15:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT14]], <1 x float> poison, <1 x i32> zeroinitializer
41 ; CHECK-NEXT: [[TMP13:%.*]] = fmul contract <1 x float> [[BLOCK13]], [[SPLAT_SPLAT15]]
42 ; CHECK-NEXT: [[BLOCK16:%.*]] = shufflevector <2 x float> [[SPLIT1]], <2 x float> poison, <1 x i32> zeroinitializer
43 ; CHECK-NEXT: [[TMP14:%.*]] = extractelement <2 x float> [[SPLIT3]], i64 1
44 ; CHECK-NEXT: [[SPLAT_SPLATINSERT17:%.*]] = insertelement <1 x float> poison, float [[TMP14]], i32 0
45 ; CHECK-NEXT: [[SPLAT_SPLAT18:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT17]], <1 x float> poison, <1 x i32> zeroinitializer
46 ; CHECK-NEXT: [[TMP15:%.*]] = call contract <1 x float> @llvm.fmuladd.v1f32(<1 x float> [[BLOCK16]], <1 x float> [[SPLAT_SPLAT18]], <1 x float> [[TMP13]])
47 ; CHECK-NEXT: [[TMP16:%.*]] = shufflevector <1 x float> [[TMP15]], <1 x float> poison, <2 x i32> <i32 0, i32 undef>
48 ; CHECK-NEXT: [[TMP17:%.*]] = shufflevector <2 x float> undef, <2 x float> [[TMP16]], <2 x i32> <i32 2, i32 1>
49 ; CHECK-NEXT: [[BLOCK19:%.*]] = shufflevector <2 x float> [[SPLIT]], <2 x float> poison, <1 x i32> <i32 1>
50 ; CHECK-NEXT: [[TMP18:%.*]] = extractelement <2 x float> [[SPLIT3]], i64 0
51 ; CHECK-NEXT: [[SPLAT_SPLATINSERT20:%.*]] = insertelement <1 x float> poison, float [[TMP18]], i32 0
52 ; CHECK-NEXT: [[SPLAT_SPLAT21:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT20]], <1 x float> poison, <1 x i32> zeroinitializer
53 ; CHECK-NEXT: [[TMP19:%.*]] = fmul contract <1 x float> [[BLOCK19]], [[SPLAT_SPLAT21]]
54 ; CHECK-NEXT: [[BLOCK22:%.*]] = shufflevector <2 x float> [[SPLIT1]], <2 x float> poison, <1 x i32> <i32 1>
55 ; CHECK-NEXT: [[TMP20:%.*]] = extractelement <2 x float> [[SPLIT3]], i64 1
56 ; CHECK-NEXT: [[SPLAT_SPLATINSERT23:%.*]] = insertelement <1 x float> poison, float [[TMP20]], i32 0
57 ; CHECK-NEXT: [[SPLAT_SPLAT24:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT23]], <1 x float> poison, <1 x i32> zeroinitializer
58 ; CHECK-NEXT: [[TMP21:%.*]] = call contract <1 x float> @llvm.fmuladd.v1f32(<1 x float> [[BLOCK22]], <1 x float> [[SPLAT_SPLAT24]], <1 x float> [[TMP19]])
59 ; CHECK-NEXT: [[TMP22:%.*]] = shufflevector <1 x float> [[TMP21]], <1 x float> poison, <2 x i32> <i32 0, i32 undef>
60 ; CHECK-NEXT: [[TMP23:%.*]] = shufflevector <2 x float> [[TMP17]], <2 x float> [[TMP22]], <2 x i32> <i32 0, i32 2>
61 ; CHECK-NEXT: [[TMP24:%.*]] = shufflevector <2 x float> [[TMP11]], <2 x float> [[TMP23]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
62 ; CHECK-NEXT: ret <4 x float> [[TMP24]]
65 %c = call contract <4 x float> @llvm.matrix.multiply.v4f32.v4f32.v4f32(<4 x float> %a, <4 x float> %b, i32 2, i32 2, i32 2)
69 declare <4 x float> @llvm.matrix.multiply.v4f32.v4f32.v4f32(<4 x float>, <4 x float>, i32, i32, i32)