• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
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
4
5
6define <4 x i32> @multiply_2x2(<4 x i32> %a, <4 x i32> %b) {
7; CHECK-LABEL: @multiply_2x2(
8; CHECK-NEXT:  entry:
9; CHECK-NEXT:    [[SPLIT:%.*]] = shufflevector <4 x i32> [[A:%.*]], <4 x i32> undef, <2 x i32> <i32 0, i32 1>
10; CHECK-NEXT:    [[SPLIT1:%.*]] = shufflevector <4 x i32> [[A]], <4 x i32> undef, <2 x i32> <i32 2, i32 3>
11; CHECK-NEXT:    [[SPLIT2:%.*]] = shufflevector <4 x i32> [[B:%.*]], <4 x i32> undef, <2 x i32> <i32 0, i32 1>
12; CHECK-NEXT:    [[SPLIT3:%.*]] = shufflevector <4 x i32> [[B]], <4 x i32> undef, <2 x i32> <i32 2, i32 3>
13; CHECK-NEXT:    [[BLOCK:%.*]] = shufflevector <2 x i32> [[SPLIT]], <2 x i32> undef, <1 x i32> zeroinitializer
14; CHECK-NEXT:    [[TMP0:%.*]] = extractelement <2 x i32> [[SPLIT2]], i64 0
15; CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x i32> undef, i32 [[TMP0]], i32 0
16; CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT]], <1 x i32> undef, <1 x i32> zeroinitializer
17; CHECK-NEXT:    [[TMP1:%.*]] = mul <1 x i32> [[BLOCK]], [[SPLAT_SPLAT]]
18; CHECK-NEXT:    [[BLOCK4:%.*]] = shufflevector <2 x i32> [[SPLIT1]], <2 x i32> undef, <1 x i32> zeroinitializer
19; CHECK-NEXT:    [[TMP2:%.*]] = extractelement <2 x i32> [[SPLIT2]], i64 1
20; CHECK-NEXT:    [[SPLAT_SPLATINSERT5:%.*]] = insertelement <1 x i32> undef, i32 [[TMP2]], i32 0
21; CHECK-NEXT:    [[SPLAT_SPLAT6:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT5]], <1 x i32> undef, <1 x i32> zeroinitializer
22; CHECK-NEXT:    [[TMP3:%.*]] = mul <1 x i32> [[BLOCK4]], [[SPLAT_SPLAT6]]
23; CHECK-NEXT:    [[TMP4:%.*]] = add <1 x i32> [[TMP1]], [[TMP3]]
24; CHECK-NEXT:    [[TMP5:%.*]] = shufflevector <1 x i32> [[TMP4]], <1 x i32> undef, <2 x i32> <i32 0, i32 undef>
25; CHECK-NEXT:    [[TMP6:%.*]] = shufflevector <2 x i32> undef, <2 x i32> [[TMP5]], <2 x i32> <i32 2, i32 1>
26; CHECK-NEXT:    [[BLOCK7:%.*]] = shufflevector <2 x i32> [[SPLIT]], <2 x i32> undef, <1 x i32> <i32 1>
27; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <2 x i32> [[SPLIT2]], i64 0
28; CHECK-NEXT:    [[SPLAT_SPLATINSERT8:%.*]] = insertelement <1 x i32> undef, i32 [[TMP7]], i32 0
29; CHECK-NEXT:    [[SPLAT_SPLAT9:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT8]], <1 x i32> undef, <1 x i32> zeroinitializer
30; CHECK-NEXT:    [[TMP8:%.*]] = mul <1 x i32> [[BLOCK7]], [[SPLAT_SPLAT9]]
31; CHECK-NEXT:    [[BLOCK10:%.*]] = shufflevector <2 x i32> [[SPLIT1]], <2 x i32> undef, <1 x i32> <i32 1>
32; CHECK-NEXT:    [[TMP9:%.*]] = extractelement <2 x i32> [[SPLIT2]], i64 1
33; CHECK-NEXT:    [[SPLAT_SPLATINSERT11:%.*]] = insertelement <1 x i32> undef, i32 [[TMP9]], i32 0
34; CHECK-NEXT:    [[SPLAT_SPLAT12:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT11]], <1 x i32> undef, <1 x i32> zeroinitializer
35; CHECK-NEXT:    [[TMP10:%.*]] = mul <1 x i32> [[BLOCK10]], [[SPLAT_SPLAT12]]
36; CHECK-NEXT:    [[TMP11:%.*]] = add <1 x i32> [[TMP8]], [[TMP10]]
37; CHECK-NEXT:    [[TMP12:%.*]] = shufflevector <1 x i32> [[TMP11]], <1 x i32> undef, <2 x i32> <i32 0, i32 undef>
38; CHECK-NEXT:    [[TMP13:%.*]] = shufflevector <2 x i32> [[TMP6]], <2 x i32> [[TMP12]], <2 x i32> <i32 0, i32 2>
39; CHECK-NEXT:    [[BLOCK13:%.*]] = shufflevector <2 x i32> [[SPLIT]], <2 x i32> undef, <1 x i32> zeroinitializer
40; CHECK-NEXT:    [[TMP14:%.*]] = extractelement <2 x i32> [[SPLIT3]], i64 0
41; CHECK-NEXT:    [[SPLAT_SPLATINSERT14:%.*]] = insertelement <1 x i32> undef, i32 [[TMP14]], i32 0
42; CHECK-NEXT:    [[SPLAT_SPLAT15:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT14]], <1 x i32> undef, <1 x i32> zeroinitializer
43; CHECK-NEXT:    [[TMP15:%.*]] = mul <1 x i32> [[BLOCK13]], [[SPLAT_SPLAT15]]
44; CHECK-NEXT:    [[BLOCK16:%.*]] = shufflevector <2 x i32> [[SPLIT1]], <2 x i32> undef, <1 x i32> zeroinitializer
45; CHECK-NEXT:    [[TMP16:%.*]] = extractelement <2 x i32> [[SPLIT3]], i64 1
46; CHECK-NEXT:    [[SPLAT_SPLATINSERT17:%.*]] = insertelement <1 x i32> undef, i32 [[TMP16]], i32 0
47; CHECK-NEXT:    [[SPLAT_SPLAT18:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT17]], <1 x i32> undef, <1 x i32> zeroinitializer
48; CHECK-NEXT:    [[TMP17:%.*]] = mul <1 x i32> [[BLOCK16]], [[SPLAT_SPLAT18]]
49; CHECK-NEXT:    [[TMP18:%.*]] = add <1 x i32> [[TMP15]], [[TMP17]]
50; CHECK-NEXT:    [[TMP19:%.*]] = shufflevector <1 x i32> [[TMP18]], <1 x i32> undef, <2 x i32> <i32 0, i32 undef>
51; CHECK-NEXT:    [[TMP20:%.*]] = shufflevector <2 x i32> undef, <2 x i32> [[TMP19]], <2 x i32> <i32 2, i32 1>
52; CHECK-NEXT:    [[BLOCK19:%.*]] = shufflevector <2 x i32> [[SPLIT]], <2 x i32> undef, <1 x i32> <i32 1>
53; CHECK-NEXT:    [[TMP21:%.*]] = extractelement <2 x i32> [[SPLIT3]], i64 0
54; CHECK-NEXT:    [[SPLAT_SPLATINSERT20:%.*]] = insertelement <1 x i32> undef, i32 [[TMP21]], i32 0
55; CHECK-NEXT:    [[SPLAT_SPLAT21:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT20]], <1 x i32> undef, <1 x i32> zeroinitializer
56; CHECK-NEXT:    [[TMP22:%.*]] = mul <1 x i32> [[BLOCK19]], [[SPLAT_SPLAT21]]
57; CHECK-NEXT:    [[BLOCK22:%.*]] = shufflevector <2 x i32> [[SPLIT1]], <2 x i32> undef, <1 x i32> <i32 1>
58; CHECK-NEXT:    [[TMP23:%.*]] = extractelement <2 x i32> [[SPLIT3]], i64 1
59; CHECK-NEXT:    [[SPLAT_SPLATINSERT23:%.*]] = insertelement <1 x i32> undef, i32 [[TMP23]], i32 0
60; CHECK-NEXT:    [[SPLAT_SPLAT24:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT23]], <1 x i32> undef, <1 x i32> zeroinitializer
61; CHECK-NEXT:    [[TMP24:%.*]] = mul <1 x i32> [[BLOCK22]], [[SPLAT_SPLAT24]]
62; CHECK-NEXT:    [[TMP25:%.*]] = add <1 x i32> [[TMP22]], [[TMP24]]
63; CHECK-NEXT:    [[TMP26:%.*]] = shufflevector <1 x i32> [[TMP25]], <1 x i32> undef, <2 x i32> <i32 0, i32 undef>
64; CHECK-NEXT:    [[TMP27:%.*]] = shufflevector <2 x i32> [[TMP20]], <2 x i32> [[TMP26]], <2 x i32> <i32 0, i32 2>
65; CHECK-NEXT:    [[TMP28:%.*]] = shufflevector <2 x i32> [[TMP13]], <2 x i32> [[TMP27]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
66; CHECK-NEXT:    ret <4 x i32> [[TMP28]]
67;
68entry:
69  %c = call <4 x i32> @llvm.matrix.multiply.v4i32.v4i32.v4i32(<4 x i32> %a, <4 x i32> %b, i32 2, i32 2, i32 2)
70  ret <4 x i32> %c
71}
72
73declare <4 x i32> @llvm.matrix.multiply.v4i32.v4i32.v4i32(<4 x i32>, <4 x i32>, i32, i32, i32)
74
75define <4 x i32> @multiply_1x2(<2 x i32> %a, <2 x i32> %b) {
76; CHECK-LABEL: @multiply_1x2(
77; CHECK-NEXT:  entry:
78; CHECK-NEXT:    [[SPLIT:%.*]] = shufflevector <2 x i32> [[A:%.*]], <2 x i32> undef, <2 x i32> <i32 0, i32 1>
79; CHECK-NEXT:    [[SPLIT1:%.*]] = shufflevector <2 x i32> [[B:%.*]], <2 x i32> undef, <1 x i32> zeroinitializer
80; CHECK-NEXT:    [[SPLIT2:%.*]] = shufflevector <2 x i32> [[B]], <2 x i32> undef, <1 x i32> <i32 1>
81; CHECK-NEXT:    [[BLOCK:%.*]] = shufflevector <2 x i32> [[SPLIT]], <2 x i32> undef, <1 x i32> zeroinitializer
82; CHECK-NEXT:    [[TMP0:%.*]] = extractelement <1 x i32> [[SPLIT1]], i64 0
83; CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x i32> undef, i32 [[TMP0]], i32 0
84; CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT]], <1 x i32> undef, <1 x i32> zeroinitializer
85; CHECK-NEXT:    [[TMP1:%.*]] = mul <1 x i32> [[BLOCK]], [[SPLAT_SPLAT]]
86; CHECK-NEXT:    [[TMP2:%.*]] = shufflevector <1 x i32> [[TMP1]], <1 x i32> undef, <2 x i32> <i32 0, i32 undef>
87; CHECK-NEXT:    [[TMP3:%.*]] = shufflevector <2 x i32> undef, <2 x i32> [[TMP2]], <2 x i32> <i32 2, i32 1>
88; CHECK-NEXT:    [[BLOCK3:%.*]] = shufflevector <2 x i32> [[SPLIT]], <2 x i32> undef, <1 x i32> <i32 1>
89; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <1 x i32> [[SPLIT1]], i64 0
90; CHECK-NEXT:    [[SPLAT_SPLATINSERT4:%.*]] = insertelement <1 x i32> undef, i32 [[TMP4]], i32 0
91; CHECK-NEXT:    [[SPLAT_SPLAT5:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT4]], <1 x i32> undef, <1 x i32> zeroinitializer
92; CHECK-NEXT:    [[TMP5:%.*]] = mul <1 x i32> [[BLOCK3]], [[SPLAT_SPLAT5]]
93; CHECK-NEXT:    [[TMP6:%.*]] = shufflevector <1 x i32> [[TMP5]], <1 x i32> undef, <2 x i32> <i32 0, i32 undef>
94; CHECK-NEXT:    [[TMP7:%.*]] = shufflevector <2 x i32> [[TMP3]], <2 x i32> [[TMP6]], <2 x i32> <i32 0, i32 2>
95; CHECK-NEXT:    [[BLOCK6:%.*]] = shufflevector <2 x i32> [[SPLIT]], <2 x i32> undef, <1 x i32> zeroinitializer
96; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <1 x i32> [[SPLIT2]], i64 0
97; CHECK-NEXT:    [[SPLAT_SPLATINSERT7:%.*]] = insertelement <1 x i32> undef, i32 [[TMP8]], i32 0
98; CHECK-NEXT:    [[SPLAT_SPLAT8:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT7]], <1 x i32> undef, <1 x i32> zeroinitializer
99; CHECK-NEXT:    [[TMP9:%.*]] = mul <1 x i32> [[BLOCK6]], [[SPLAT_SPLAT8]]
100; CHECK-NEXT:    [[TMP10:%.*]] = shufflevector <1 x i32> [[TMP9]], <1 x i32> undef, <2 x i32> <i32 0, i32 undef>
101; CHECK-NEXT:    [[TMP11:%.*]] = shufflevector <2 x i32> undef, <2 x i32> [[TMP10]], <2 x i32> <i32 2, i32 1>
102; CHECK-NEXT:    [[BLOCK9:%.*]] = shufflevector <2 x i32> [[SPLIT]], <2 x i32> undef, <1 x i32> <i32 1>
103; CHECK-NEXT:    [[TMP12:%.*]] = extractelement <1 x i32> [[SPLIT2]], i64 0
104; CHECK-NEXT:    [[SPLAT_SPLATINSERT10:%.*]] = insertelement <1 x i32> undef, i32 [[TMP12]], i32 0
105; CHECK-NEXT:    [[SPLAT_SPLAT11:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT10]], <1 x i32> undef, <1 x i32> zeroinitializer
106; CHECK-NEXT:    [[TMP13:%.*]] = mul <1 x i32> [[BLOCK9]], [[SPLAT_SPLAT11]]
107; CHECK-NEXT:    [[TMP14:%.*]] = shufflevector <1 x i32> [[TMP13]], <1 x i32> undef, <2 x i32> <i32 0, i32 undef>
108; CHECK-NEXT:    [[TMP15:%.*]] = shufflevector <2 x i32> [[TMP11]], <2 x i32> [[TMP14]], <2 x i32> <i32 0, i32 2>
109; CHECK-NEXT:    [[TMP16:%.*]] = shufflevector <2 x i32> [[TMP7]], <2 x i32> [[TMP15]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
110; CHECK-NEXT:    ret <4 x i32> [[TMP16]]
111;
112entry:
113  %c = call <4 x i32> @llvm.matrix.multiply.v4i32.v2i32.v2i32(<2 x i32> %a, <2 x i32> %b, i32 2, i32 1, i32 2)
114  ret <4 x i32> %c
115}
116
117declare <4 x i32> @llvm.matrix.multiply.v4i32.v2i32.v2i32(<2 x i32>, <2 x i32>, i32, i32, i32)
118
119define <9 x i32> @multiply_2x3(<6 x i32> %a, <6 x i32> %b) {
120; CHECK-LABEL: @multiply_2x3(
121; CHECK-NEXT:  entry:
122; CHECK-NEXT:    [[SPLIT:%.*]] = shufflevector <6 x i32> [[A:%.*]], <6 x i32> undef, <3 x i32> <i32 0, i32 1, i32 2>
123; CHECK-NEXT:    [[SPLIT1:%.*]] = shufflevector <6 x i32> [[A]], <6 x i32> undef, <3 x i32> <i32 3, i32 4, i32 5>
124; CHECK-NEXT:    [[SPLIT2:%.*]] = shufflevector <6 x i32> [[B:%.*]], <6 x i32> undef, <2 x i32> <i32 0, i32 1>
125; CHECK-NEXT:    [[SPLIT3:%.*]] = shufflevector <6 x i32> [[B]], <6 x i32> undef, <2 x i32> <i32 2, i32 3>
126; CHECK-NEXT:    [[SPLIT4:%.*]] = shufflevector <6 x i32> [[B]], <6 x i32> undef, <2 x i32> <i32 4, i32 5>
127; CHECK-NEXT:    [[BLOCK:%.*]] = shufflevector <3 x i32> [[SPLIT]], <3 x i32> undef, <1 x i32> zeroinitializer
128; CHECK-NEXT:    [[TMP0:%.*]] = extractelement <2 x i32> [[SPLIT2]], i64 0
129; CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x i32> undef, i32 [[TMP0]], i32 0
130; CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT]], <1 x i32> undef, <1 x i32> zeroinitializer
131; CHECK-NEXT:    [[TMP1:%.*]] = mul <1 x i32> [[BLOCK]], [[SPLAT_SPLAT]]
132; CHECK-NEXT:    [[BLOCK5:%.*]] = shufflevector <3 x i32> [[SPLIT1]], <3 x i32> undef, <1 x i32> zeroinitializer
133; CHECK-NEXT:    [[TMP2:%.*]] = extractelement <2 x i32> [[SPLIT2]], i64 1
134; CHECK-NEXT:    [[SPLAT_SPLATINSERT6:%.*]] = insertelement <1 x i32> undef, i32 [[TMP2]], i32 0
135; CHECK-NEXT:    [[SPLAT_SPLAT7:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT6]], <1 x i32> undef, <1 x i32> zeroinitializer
136; CHECK-NEXT:    [[TMP3:%.*]] = mul <1 x i32> [[BLOCK5]], [[SPLAT_SPLAT7]]
137; CHECK-NEXT:    [[TMP4:%.*]] = add <1 x i32> [[TMP1]], [[TMP3]]
138; CHECK-NEXT:    [[TMP5:%.*]] = shufflevector <1 x i32> [[TMP4]], <1 x i32> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
139; CHECK-NEXT:    [[TMP6:%.*]] = shufflevector <3 x i32> undef, <3 x i32> [[TMP5]], <3 x i32> <i32 3, i32 1, i32 2>
140; CHECK-NEXT:    [[BLOCK8:%.*]] = shufflevector <3 x i32> [[SPLIT]], <3 x i32> undef, <1 x i32> <i32 1>
141; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <2 x i32> [[SPLIT2]], i64 0
142; CHECK-NEXT:    [[SPLAT_SPLATINSERT9:%.*]] = insertelement <1 x i32> undef, i32 [[TMP7]], i32 0
143; CHECK-NEXT:    [[SPLAT_SPLAT10:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT9]], <1 x i32> undef, <1 x i32> zeroinitializer
144; CHECK-NEXT:    [[TMP8:%.*]] = mul <1 x i32> [[BLOCK8]], [[SPLAT_SPLAT10]]
145; CHECK-NEXT:    [[BLOCK11:%.*]] = shufflevector <3 x i32> [[SPLIT1]], <3 x i32> undef, <1 x i32> <i32 1>
146; CHECK-NEXT:    [[TMP9:%.*]] = extractelement <2 x i32> [[SPLIT2]], i64 1
147; CHECK-NEXT:    [[SPLAT_SPLATINSERT12:%.*]] = insertelement <1 x i32> undef, i32 [[TMP9]], i32 0
148; CHECK-NEXT:    [[SPLAT_SPLAT13:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT12]], <1 x i32> undef, <1 x i32> zeroinitializer
149; CHECK-NEXT:    [[TMP10:%.*]] = mul <1 x i32> [[BLOCK11]], [[SPLAT_SPLAT13]]
150; CHECK-NEXT:    [[TMP11:%.*]] = add <1 x i32> [[TMP8]], [[TMP10]]
151; CHECK-NEXT:    [[TMP12:%.*]] = shufflevector <1 x i32> [[TMP11]], <1 x i32> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
152; CHECK-NEXT:    [[TMP13:%.*]] = shufflevector <3 x i32> [[TMP6]], <3 x i32> [[TMP12]], <3 x i32> <i32 0, i32 3, i32 2>
153; CHECK-NEXT:    [[BLOCK14:%.*]] = shufflevector <3 x i32> [[SPLIT]], <3 x i32> undef, <1 x i32> <i32 2>
154; CHECK-NEXT:    [[TMP14:%.*]] = extractelement <2 x i32> [[SPLIT2]], i64 0
155; CHECK-NEXT:    [[SPLAT_SPLATINSERT15:%.*]] = insertelement <1 x i32> undef, i32 [[TMP14]], i32 0
156; CHECK-NEXT:    [[SPLAT_SPLAT16:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT15]], <1 x i32> undef, <1 x i32> zeroinitializer
157; CHECK-NEXT:    [[TMP15:%.*]] = mul <1 x i32> [[BLOCK14]], [[SPLAT_SPLAT16]]
158; CHECK-NEXT:    [[BLOCK17:%.*]] = shufflevector <3 x i32> [[SPLIT1]], <3 x i32> undef, <1 x i32> <i32 2>
159; CHECK-NEXT:    [[TMP16:%.*]] = extractelement <2 x i32> [[SPLIT2]], i64 1
160; CHECK-NEXT:    [[SPLAT_SPLATINSERT18:%.*]] = insertelement <1 x i32> undef, i32 [[TMP16]], i32 0
161; CHECK-NEXT:    [[SPLAT_SPLAT19:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT18]], <1 x i32> undef, <1 x i32> zeroinitializer
162; CHECK-NEXT:    [[TMP17:%.*]] = mul <1 x i32> [[BLOCK17]], [[SPLAT_SPLAT19]]
163; CHECK-NEXT:    [[TMP18:%.*]] = add <1 x i32> [[TMP15]], [[TMP17]]
164; CHECK-NEXT:    [[TMP19:%.*]] = shufflevector <1 x i32> [[TMP18]], <1 x i32> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
165; CHECK-NEXT:    [[TMP20:%.*]] = shufflevector <3 x i32> [[TMP13]], <3 x i32> [[TMP19]], <3 x i32> <i32 0, i32 1, i32 3>
166; CHECK-NEXT:    [[BLOCK20:%.*]] = shufflevector <3 x i32> [[SPLIT]], <3 x i32> undef, <1 x i32> zeroinitializer
167; CHECK-NEXT:    [[TMP21:%.*]] = extractelement <2 x i32> [[SPLIT3]], i64 0
168; CHECK-NEXT:    [[SPLAT_SPLATINSERT21:%.*]] = insertelement <1 x i32> undef, i32 [[TMP21]], i32 0
169; CHECK-NEXT:    [[SPLAT_SPLAT22:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT21]], <1 x i32> undef, <1 x i32> zeroinitializer
170; CHECK-NEXT:    [[TMP22:%.*]] = mul <1 x i32> [[BLOCK20]], [[SPLAT_SPLAT22]]
171; CHECK-NEXT:    [[BLOCK23:%.*]] = shufflevector <3 x i32> [[SPLIT1]], <3 x i32> undef, <1 x i32> zeroinitializer
172; CHECK-NEXT:    [[TMP23:%.*]] = extractelement <2 x i32> [[SPLIT3]], i64 1
173; CHECK-NEXT:    [[SPLAT_SPLATINSERT24:%.*]] = insertelement <1 x i32> undef, i32 [[TMP23]], i32 0
174; CHECK-NEXT:    [[SPLAT_SPLAT25:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT24]], <1 x i32> undef, <1 x i32> zeroinitializer
175; CHECK-NEXT:    [[TMP24:%.*]] = mul <1 x i32> [[BLOCK23]], [[SPLAT_SPLAT25]]
176; CHECK-NEXT:    [[TMP25:%.*]] = add <1 x i32> [[TMP22]], [[TMP24]]
177; CHECK-NEXT:    [[TMP26:%.*]] = shufflevector <1 x i32> [[TMP25]], <1 x i32> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
178; CHECK-NEXT:    [[TMP27:%.*]] = shufflevector <3 x i32> undef, <3 x i32> [[TMP26]], <3 x i32> <i32 3, i32 1, i32 2>
179; CHECK-NEXT:    [[BLOCK26:%.*]] = shufflevector <3 x i32> [[SPLIT]], <3 x i32> undef, <1 x i32> <i32 1>
180; CHECK-NEXT:    [[TMP28:%.*]] = extractelement <2 x i32> [[SPLIT3]], i64 0
181; CHECK-NEXT:    [[SPLAT_SPLATINSERT27:%.*]] = insertelement <1 x i32> undef, i32 [[TMP28]], i32 0
182; CHECK-NEXT:    [[SPLAT_SPLAT28:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT27]], <1 x i32> undef, <1 x i32> zeroinitializer
183; CHECK-NEXT:    [[TMP29:%.*]] = mul <1 x i32> [[BLOCK26]], [[SPLAT_SPLAT28]]
184; CHECK-NEXT:    [[BLOCK29:%.*]] = shufflevector <3 x i32> [[SPLIT1]], <3 x i32> undef, <1 x i32> <i32 1>
185; CHECK-NEXT:    [[TMP30:%.*]] = extractelement <2 x i32> [[SPLIT3]], i64 1
186; CHECK-NEXT:    [[SPLAT_SPLATINSERT30:%.*]] = insertelement <1 x i32> undef, i32 [[TMP30]], i32 0
187; CHECK-NEXT:    [[SPLAT_SPLAT31:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT30]], <1 x i32> undef, <1 x i32> zeroinitializer
188; CHECK-NEXT:    [[TMP31:%.*]] = mul <1 x i32> [[BLOCK29]], [[SPLAT_SPLAT31]]
189; CHECK-NEXT:    [[TMP32:%.*]] = add <1 x i32> [[TMP29]], [[TMP31]]
190; CHECK-NEXT:    [[TMP33:%.*]] = shufflevector <1 x i32> [[TMP32]], <1 x i32> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
191; CHECK-NEXT:    [[TMP34:%.*]] = shufflevector <3 x i32> [[TMP27]], <3 x i32> [[TMP33]], <3 x i32> <i32 0, i32 3, i32 2>
192; CHECK-NEXT:    [[BLOCK32:%.*]] = shufflevector <3 x i32> [[SPLIT]], <3 x i32> undef, <1 x i32> <i32 2>
193; CHECK-NEXT:    [[TMP35:%.*]] = extractelement <2 x i32> [[SPLIT3]], i64 0
194; CHECK-NEXT:    [[SPLAT_SPLATINSERT33:%.*]] = insertelement <1 x i32> undef, i32 [[TMP35]], i32 0
195; CHECK-NEXT:    [[SPLAT_SPLAT34:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT33]], <1 x i32> undef, <1 x i32> zeroinitializer
196; CHECK-NEXT:    [[TMP36:%.*]] = mul <1 x i32> [[BLOCK32]], [[SPLAT_SPLAT34]]
197; CHECK-NEXT:    [[BLOCK35:%.*]] = shufflevector <3 x i32> [[SPLIT1]], <3 x i32> undef, <1 x i32> <i32 2>
198; CHECK-NEXT:    [[TMP37:%.*]] = extractelement <2 x i32> [[SPLIT3]], i64 1
199; CHECK-NEXT:    [[SPLAT_SPLATINSERT36:%.*]] = insertelement <1 x i32> undef, i32 [[TMP37]], i32 0
200; CHECK-NEXT:    [[SPLAT_SPLAT37:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT36]], <1 x i32> undef, <1 x i32> zeroinitializer
201; CHECK-NEXT:    [[TMP38:%.*]] = mul <1 x i32> [[BLOCK35]], [[SPLAT_SPLAT37]]
202; CHECK-NEXT:    [[TMP39:%.*]] = add <1 x i32> [[TMP36]], [[TMP38]]
203; CHECK-NEXT:    [[TMP40:%.*]] = shufflevector <1 x i32> [[TMP39]], <1 x i32> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
204; CHECK-NEXT:    [[TMP41:%.*]] = shufflevector <3 x i32> [[TMP34]], <3 x i32> [[TMP40]], <3 x i32> <i32 0, i32 1, i32 3>
205; CHECK-NEXT:    [[BLOCK38:%.*]] = shufflevector <3 x i32> [[SPLIT]], <3 x i32> undef, <1 x i32> zeroinitializer
206; CHECK-NEXT:    [[TMP42:%.*]] = extractelement <2 x i32> [[SPLIT4]], i64 0
207; CHECK-NEXT:    [[SPLAT_SPLATINSERT39:%.*]] = insertelement <1 x i32> undef, i32 [[TMP42]], i32 0
208; CHECK-NEXT:    [[SPLAT_SPLAT40:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT39]], <1 x i32> undef, <1 x i32> zeroinitializer
209; CHECK-NEXT:    [[TMP43:%.*]] = mul <1 x i32> [[BLOCK38]], [[SPLAT_SPLAT40]]
210; CHECK-NEXT:    [[BLOCK41:%.*]] = shufflevector <3 x i32> [[SPLIT1]], <3 x i32> undef, <1 x i32> zeroinitializer
211; CHECK-NEXT:    [[TMP44:%.*]] = extractelement <2 x i32> [[SPLIT4]], i64 1
212; CHECK-NEXT:    [[SPLAT_SPLATINSERT42:%.*]] = insertelement <1 x i32> undef, i32 [[TMP44]], i32 0
213; CHECK-NEXT:    [[SPLAT_SPLAT43:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT42]], <1 x i32> undef, <1 x i32> zeroinitializer
214; CHECK-NEXT:    [[TMP45:%.*]] = mul <1 x i32> [[BLOCK41]], [[SPLAT_SPLAT43]]
215; CHECK-NEXT:    [[TMP46:%.*]] = add <1 x i32> [[TMP43]], [[TMP45]]
216; CHECK-NEXT:    [[TMP47:%.*]] = shufflevector <1 x i32> [[TMP46]], <1 x i32> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
217; CHECK-NEXT:    [[TMP48:%.*]] = shufflevector <3 x i32> undef, <3 x i32> [[TMP47]], <3 x i32> <i32 3, i32 1, i32 2>
218; CHECK-NEXT:    [[BLOCK44:%.*]] = shufflevector <3 x i32> [[SPLIT]], <3 x i32> undef, <1 x i32> <i32 1>
219; CHECK-NEXT:    [[TMP49:%.*]] = extractelement <2 x i32> [[SPLIT4]], i64 0
220; CHECK-NEXT:    [[SPLAT_SPLATINSERT45:%.*]] = insertelement <1 x i32> undef, i32 [[TMP49]], i32 0
221; CHECK-NEXT:    [[SPLAT_SPLAT46:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT45]], <1 x i32> undef, <1 x i32> zeroinitializer
222; CHECK-NEXT:    [[TMP50:%.*]] = mul <1 x i32> [[BLOCK44]], [[SPLAT_SPLAT46]]
223; CHECK-NEXT:    [[BLOCK47:%.*]] = shufflevector <3 x i32> [[SPLIT1]], <3 x i32> undef, <1 x i32> <i32 1>
224; CHECK-NEXT:    [[TMP51:%.*]] = extractelement <2 x i32> [[SPLIT4]], i64 1
225; CHECK-NEXT:    [[SPLAT_SPLATINSERT48:%.*]] = insertelement <1 x i32> undef, i32 [[TMP51]], i32 0
226; CHECK-NEXT:    [[SPLAT_SPLAT49:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT48]], <1 x i32> undef, <1 x i32> zeroinitializer
227; CHECK-NEXT:    [[TMP52:%.*]] = mul <1 x i32> [[BLOCK47]], [[SPLAT_SPLAT49]]
228; CHECK-NEXT:    [[TMP53:%.*]] = add <1 x i32> [[TMP50]], [[TMP52]]
229; CHECK-NEXT:    [[TMP54:%.*]] = shufflevector <1 x i32> [[TMP53]], <1 x i32> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
230; CHECK-NEXT:    [[TMP55:%.*]] = shufflevector <3 x i32> [[TMP48]], <3 x i32> [[TMP54]], <3 x i32> <i32 0, i32 3, i32 2>
231; CHECK-NEXT:    [[BLOCK50:%.*]] = shufflevector <3 x i32> [[SPLIT]], <3 x i32> undef, <1 x i32> <i32 2>
232; CHECK-NEXT:    [[TMP56:%.*]] = extractelement <2 x i32> [[SPLIT4]], i64 0
233; CHECK-NEXT:    [[SPLAT_SPLATINSERT51:%.*]] = insertelement <1 x i32> undef, i32 [[TMP56]], i32 0
234; CHECK-NEXT:    [[SPLAT_SPLAT52:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT51]], <1 x i32> undef, <1 x i32> zeroinitializer
235; CHECK-NEXT:    [[TMP57:%.*]] = mul <1 x i32> [[BLOCK50]], [[SPLAT_SPLAT52]]
236; CHECK-NEXT:    [[BLOCK53:%.*]] = shufflevector <3 x i32> [[SPLIT1]], <3 x i32> undef, <1 x i32> <i32 2>
237; CHECK-NEXT:    [[TMP58:%.*]] = extractelement <2 x i32> [[SPLIT4]], i64 1
238; CHECK-NEXT:    [[SPLAT_SPLATINSERT54:%.*]] = insertelement <1 x i32> undef, i32 [[TMP58]], i32 0
239; CHECK-NEXT:    [[SPLAT_SPLAT55:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT54]], <1 x i32> undef, <1 x i32> zeroinitializer
240; CHECK-NEXT:    [[TMP59:%.*]] = mul <1 x i32> [[BLOCK53]], [[SPLAT_SPLAT55]]
241; CHECK-NEXT:    [[TMP60:%.*]] = add <1 x i32> [[TMP57]], [[TMP59]]
242; CHECK-NEXT:    [[TMP61:%.*]] = shufflevector <1 x i32> [[TMP60]], <1 x i32> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
243; CHECK-NEXT:    [[TMP62:%.*]] = shufflevector <3 x i32> [[TMP55]], <3 x i32> [[TMP61]], <3 x i32> <i32 0, i32 1, i32 3>
244; CHECK-NEXT:    [[TMP63:%.*]] = shufflevector <3 x i32> [[TMP20]], <3 x i32> [[TMP41]], <6 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5>
245; CHECK-NEXT:    [[TMP64:%.*]] = shufflevector <3 x i32> [[TMP62]], <3 x i32> undef, <6 x i32> <i32 0, i32 1, i32 2, i32 undef, i32 undef, i32 undef>
246; CHECK-NEXT:    [[TMP65:%.*]] = shufflevector <6 x i32> [[TMP63]], <6 x i32> [[TMP64]], <9 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8>
247; CHECK-NEXT:    ret <9 x i32> [[TMP65]]
248;
249entry:
250  %c = call <9 x i32> @llvm.matrix.multiply.v6i32.v6i32.v6i32(<6 x i32> %a, <6 x i32> %b, i32 3, i32 2, i32 3)
251  ret <9 x i32> %c
252}
253
254declare <9 x i32> @llvm.matrix.multiply.v6i32.v6i32.v6i32(<6 x i32>, <6 x i32>, i32, i32, i32)
255