1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
|
; RUN: llc -march=amdgcn -mcpu=gfx90a -verify-machineinstrs < %s | FileCheck --enable-var-scope --check-prefixes=GCN %s
; RUN: llc -global-isel -march=amdgcn -mcpu=gfx90a -verify-machineinstrs < %s | FileCheck --enable-var-scope --check-prefixes=GCN %s
declare <32 x float> @llvm.amdgcn.mfma.f32.32x32x2bf16(<2 x i16>, <2 x i16>, <32 x float>, i32, i32, i32)
declare <16 x float> @llvm.amdgcn.mfma.f32.16x16x2bf16(<2 x i16>, <2 x i16>, <16 x float>, i32, i32, i32)
declare <4 x float> @llvm.amdgcn.mfma.f32.4x4x2bf16(<2 x i16>, <2 x i16>, <4 x float>, i32, i32, i32)
declare <16 x float> @llvm.amdgcn.mfma.f32.32x32x4bf16(<2 x i16>, <2 x i16>, <16 x float>, i32, i32, i32)
declare <4 x float> @llvm.amdgcn.mfma.f32.16x16x8bf16(<2 x i16>, <2 x i16>, <4 x float>, i32, i32, i32)
declare <32 x float> @llvm.amdgcn.mfma.f32.32x32x4bf16.1k(<4 x i16>, <4 x i16>, <32 x float>, i32, i32, i32)
declare <16 x float> @llvm.amdgcn.mfma.f32.16x16x4bf16.1k(<4 x i16>, <4 x i16>, <16 x float>, i32, i32, i32)
declare <4 x float> @llvm.amdgcn.mfma.f32.4x4x4bf16.1k(<4 x i16>, <4 x i16>, <4 x float>, i32, i32, i32)
declare <16 x float> @llvm.amdgcn.mfma.f32.32x32x8bf16.1k(<4 x i16>, <4 x i16>, <16 x float>, i32, i32, i32)
declare <4 x float> @llvm.amdgcn.mfma.f32.16x16x16bf16.1k(<4 x i16>, <4 x i16>, <4 x float>, i32, i32, i32)
declare <4 x double> @llvm.amdgcn.mfma.f64.16x16x4f64(double, double, <4 x double>, i32, i32, i32)
declare double @llvm.amdgcn.mfma.f64.4x4x4f64(double, double, double, i32, i32, i32)
declare <16 x i32> @llvm.amdgcn.mfma.i32.32x32x8i8(i32, i32, <16 x i32>, i32, i32, i32)
declare <4 x i32> @llvm.amdgcn.mfma.i32.16x16x16i8(i32, i32, <4 x i32>, i32, i32, i32)
; GCN-LABEL: {{^}}test_mfma_f32_32x32x2bf16:
; GCN: v_mfma_f32_32x32x2bf16 v[{{[0-9]+:[0-9]+}}], v{{[0-9]+}}, v{{[0-9]+}}, v[{{[0-9]+:[0-9]+}}]
define amdgpu_kernel void @test_mfma_f32_32x32x2bf16(ptr addrspace(1) %arg) {
bb:
%in.1 = load <32 x float>, ptr addrspace(1) %arg
%a = bitcast i32 1 to <2 x i16>
%b = bitcast i32 2 to <2 x i16>
%mai.1 = tail call <32 x float> @llvm.amdgcn.mfma.f32.32x32x2bf16(<2 x i16> %a, <2 x i16> %b, <32 x float> %in.1, i32 0, i32 0, i32 0)
store <32 x float> %mai.1, ptr addrspace(1) %arg
ret void
}
; GCN-LABEL: {{^}}test_mfma_f32_16x16x2bf16:
; GCN: v_mfma_f32_16x16x2bf16 v[{{[0-9]+:[0-9]+}}], v{{[0-9]+}}, v{{[0-9]+}}, v[{{[0-9]+:[0-9]+}}]
define amdgpu_kernel void @test_mfma_f32_16x16x2bf16(ptr addrspace(1) %arg) {
bb:
%in.1 = load <16 x float>, ptr addrspace(1) %arg
%mai.1 = tail call <16 x float> @llvm.amdgcn.mfma.f32.16x16x2bf16(<2 x i16> undef, <2 x i16> undef, <16 x float> %in.1, i32 0, i32 0, i32 0)
store <16 x float> %mai.1, ptr addrspace(1) %arg
ret void
}
; GCN-LABEL: {{^}}test_mfma_f32_4x4x2bf16:
; GCN: v_mfma_f32_4x4x2bf16 v[{{[0-9:]+:[0-9]+}}], v{{[0-9]+}}, v{{[0-9]+}}, v[{{[0-9]+:[0-9]+}}]
define amdgpu_kernel void @test_mfma_f32_4x4x2bf16(ptr addrspace(1) %arg) {
bb:
%in.1 = load <4 x float>, ptr addrspace(1) %arg
%mai.1 = tail call <4 x float> @llvm.amdgcn.mfma.f32.4x4x2bf16(<2 x i16> undef, <2 x i16> undef, <4 x float> %in.1, i32 0, i32 0, i32 0)
store <4 x float> %mai.1, ptr addrspace(1) %arg
ret void
}
; GCN-LABEL: {{^}}test_mfma_f32_32x32x4bf16:
; GCN: v_mfma_f32_32x32x4bf16 v[{{[0-9]+:[0-9]+}}], v{{[0-9]+}}, v{{[0-9]+}}, v[{{[0-9]+:[0-9]+}}]
define amdgpu_kernel void @test_mfma_f32_32x32x4bf16(ptr addrspace(1) %arg) {
bb:
%in.1 = load <16 x float>, ptr addrspace(1) %arg
%mai.1 = tail call <16 x float> @llvm.amdgcn.mfma.f32.32x32x4bf16(<2 x i16> undef, <2 x i16> undef, <16 x float> %in.1, i32 0, i32 0, i32 0)
store <16 x float> %mai.1, ptr addrspace(1) %arg
ret void
}
; GCN-LABEL: {{^}}test_mfma_f32_16x16x8bf16:
; GCN: v_mfma_f32_16x16x8bf16 v[{{[0-9:]+:[0-9]+}}], v{{[0-9]+}}, v{{[0-9]+}}, v[{{[0-9]+:[0-9]+}}]
define amdgpu_kernel void @test_mfma_f32_16x16x8bf16(ptr addrspace(1) %arg) {
bb:
%in.1 = load <4 x float>, ptr addrspace(1) %arg
%mai.1 = tail call <4 x float> @llvm.amdgcn.mfma.f32.16x16x8bf16(<2 x i16> undef, <2 x i16> undef, <4 x float> %in.1, i32 0, i32 0, i32 0)
store <4 x float> %mai.1, ptr addrspace(1) %arg
ret void
}
; GCN-LABEL: {{^}}test_mfma_f32_32x32x4bf16_1k:
; GCN: v_mfma_f32_32x32x4bf16_1k v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
define amdgpu_kernel void @test_mfma_f32_32x32x4bf16_1k(ptr addrspace(1) %arg) {
bb:
%in.1 = load <32 x float>, ptr addrspace(1) %arg
%mai.1 = tail call <32 x float> @llvm.amdgcn.mfma.f32.32x32x4bf16.1k(<4 x i16> undef, <4 x i16> undef, <32 x float> %in.1, i32 0, i32 0, i32 0)
store <32 x float> %mai.1, ptr addrspace(1) %arg
ret void
}
; GCN-LABEL: {{^}}test_mfma_f32_16x16x4bf16_1k:
; GCN: v_mfma_f32_16x16x4bf16_1k v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
define amdgpu_kernel void @test_mfma_f32_16x16x4bf16_1k(ptr addrspace(1) %arg) {
bb:
%in.1 = load <16 x float>, ptr addrspace(1) %arg
%mai.1 = tail call <16 x float> @llvm.amdgcn.mfma.f32.16x16x4bf16.1k(<4 x i16> undef, <4 x i16> undef, <16 x float> %in.1, i32 0, i32 0, i32 0)
store <16 x float> %mai.1, ptr addrspace(1) %arg
ret void
}
; GCN-LABEL: {{^}}test_mfma_f32_4x4x4bf16_1k:
; GCN: v_mfma_f32_4x4x4bf16_1k v[{{[0-9:]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
define amdgpu_kernel void @test_mfma_f32_4x4x4bf16_1k(ptr addrspace(1) %arg) {
bb:
%in.1 = load <4 x float>, ptr addrspace(1) %arg
%mai.1 = tail call <4 x float> @llvm.amdgcn.mfma.f32.4x4x4bf16.1k(<4 x i16> undef, <4 x i16> undef, <4 x float> %in.1, i32 0, i32 0, i32 0)
store <4 x float> %mai.1, ptr addrspace(1) %arg
ret void
}
; GCN-LABEL: {{^}}test_mfma_f32_32x32x8bf16_1k:
; GCN: v_mfma_f32_32x32x8bf16_1k v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
define amdgpu_kernel void @test_mfma_f32_32x32x8bf16_1k(ptr addrspace(1) %arg) {
bb:
%in.1 = load <16 x float>, ptr addrspace(1) %arg
%mai.1 = tail call <16 x float> @llvm.amdgcn.mfma.f32.32x32x8bf16.1k(<4 x i16> undef, <4 x i16> undef, <16 x float> %in.1, i32 0, i32 0, i32 0)
store <16 x float> %mai.1, ptr addrspace(1) %arg
ret void
}
; GCN-LABEL: {{^}}test_mfma_f32_16x16x16bf16_1k:
; GCN: v_mfma_f32_16x16x16bf16_1k v[{{[0-9:]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
define amdgpu_kernel void @test_mfma_f32_16x16x16bf16_1k(ptr addrspace(1) %arg) {
bb:
%in.1 = load <4 x float>, ptr addrspace(1) %arg
%mai.1 = tail call <4 x float> @llvm.amdgcn.mfma.f32.16x16x16bf16.1k(<4 x i16> undef, <4 x i16> undef, <4 x float> %in.1, i32 0, i32 0, i32 0)
store <4 x float> %mai.1, ptr addrspace(1) %arg
ret void
}
; GCN-LABEL: {{^}}test_mfma_f64_4x4x4f64:
; GCN: v_mfma_f64_4x4x4f64 v[{{[0-9:]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9:]+:[0-9]+}}
define amdgpu_kernel void @test_mfma_f64_4x4x4f64(ptr addrspace(1) %arg) {
bb:
%mai.1 = tail call double @llvm.amdgcn.mfma.f64.4x4x4f64(double 1.0, double 1.0, double 128.0, i32 0, i32 0, i32 0)
store double %mai.1, ptr addrspace(1) %arg
ret void
}
; GCN-LABEL: {{^}}test_mfma_f64_16x16x4f64:
; GCN: v_mfma_f64_16x16x4f64 v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}], v[{{[0-9]+:[0-9]+}}]
define amdgpu_kernel void @test_mfma_f64_16x16x4f64(ptr addrspace(1) %arg) {
bb:
%in.1 = load <4 x double>, ptr addrspace(1) %arg
%mai.1 = tail call <4 x double> @llvm.amdgcn.mfma.f64.16x16x4f64(double 1.0, double 1.0, <4 x double> %in.1, i32 0, i32 0, i32 0)
store <4 x double> %mai.1, ptr addrspace(1) %arg
ret void
}
; GCN-LABEL: {{^}}test_mfma_i32_32x32x8i8:
; GCN: v_mfma_i32_32x32x8i8 v[{{[0-9]+:[0-9]+}}], v{{[0-9]+}}, v{{[0-9]+}}, v[{{[0-9]+:[0-9]+}}]
define amdgpu_kernel void @test_mfma_i32_32x32x8i8(ptr addrspace(1) %arg) {
bb:
%in.1 = load <16 x i32>, ptr addrspace(1) %arg
%mai.1 = tail call <16 x i32> @llvm.amdgcn.mfma.i32.32x32x8i8(i32 1, i32 1, <16 x i32> %in.1, i32 0, i32 0, i32 0)
store <16 x i32> %mai.1, ptr addrspace(1) %arg
ret void
}
; GCN-LABEL: {{^}}test_mfma_i32_16x16x16i8:
; GCN: v_mfma_i32_16x16x16i8 v[{{[0-9:]+:[0-9]+}}], v{{[0-9]+}}, v{{[0-9]+}}, v[{{[0-9]+:[0-9]+}}]
define amdgpu_kernel void @test_mfma_i32_16x16x16i8(ptr addrspace(1) %arg) {
bb:
%in.1 = load <4 x i32>, ptr addrspace(1) %arg
%mai.1 = tail call <4 x i32> @llvm.amdgcn.mfma.i32.16x16x16i8(i32 1, i32 1, <4 x i32> %in.1, i32 0, i32 0, i32 0)
store <4 x i32> %mai.1, ptr addrspace(1) %arg
ret void
}
|
