load-local-f32.ll
4.25 KB
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
; RUN: llc -march=amdgcn -verify-machineinstrs < %s | FileCheck -check-prefixes=GCN,SICIVI,FUNC %s
; RUN: llc -march=amdgcn -mcpu=tonga -mattr=-enable-ds128 -verify-machineinstrs < %s | FileCheck -check-prefixes=GCN,SICIVI,FUNC %s
; RUN: llc -march=r600 -mcpu=redwood < %s | FileCheck -check-prefixes=EG,FUNC %s
; Testing for ds_read/write_128
; RUN: llc -march=amdgcn -mcpu=tahiti -mattr=+enable-ds128 < %s | FileCheck -check-prefixes=SI,FUNC %s
; RUN: llc -march=amdgcn -mcpu=tonga -mattr=+enable-ds128 < %s | FileCheck -check-prefixes=CIVI,FUNC %s
; RUN: llc -march=amdgcn -mcpu=gfx900 -mattr=+enable-ds128 < %s | FileCheck -check-prefixes=CIVI,FUNC %s
; FUNC-LABEL: {{^}}load_f32_local:
; SICIVI: s_mov_b32 m0
; GFX9-NOT: m0
; GCN: ds_read_b32
; EG: LDS_READ_RET
define amdgpu_kernel void @load_f32_local(float addrspace(1)* %out, float addrspace(3)* %in) #0 {
entry:
%tmp0 = load float, float addrspace(3)* %in
store float %tmp0, float addrspace(1)* %out
ret void
}
; FUNC-LABEL: {{^}}load_v2f32_local:
; SICIVI: s_mov_b32 m0
; GFX9-NOT: m0
; GCN: ds_read_b64
; EG: LDS_READ_RET
; EG: LDS_READ_RET
define amdgpu_kernel void @load_v2f32_local(<2 x float> addrspace(1)* %out, <2 x float> addrspace(3)* %in) #0 {
entry:
%tmp0 = load <2 x float>, <2 x float> addrspace(3)* %in
store <2 x float> %tmp0, <2 x float> addrspace(1)* %out
ret void
}
; FIXME: should this do a read2_b64?
; FUNC-LABEL: {{^}}local_load_v3f32:
; SICIVI: s_mov_b32 m0
; GFX9-NOT: m0
; SI-DAG: ds_read_b32 v{{[0-9]+}}, v{{[0-9]+}} offset:8
; SI-DAG: ds_read_b64 v{{\[[0-9]+:[0-9]+\]}}, v{{[0-9]+$}}
; CIVI-DAG: ds_read_b96 v{{\[[0-9]+:[0-9]+\]}}, v{{[0-9]+$}}
; GCN: s_waitcnt
; SI-DAG: ds_write_b64
; SI-DAG: ds_write_b32 v{{[0-9]+}}, v{{[0-9]+}} offset:8{{$}}
; CIVI-DAG: ds_write_b96 v{{[0-9]+}}, v{{\[[0-9]+:[0-9]+\]}}
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
define amdgpu_kernel void @local_load_v3f32(<3 x float> addrspace(3)* %out, <3 x float> addrspace(3)* %in) #0 {
entry:
%tmp0 = load <3 x float>, <3 x float> addrspace(3)* %in
store <3 x float> %tmp0, <3 x float> addrspace(3)* %out
ret void
}
; FUNC-LABEL: {{^}}local_load_v4f32:
; SICIVI: s_mov_b32 m0
; GFX9-NOT: m0
; GCN: ds_read2_b64
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
define amdgpu_kernel void @local_load_v4f32(<4 x float> addrspace(3)* %out, <4 x float> addrspace(3)* %in) #0 {
entry:
%tmp0 = load <4 x float>, <4 x float> addrspace(3)* %in
store <4 x float> %tmp0, <4 x float> addrspace(3)* %out
ret void
}
; FUNC-LABEL: {{^}}local_load_v8f32:
; SICIVI: s_mov_b32 m0
; GFX9-NOT: m0
; GCN: ds_read2_b64
; GCN: ds_read2_b64
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
define amdgpu_kernel void @local_load_v8f32(<8 x float> addrspace(3)* %out, <8 x float> addrspace(3)* %in) #0 {
entry:
%tmp0 = load <8 x float>, <8 x float> addrspace(3)* %in
store <8 x float> %tmp0, <8 x float> addrspace(3)* %out
ret void
}
; FUNC-LABEL: {{^}}local_load_v16f32:
; SICIVI: s_mov_b32 m0
; GFX9-NOT: m0
; GCN: ds_read2_b64
; GCN: ds_read2_b64
; GCN: ds_read2_b64
; GCN: ds_read2_b64
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
define amdgpu_kernel void @local_load_v16f32(<16 x float> addrspace(3)* %out, <16 x float> addrspace(3)* %in) #0 {
entry:
%tmp0 = load <16 x float>, <16 x float> addrspace(3)* %in
store <16 x float> %tmp0, <16 x float> addrspace(3)* %out
ret void
}
; Tests if ds_read/write_b128 gets generated for the 16 byte aligned load.
; FUNC-LABEL: {{^}}local_v4f32_to_128:
; SI-NOT: ds_read_b128
; SI-NOT: ds_write_b128
; CIVI: ds_read_b128
; CIVI: ds_write_b128
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
; EG: LDS_READ_RET
define amdgpu_kernel void @local_v4f32_to_128(<4 x float> addrspace(3)* %out, <4 x float> addrspace(3)* %in) {
%ld = load <4 x float>, <4 x float> addrspace(3)* %in, align 16
store <4 x float> %ld, <4 x float> addrspace(3)* %out, align 16
ret void
}
attributes #0 = { nounwind }