1; RUN: llc -march=amdgcn -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=FUNC %s
2; RUN: llc -march=amdgcn -mcpu=tonga -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=FUNC %s
3; RUN: llc -march=r600 -mcpu=redwood < %s | FileCheck -check-prefix=EG -check-prefix=FUNC %s
4
5
6; FUNC-LABEL: {{^}}local_load_i32:
7; GCN-NOT: s_wqm_b64
8; GCN: s_mov_b32 m0, -1
9; GCN: ds_read_b32
10
11; EG: LDS_READ_RET
12define void @local_load_i32(i32 addrspace(3)* %out, i32 addrspace(3)* %in) #0 {
13entry:
14  %ld = load i32, i32 addrspace(3)* %in
15  store i32 %ld, i32 addrspace(3)* %out
16  ret void
17}
18
19; FUNC-LABEL: {{^}}local_load_v2i32:
20; GCN: ds_read_b64
21define void @local_load_v2i32(<2 x i32> addrspace(3)* %out, <2 x i32> addrspace(3)* %in) #0 {
22entry:
23  %ld = load <2 x i32>, <2 x i32> addrspace(3)* %in
24  store <2 x i32> %ld, <2 x i32> addrspace(3)* %out
25  ret void
26}
27
28; FUNC-LABEL: {{^}}local_load_v3i32:
29; GCN-DAG: ds_read_b64
30; GCN-DAG: ds_read_b32
31define void @local_load_v3i32(<3 x i32> addrspace(3)* %out, <3 x i32> addrspace(3)* %in) #0 {
32entry:
33  %ld = load <3 x i32>, <3 x i32> addrspace(3)* %in
34  store <3 x i32> %ld, <3 x i32> addrspace(3)* %out
35  ret void
36}
37
38; FUNC-LABEL: {{^}}local_load_v4i32:
39; GCN: ds_read2_b64 v{{\[[0-9]+:[0-9]+\]}}, v{{[0-9]+}} offset0:1{{$}}
40
41define void @local_load_v4i32(<4 x i32> addrspace(3)* %out, <4 x i32> addrspace(3)* %in) #0 {
42entry:
43  %ld = load <4 x i32>, <4 x i32> addrspace(3)* %in
44  store <4 x i32> %ld, <4 x i32> addrspace(3)* %out
45  ret void
46}
47
48; FUNC-LABEL: {{^}}local_load_v8i32:
49; GCN-DAG: ds_read2_b64 v{{\[[0-9]+:[0-9]+\]}}, v{{[0-9]+}} offset0:3 offset1:2{{$}}
50; GCN-DAG: ds_read2_b64 v{{\[[0-9]+:[0-9]+\]}}, v{{[0-9]+}} offset0:1{{$}}
51define void @local_load_v8i32(<8 x i32> addrspace(3)* %out, <8 x i32> addrspace(3)* %in) #0 {
52entry:
53  %ld = load <8 x i32>, <8 x i32> addrspace(3)* %in
54  store <8 x i32> %ld, <8 x i32> addrspace(3)* %out
55  ret void
56}
57
58; FUNC-LABEL: {{^}}local_load_v16i32:
59; GCN-DAG: ds_read2_b64 v{{\[[0-9]+:[0-9]+\]}}, v{{[0-9]+}} offset0:3 offset1:4{{$}}
60; GCN-DAG: ds_read2_b64 v{{\[[0-9]+:[0-9]+\]}}, v{{[0-9]+}} offset0:5 offset1:6{{$}}
61; GCN-DAG: ds_read2_b64 v{{\[[0-9]+:[0-9]+\]}}, v{{[0-9]+}} offset0:7{{$}}
62; GCN-DAG: ds_read2_b64 v{{\[[0-9]+:[0-9]+\]}}, v{{[0-9]+}} offset0:1 offset1:2{{$}}
63define void @local_load_v16i32(<16 x i32> addrspace(3)* %out, <16 x i32> addrspace(3)* %in) #0 {
64entry:
65  %ld = load <16 x i32>, <16 x i32> addrspace(3)* %in
66  store <16 x i32> %ld, <16 x i32> addrspace(3)* %out
67  ret void
68}
69
70; FUNC-LABEL: {{^}}local_zextload_i32_to_i64:
71define void @local_zextload_i32_to_i64(i64 addrspace(3)* %out, i32 addrspace(3)* %in) #0 {
72  %ld = load i32, i32 addrspace(3)* %in
73  %ext = zext i32 %ld to i64
74  store i64 %ext, i64 addrspace(3)* %out
75  ret void
76}
77
78; FUNC-LABEL: {{^}}local_sextload_i32_to_i64:
79define void @local_sextload_i32_to_i64(i64 addrspace(3)* %out, i32 addrspace(3)* %in) #0 {
80  %ld = load i32, i32 addrspace(3)* %in
81  %ext = sext i32 %ld to i64
82  store i64 %ext, i64 addrspace(3)* %out
83  ret void
84}
85
86; FUNC-LABEL: {{^}}local_zextload_v1i32_to_v1i64:
87define void @local_zextload_v1i32_to_v1i64(<1 x i64> addrspace(3)* %out, <1 x i32> addrspace(3)* %in) #0 {
88  %ld = load <1 x i32>, <1 x i32> addrspace(3)* %in
89  %ext = zext <1 x i32> %ld to <1 x i64>
90  store <1 x i64> %ext, <1 x i64> addrspace(3)* %out
91  ret void
92}
93
94; FUNC-LABEL: {{^}}local_sextload_v1i32_to_v1i64:
95define void @local_sextload_v1i32_to_v1i64(<1 x i64> addrspace(3)* %out, <1 x i32> addrspace(3)* %in) #0 {
96  %ld = load <1 x i32>, <1 x i32> addrspace(3)* %in
97  %ext = sext <1 x i32> %ld to <1 x i64>
98  store <1 x i64> %ext, <1 x i64> addrspace(3)* %out
99  ret void
100}
101
102; FUNC-LABEL: {{^}}local_zextload_v2i32_to_v2i64:
103define void @local_zextload_v2i32_to_v2i64(<2 x i64> addrspace(3)* %out, <2 x i32> addrspace(3)* %in) #0 {
104  %ld = load <2 x i32>, <2 x i32> addrspace(3)* %in
105  %ext = zext <2 x i32> %ld to <2 x i64>
106  store <2 x i64> %ext, <2 x i64> addrspace(3)* %out
107  ret void
108}
109
110; FUNC-LABEL: {{^}}local_sextload_v2i32_to_v2i64:
111define void @local_sextload_v2i32_to_v2i64(<2 x i64> addrspace(3)* %out, <2 x i32> addrspace(3)* %in) #0 {
112  %ld = load <2 x i32>, <2 x i32> addrspace(3)* %in
113  %ext = sext <2 x i32> %ld to <2 x i64>
114  store <2 x i64> %ext, <2 x i64> addrspace(3)* %out
115  ret void
116}
117
118; FUNC-LABEL: {{^}}local_zextload_v4i32_to_v4i64:
119define void @local_zextload_v4i32_to_v4i64(<4 x i64> addrspace(3)* %out, <4 x i32> addrspace(3)* %in) #0 {
120  %ld = load <4 x i32>, <4 x i32> addrspace(3)* %in
121  %ext = zext <4 x i32> %ld to <4 x i64>
122  store <4 x i64> %ext, <4 x i64> addrspace(3)* %out
123  ret void
124}
125
126; FUNC-LABEL: {{^}}local_sextload_v4i32_to_v4i64:
127define void @local_sextload_v4i32_to_v4i64(<4 x i64> addrspace(3)* %out, <4 x i32> addrspace(3)* %in) #0 {
128  %ld = load <4 x i32>, <4 x i32> addrspace(3)* %in
129  %ext = sext <4 x i32> %ld to <4 x i64>
130  store <4 x i64> %ext, <4 x i64> addrspace(3)* %out
131  ret void
132}
133
134; FUNC-LABEL: {{^}}local_zextload_v8i32_to_v8i64:
135define void @local_zextload_v8i32_to_v8i64(<8 x i64> addrspace(3)* %out, <8 x i32> addrspace(3)* %in) #0 {
136  %ld = load <8 x i32>, <8 x i32> addrspace(3)* %in
137  %ext = zext <8 x i32> %ld to <8 x i64>
138  store <8 x i64> %ext, <8 x i64> addrspace(3)* %out
139  ret void
140}
141
142; FUNC-LABEL: {{^}}local_sextload_v8i32_to_v8i64:
143define void @local_sextload_v8i32_to_v8i64(<8 x i64> addrspace(3)* %out, <8 x i32> addrspace(3)* %in) #0 {
144  %ld = load <8 x i32>, <8 x i32> addrspace(3)* %in
145  %ext = sext <8 x i32> %ld to <8 x i64>
146  store <8 x i64> %ext, <8 x i64> addrspace(3)* %out
147  ret void
148}
149
150; FUNC-LABEL: {{^}}local_sextload_v16i32_to_v16i64:
151define void @local_sextload_v16i32_to_v16i64(<16 x i64> addrspace(3)* %out, <16 x i32> addrspace(3)* %in) #0 {
152  %ld = load <16 x i32>, <16 x i32> addrspace(3)* %in
153  %ext = sext <16 x i32> %ld to <16 x i64>
154  store <16 x i64> %ext, <16 x i64> addrspace(3)* %out
155  ret void
156}
157
158; FUNC-LABEL: {{^}}local_zextload_v16i32_to_v16i64
159define void @local_zextload_v16i32_to_v16i64(<16 x i64> addrspace(3)* %out, <16 x i32> addrspace(3)* %in) #0 {
160  %ld = load <16 x i32>, <16 x i32> addrspace(3)* %in
161  %ext = zext <16 x i32> %ld to <16 x i64>
162  store <16 x i64> %ext, <16 x i64> addrspace(3)* %out
163  ret void
164}
165
166; FUNC-LABEL: {{^}}local_sextload_v32i32_to_v32i64:
167define void @local_sextload_v32i32_to_v32i64(<32 x i64> addrspace(3)* %out, <32 x i32> addrspace(3)* %in) #0 {
168  %ld = load <32 x i32>, <32 x i32> addrspace(3)* %in
169  %ext = sext <32 x i32> %ld to <32 x i64>
170  store <32 x i64> %ext, <32 x i64> addrspace(3)* %out
171  ret void
172}
173
174; FUNC-LABEL: {{^}}local_zextload_v32i32_to_v32i64:
175define void @local_zextload_v32i32_to_v32i64(<32 x i64> addrspace(3)* %out, <32 x i32> addrspace(3)* %in) #0 {
176  %ld = load <32 x i32>, <32 x i32> addrspace(3)* %in
177  %ext = zext <32 x i32> %ld to <32 x i64>
178  store <32 x i64> %ext, <32 x i64> addrspace(3)* %out
179  ret void
180}
181
182attributes #0 = { nounwind }
183