1;RUN: llc < %s -march=amdgcn -mcpu=verde -verify-machineinstrs | FileCheck -check-prefix=GCN -check-prefix=PREGFX10 %s
2;RUN: llc < %s -march=amdgcn -mcpu=tonga -verify-machineinstrs | FileCheck -check-prefix=GCN -check-prefix=PREGFX10 %s
3;RUN: llc < %s -march=amdgcn -mcpu=gfx1010 -verify-machineinstrs | FileCheck -check-prefix=GCN -check-prefix=GFX10 %s
4;RUN: llc < %s -march=amdgcn -mcpu=gfx1100 -verify-machineinstrs | FileCheck -check-prefix=GCN -check-prefix=GFX11 %s
5
6; GCN-LABEL: {{^}}tbuffer_load:
7; PREGFX10: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 format:[BUF_DATA_FORMAT_32_32_32_32,BUF_NUM_FORMAT_UINT]
8; PREGFX10: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 format:[BUF_DATA_FORMAT_RESERVED_15,BUF_NUM_FORMAT_SSCALED] glc
9; PREGFX10: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 format:[BUF_DATA_FORMAT_10_11_11,BUF_NUM_FORMAT_SNORM] slc
10; PREGFX10: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 format:[BUF_DATA_FORMAT_10_11_11,BUF_NUM_FORMAT_SNORM] glc
11; GFX10-DAG: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 format:78
12; GFX10-DAG: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 format:[BUF_FMT_32_32_SINT] glc
13; GFX10-DAG: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 format:[BUF_FMT_32_FLOAT] slc
14; GFX10-DAG: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 format:[BUF_FMT_32_FLOAT] glc dlc
15; GFX11-DAG: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 format:78
16; GFX11-DAG: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 format:[BUF_FMT_32_32_32_32_FLOAT] glc
17; GFX11-DAG: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 format:[BUF_FMT_32_FLOAT] slc
18; GFX11-DAG: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 format:[BUF_FMT_32_FLOAT] glc dlc
19; GCN: s_waitcnt
20define amdgpu_vs {<4 x float>, <4 x float>, <4 x float>, <4 x float>} @tbuffer_load(<4 x i32> inreg) {
21main_body:
22    %vdata     = call <4 x i32> @llvm.amdgcn.raw.tbuffer.load.v4i32(<4 x i32> %0, i32 0, i32 0, i32 78, i32 0)
23    %vdata_glc = call <4 x i32> @llvm.amdgcn.raw.tbuffer.load.v4i32(<4 x i32> %0, i32 0, i32 0, i32 63, i32 1)
24    %vdata_slc = call <4 x i32> @llvm.amdgcn.raw.tbuffer.load.v4i32(<4 x i32> %0, i32 0, i32 0, i32 22, i32 2)
25    %vdata_f32 = call <4 x float> @llvm.amdgcn.raw.tbuffer.load.v4f32(<4 x i32> %0, i32 0, i32 0, i32 22, i32 5)
26    %vdata.f     = bitcast <4 x i32> %vdata to <4 x float>
27    %vdata_glc.f = bitcast <4 x i32> %vdata_glc to <4 x float>
28    %vdata_slc.f = bitcast <4 x i32> %vdata_slc to <4 x float>
29    %r0 = insertvalue {<4 x float>, <4 x float>, <4 x float>, <4 x float>} undef, <4 x float> %vdata.f, 0
30    %r1 = insertvalue {<4 x float>, <4 x float>, <4 x float>, <4 x float>} %r0, <4 x float> %vdata_glc.f, 1
31    %r2 = insertvalue {<4 x float>, <4 x float>, <4 x float>, <4 x float>} %r1, <4 x float> %vdata_slc.f, 2
32    %r3 = insertvalue {<4 x float>, <4 x float>, <4 x float>, <4 x float>} %r2, <4 x float> %vdata_f32, 3
33    ret {<4 x float>, <4 x float>, <4 x float>, <4 x float>} %r3
34}
35
36; GCN-LABEL: {{^}}tbuffer_load_immoffs:
37; PREGFX10: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 format:[BUF_DATA_FORMAT_32_32_32_32,BUF_NUM_FORMAT_UINT] offset:42
38; GFX10: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 format:78 offset:42
39; GFX11: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 format:78 offset:42
40define amdgpu_vs <4 x float> @tbuffer_load_immoffs(<4 x i32> inreg) {
41main_body:
42    %vdata   = call <4 x i32> @llvm.amdgcn.raw.tbuffer.load.v4i32(<4 x i32> %0, i32 42, i32 0, i32 78, i32 0)
43    %vdata.f = bitcast <4 x i32> %vdata to <4 x float>
44    ret <4 x float> %vdata.f
45}
46
47; GCN-LABEL: {{^}}tbuffer_load_immoffs_large
48; PREGFX10: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, 61 format:[BUF_DATA_FORMAT_RESERVED_15,BUF_NUM_FORMAT_USCALED] offset:4095
49; PREGFX10: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, {{s[0-9]+}} format:[BUF_DATA_FORMAT_32_32_32_32,BUF_NUM_FORMAT_SSCALED] offset:73
50; PREGFX10: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, {{s[0-9]+}} format:[BUF_DATA_FORMAT_32_32_32,BUF_NUM_FORMAT_UINT] offset:1
51; GFX10-DAG: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, 61 format:[BUF_FMT_10_10_10_2_SSCALED] offset:4095
52; GFX10-DAG: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, {{s[0-9]+}} format:[BUF_FMT_32_32_UINT] offset:73
53; GFX10-DAG: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, {{s[0-9]+}} format:[BUF_FMT_32_32_32_32_FLOAT] offset:1
54; GFX11-DAG: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, 61 format:[BUF_FMT_8_8_8_8_SINT] offset:4095
55; GFX11-DAG: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, {{s[0-9]+}} format:[BUF_FMT_32_32_32_32_SINT] offset:73
56; GFX11-DAG: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, {{s[0-9]+}} format:77 offset:1
57; GCN: s_waitcnt
58define amdgpu_vs {<4 x float>, <4 x float>, <4 x float>} @tbuffer_load_immoffs_large(<4 x i32> inreg, i32 inreg %soffs) {
59    %vdata     = call <4 x i32> @llvm.amdgcn.raw.tbuffer.load.v4i32(<4 x i32> %0, i32 4095, i32 61, i32 47, i32 0)
60    %vdata_glc = call <4 x i32> @llvm.amdgcn.raw.tbuffer.load.v4i32(<4 x i32> %0, i32 73, i32 %soffs, i32 62, i32 0)
61    %vdata_slc = call <4 x i32> @llvm.amdgcn.raw.tbuffer.load.v4i32(<4 x i32> %0, i32 1, i32 %soffs, i32 77, i32 0)
62    %vdata.f     = bitcast <4 x i32> %vdata to <4 x float>
63    %vdata_glc.f = bitcast <4 x i32> %vdata_glc to <4 x float>
64    %vdata_slc.f = bitcast <4 x i32> %vdata_slc to <4 x float>
65    %r0 = insertvalue {<4 x float>, <4 x float>, <4 x float>} undef, <4 x float> %vdata.f, 0
66    %r1 = insertvalue {<4 x float>, <4 x float>, <4 x float>} %r0, <4 x float> %vdata_glc.f, 1
67    %r2 = insertvalue {<4 x float>, <4 x float>, <4 x float>} %r1, <4 x float> %vdata_slc.f, 2
68    ret {<4 x float>, <4 x float>, <4 x float>} %r2
69}
70
71; GCN-LABEL: {{^}}tbuffer_load_ofs:
72; PREGFX10: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, 0 format:[BUF_DATA_FORMAT_32_32_32_32,BUF_NUM_FORMAT_UINT] offen
73; GFX10: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, 0 format:78 offen
74; GFX11: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, 0 format:78 offen
75define amdgpu_vs <4 x float> @tbuffer_load_ofs(<4 x i32> inreg, i32 %voffs) {
76main_body:
77    %vdata   = call <4 x i32> @llvm.amdgcn.raw.tbuffer.load.v4i32(<4 x i32> %0, i32 %voffs, i32 0, i32 78, i32 0)
78    %vdata.f = bitcast <4 x i32> %vdata to <4 x float>
79    ret <4 x float> %vdata.f
80}
81
82; GCN-LABEL: {{^}}tbuffer_load_ofs_imm:
83; PREGFX10: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, 0 format:[BUF_DATA_FORMAT_32_32_32_32,BUF_NUM_FORMAT_UINT] offen offset:52
84; GFX10: tbuffer_load_format_xyzw {{v\[[0-9]+:[0-9]+\]}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, 0 format:78 offen offset:52
85define amdgpu_vs <4 x float> @tbuffer_load_ofs_imm(<4 x i32> inreg, i32 %voffs) {
86main_body:
87    %ofs = add i32 %voffs, 52
88    %vdata   = call <4 x i32> @llvm.amdgcn.raw.tbuffer.load.v4i32(<4 x i32> %0, i32 %ofs, i32 0, i32 78, i32 0)
89    %vdata.f = bitcast <4 x i32> %vdata to <4 x float>
90    ret <4 x float> %vdata.f
91}
92
93; GCN-LABEL: {{^}}buffer_load_xy:
94; PREGFX10: tbuffer_load_format_xy {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 format:[BUF_DATA_FORMAT_32_32_32,BUF_NUM_FORMAT_UINT]
95; GFX10: tbuffer_load_format_xy {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 format:[BUF_FMT_32_32_32_32_FLOAT]
96; GFX11: tbuffer_load_format_xy {{v\[[0-9]+:[0-9]+\]}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 format:77
97define amdgpu_vs <2 x float> @buffer_load_xy(<4 x i32> inreg %rsrc) {
98    %vdata = call <2 x i32> @llvm.amdgcn.raw.tbuffer.load.v2i32(<4 x i32> %rsrc, i32 0, i32 0, i32 77, i32 0)
99    %vdata.f = bitcast <2 x i32> %vdata to <2 x float>
100    ret <2 x float> %vdata.f
101}
102
103; GCN-LABEL: {{^}}buffer_load_x:
104; PREGFX10: tbuffer_load_format_x {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 format:[BUF_DATA_FORMAT_32_32_32,BUF_NUM_FORMAT_UINT]
105; GFX10: tbuffer_load_format_x {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 format:[BUF_FMT_32_32_32_32_FLOAT]
106; GFX11: tbuffer_load_format_x {{v[0-9]+}}, off, {{s\[[0-9]+:[0-9]+\]}}, 0 format:77
107define amdgpu_vs float @buffer_load_x(<4 x i32> inreg %rsrc) {
108    %vdata = call i32 @llvm.amdgcn.raw.tbuffer.load.i32(<4 x i32> %rsrc, i32 0, i32 0, i32 77, i32 0)
109    %vdata.f = bitcast i32 %vdata to float
110    ret float %vdata.f
111}
112
113declare i32 @llvm.amdgcn.raw.tbuffer.load.i32(<4 x i32>, i32, i32, i32, i32)
114declare <2 x i32> @llvm.amdgcn.raw.tbuffer.load.v2i32(<4 x i32>, i32, i32, i32, i32)
115declare <4 x i32> @llvm.amdgcn.raw.tbuffer.load.v4i32(<4 x i32>, i32, i32, i32, i32)
116declare <4 x float> @llvm.amdgcn.raw.tbuffer.load.v4f32(<4 x i32>, i32, i32, i32, i32)
117
118