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