|
1 | | -; RUN: llc -march=amdgcn -mcpu=gfx906 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=GCN,GFX906 |
| 1 | +; RUN: llc -march=amdgcn -mcpu=gfx906 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=GCN,GFX9,GFX906 |
| 2 | +; RUN: llc -march=amdgcn -mcpu=gfx940 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=GCN,GFX9,GFX940 |
2 | 3 | ; RUN: llc -march=amdgcn -mcpu=gfx1011 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=GCN,GFX10 |
3 | 4 | ; RUN: llc -march=amdgcn -mcpu=gfx1012 -verify-machineinstrs < %s | FileCheck %s --check-prefixes=GCN,GFX10 |
4 | 5 |
|
5 | 6 | declare float @llvm.amdgcn.fdot2(<2 x half> %a, <2 x half> %b, float %c, i1 %clamp) |
6 | 7 |
|
7 | 8 | ; GCN-LABEL: {{^}}test_llvm_amdgcn_fdot2_clamp |
8 | | -; GFX906: v_dot2_f32_f16 v{{[0-9]+}}, s{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} clamp{{$}} |
| 9 | +; GFX9: v_dot2_f32_f16 v{{[0-9]+}}, s{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} clamp{{$}} |
9 | 10 | ; GFX10: v_dot2_f32_f16 v{{[0-9]+}}, s{{[0-9]+}}, s{{[0-9]+}}, v{{[0-9]+}} clamp{{$}} |
10 | 11 | define amdgpu_kernel void @test_llvm_amdgcn_fdot2_clamp( |
11 | 12 | float addrspace(1)* %r, |
|
23 | 24 |
|
24 | 25 | ; GCN-LABEL: {{^}}test_llvm_amdgcn_fdot2_no_clamp |
25 | 26 | ; GFX906: v_dot2_f32_f16 v{{[0-9]+}}, s{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}{{$}} |
| 27 | +; GFX940: v_dot2c_f32_f16_e32 v{{[0-9]+}}, s{{[0-9]+}}, v{{[0-9]+}}{{$}} |
26 | 28 | ; GFX10: v_dot2c_f32_f16_e32 v{{[0-9]+}}, s{{[0-9]+}}, v{{[0-9]+}}{{$}} |
27 | 29 | define amdgpu_kernel void @test_llvm_amdgcn_fdot2_no_clamp( |
28 | 30 | float addrspace(1)* %r, |
|
38 | 40 | ret void |
39 | 41 | } |
40 | 42 |
|
41 | | -; GFX906-LABEL: {{^}}fdot2_inline_literal |
| 43 | +; GFX9-LABEL: {{^}}fdot2_inline_literal |
42 | 44 | ; GFX906: v_dot2_f32_f16 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}, 1.0 |
| 45 | +; GFX940: v_dot2c_f32_f16_e32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}{{$}} |
43 | 46 | define float @fdot2_inline_literal(<2 x half> %a, <2 x half> %b) { |
44 | 47 | %ret = tail call float @llvm.amdgcn.fdot2(<2 x half> %a, <2 x half> %b, float 1.0, i1 false) |
45 | 48 | ret float %ret |
|
0 commit comments