1 ; RUN: llc -march=amdgcn -mcpu=gfx908 -verify-machineinstrs -show-mc-encoding < %s | FileCheck -check-prefixes=GFX9,GFX908 %s
2 ; RUN: llc -march=amdgcn -mcpu=gfx90a -verify-machineinstrs -show-mc-encoding < %s | FileCheck -check-prefixes=GFX9,GFX90A %s
4 ; GFX9-DAG: buffer_load_format_xyzw v[{{[0-9:]+}}], v{{[0-9]+}}, s[{{[0-9:]+}}], 0 idxen ; encoding:
5 ; GFX9-DAG: buffer_load_format_d16_xyzw v[{{[0-9:]+}}], v{{[0-9]+}}, s[{{[0-9:]+}}], 0 idxen ; encoding:
6 ; GFX908-DAG: v_mfma_i32_4x4x4i8 a[{{[0-9:]+}}], v{{[0-9]+}}, v{{[0-9]+}}, a[{{[0-9:]+}}] ; encoding: [{{0x..,0x0.,}}
7 ; GFX90A-DAG: v_mfma_i32_4x4x4i8 a[{{[0-9:]+}}], v{{[0-9]+}}, v{{[0-9]+}}, a[{{[0-9:]+}}] ; encoding: [{{0x..,0x8.,}}
8 define amdgpu_kernel void @test(<4 x i32> %x) {
9 %id = tail call i32 @llvm.amdgcn.workitem.id.x()
10 %r1 = tail call <4 x float> @llvm.amdgcn.buffer.load.format.v4f32(<4 x i32> %x, i32 %id, i32 0, i1 zeroext false, i1 zeroext false)
11 store volatile <4 x float> %r1, <4 x float>* undef
12 %r2 = tail call <4 x half> @llvm.amdgcn.buffer.load.format.v4f16(<4 x i32> %x, i32 %id, i32 0, i1 zeroext false, i1 zeroext false)
13 store volatile <4 x half> %r2, <4 x half>* undef
14 %r3 = tail call <4 x i32> @llvm.amdgcn.mfma.i32.4x4x4i8(i32 1, i32 2, <4 x i32> %x, i32 0, i32 0, i32 0)
15 store <4 x i32> %r3, <4 x i32>* undef
19 declare i32 @llvm.amdgcn.workitem.id.x()
20 declare <4 x float> @llvm.amdgcn.buffer.load.format.v4f32(<4 x i32>, i32, i32, i1 immarg, i1 immarg)
21 declare <4 x half> @llvm.amdgcn.buffer.load.format.v4f16(<4 x i32>, i32, i32, i1 immarg, i1 immarg)
22 declare <4 x i32> @llvm.amdgcn.mfma.i32.4x4x4i8(i32, i32, <4 x i32>, i32, i32, i32)