xref: /llvm-project/llvm/test/CodeGen/AMDGPU/wmma_modifiers.ll (revision a82032918cd445e5750e171f57d4f3d7096c021a)
1; RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -mattr=+wavefrontsize64 -verify-machineinstrs < %s | FileCheck %s
2
3declare <4 x float> @llvm.amdgcn.wmma.f32.16x16x16.f16.v4f32.v16f16(<16 x half>, <16 x half>, <4 x float>)
4
5; Make sure we don't crash when trying to select modifiers in SelectVOP3PMods.
6
7define amdgpu_cs void @xyz () {
8; CHECK-LABEL: xyz:
9; CHECK: v_wmma_f32_16x16x16_f16 v[0:3], v[0:7], v[0:7], v[0:3]
10
11.entry:
12  br label %loop
13loop:
14  %ld = load <8 x float>, ptr addrspace(5) null, align 32
15  %in_shuffle = shufflevector <8 x float> %ld, <8 x float> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
16  %wmma = call <4 x float> @llvm.amdgcn.wmma.f32.16x16x16.f16.v4f32.v16f16(<16 x half> undef, <16 x half> undef, <4 x float> %in_shuffle)
17  %out_shuffle = shufflevector <4 x float> %wmma, <4 x float> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
18  store <8 x float> %out_shuffle, ptr addrspace(5) null, align 32
19  br i1 false, label %.exit, label %loop
20.exit:
21  ret void
22}
23