diff options
Diffstat (limited to 'llvm/test/CodeGen/AMDGPU/llvm.trunc.f16.ll')
| -rw-r--r-- | llvm/test/CodeGen/AMDGPU/llvm.trunc.f16.ll | 16 |
1 files changed, 6 insertions, 10 deletions
diff --git a/llvm/test/CodeGen/AMDGPU/llvm.trunc.f16.ll b/llvm/test/CodeGen/AMDGPU/llvm.trunc.f16.ll index dc7182aa0d8..9d45bbcaab9 100644 --- a/llvm/test/CodeGen/AMDGPU/llvm.trunc.f16.ll +++ b/llvm/test/CodeGen/AMDGPU/llvm.trunc.f16.ll @@ -25,21 +25,17 @@ entry: ; GCN-LABEL: {{^}}trunc_v2f16 ; GCN: buffer_load_dword v[[A_V2_F16:[0-9]+]] ; SI: v_cvt_f32_f16_e32 v[[A_F32_0:[0-9]+]], v[[A_V2_F16]] -; SI: v_lshrrev_b32_e32 v[[A_F16_1:[0-9]+]], 16, v[[A_V2_F16]] +; GCN: v_lshrrev_b32_e32 v[[A_F16_1:[0-9]+]], 16, v[[A_V2_F16]] ; SI: v_cvt_f32_f16_e32 v[[A_F32_1:[0-9]+]], v[[A_F16_1]] ; SI: v_trunc_f32_e32 v[[R_F32_0:[0-9]+]], v[[A_F32_0]] ; SI: v_cvt_f16_f32_e32 v[[R_F16_0:[0-9]+]], v[[R_F32_0]] ; SI: v_trunc_f32_e32 v[[R_F32_1:[0-9]+]], v[[A_F32_1]] ; SI: v_cvt_f16_f32_e32 v[[R_F16_1:[0-9]+]], v[[R_F32_1]] -; SI-DAG: v_lshlrev_b32_e32 v[[R_F16_HI:[0-9]+]], 16, v[[R_F16_1]] -; SI-NOT: v_and_b32 -; SI: v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_HI]], v[[R_F16_0]] - -; VI-DAG: v_trunc_f16_e32 v[[R_F16_0:[0-9]+]], v[[A_V2_F16]] -; VI-DAG: v_trunc_f16_sdwa v[[R_F16_1:[0-9]+]], v[[A_V2_F16]] dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 -; VI-NOT: v_and_b32 -; VI: v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_1]], v[[R_F16_0]] - +; VI-DAG: v_trunc_f16_e32 v[[R_F16_0:[0-9]+]], v[[A_V2_F16]] +; VI-DAG: v_trunc_f16_e32 v[[R_F16_1:[0-9]+]], v[[A_F16_1]] +; GCN-DAG: v_lshlrev_b32_e32 v[[R_F16_HI:[0-9]+]], 16, v[[R_F16_1]] +; GCN-NOT: v_and_b32 +; GCN: v_or_b32_e32 v[[R_V2_F16:[0-9]+]], v[[R_F16_HI]], v[[R_F16_0]] ; GCN: buffer_store_dword v[[R_V2_F16]] ; GCN: s_endpgm define amdgpu_kernel void @trunc_v2f16( |
