diff options
Diffstat (limited to 'llvm/test/CodeGen/AMDGPU/dagcombine-lshr-and-cmp.ll')
| -rw-r--r-- | llvm/test/CodeGen/AMDGPU/dagcombine-lshr-and-cmp.ll | 9 |
1 files changed, 5 insertions, 4 deletions
diff --git a/llvm/test/CodeGen/AMDGPU/dagcombine-lshr-and-cmp.ll b/llvm/test/CodeGen/AMDGPU/dagcombine-lshr-and-cmp.ll index 084b968..ce478d4 100644 --- a/llvm/test/CodeGen/AMDGPU/dagcombine-lshr-and-cmp.ll +++ b/llvm/test/CodeGen/AMDGPU/dagcombine-lshr-and-cmp.ll @@ -28,7 +28,6 @@ define i32 @divergent_lshr_and_cmp(i32 %x) { entry: %0 = and i32 %x, 2 %1 = icmp ne i32 %0, 0 - ; Prevent removal of truncate in SDag by inserting llvm.amdgcn.if br i1 %1, label %out.true, label %out.else out.true: @@ -43,9 +42,9 @@ define amdgpu_kernel void @uniform_opt_lshr_and_cmp(ptr addrspace(1) %out, i32 % ; GCN-LABEL: name: uniform_opt_lshr_and_cmp ; GCN: bb.0.entry: ; GCN-NEXT: successors: %bb.1(0x40000000), %bb.2(0x40000000) - ; GCN-NEXT: liveins: $sgpr4_sgpr5 + ; GCN-NEXT: liveins: $sgpr2_sgpr3 ; GCN-NEXT: {{ $}} - ; GCN-NEXT: [[COPY:%[0-9]+]]:sgpr_64(p4) = COPY $sgpr4_sgpr5 + ; GCN-NEXT: [[COPY:%[0-9]+]]:sgpr_64(p4) = COPY $sgpr2_sgpr3 ; GCN-NEXT: [[S_LOAD_DWORDX2_IMM:%[0-9]+]]:sreg_64_xexec = S_LOAD_DWORDX2_IMM [[COPY]](p4), 9, 0 :: (dereferenceable invariant load (s64) from %ir.out.kernarg.offset, align 4, addrspace 4) ; GCN-NEXT: [[S_LOAD_DWORD_IMM:%[0-9]+]]:sreg_32_xm0_xexec = S_LOAD_DWORD_IMM [[COPY]](p4), 11, 0 :: (dereferenceable invariant load (s32) from %ir.x.kernarg.offset, addrspace 4) ; GCN-NEXT: [[COPY1:%[0-9]+]]:sreg_64 = COPY [[S_LOAD_DWORDX2_IMM]] @@ -84,7 +83,6 @@ define amdgpu_kernel void @uniform_opt_lshr_and_cmp(ptr addrspace(1) %out, i32 % entry: %0 = and i32 %x, 2 %1 = icmp ne i32 %0, 0 - ; Don't optimize the truncate in the SDag away. br i1 %1, label %out.true, label %out.else out.true: @@ -96,3 +94,6 @@ out.else: store i1 %1, ptr addrspace(1) %out ret void } + +!llvm.module.flags = !{!0} +!0 = !{i32 1, !"amdgpu_code_object_version", i32 500} |