• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1; RUN: llc -march=amdgcn -enable-no-signed-zeros-fp-math=0 < %s | FileCheck -check-prefix=GCN -check-prefix=GCN-SAFE %s
2; RUN: llc -march=amdgcn -enable-no-signed-zeros-fp-math=1 < %s | FileCheck -check-prefix=GCN -check-prefix=GCN-UNSAFE %s
3; RUN: llc -march=amdgcn -enable-unsafe-fp-math < %s | FileCheck -check-prefix=GCN -check-prefix=GCN-UNSAFE %s
4
5declare i32 @llvm.amdgcn.workitem.id.x() nounwind readnone
6
7; Test that the -enable-no-signed-zeros-fp-math flag works
8
9; GCN-LABEL: {{^}}fneg_fsub_f32:
10; GCN: v_sub_f32_e32 [[SUB:v[0-9]+]], {{v[0-9]+}}, {{v[0-9]+}}
11; GCN-SAFE: v_xor_b32_e32 v{{[0-9]+}}, 0x80000000, [[SUB]]
12
13; GCN-UNSAFE-NOT: xor
14define amdgpu_kernel void @fneg_fsub_f32(float addrspace(1)* %out, float addrspace(1)* %in) #0 {
15  %tid = call i32 @llvm.amdgcn.workitem.id.x()
16  %add = add i32 %tid, 1
17  %gep = getelementptr float, float addrspace(1)* %in, i32 %tid
18  %b_ptr = getelementptr float, float addrspace(1)* %in, i32 %add
19  %a = load float, float addrspace(1)* %gep, align 4
20  %b = load float, float addrspace(1)* %b_ptr, align 4
21  %result = fsub float %a, %b
22  %neg.result = fsub float -0.0, %result
23  store float %neg.result, float addrspace(1)* %out, align 4
24  ret void
25}
26
27attributes #0 = { nounwind }
28