1 ; RUN: llc -march=amdgcn -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=GFX8 -check-prefix=NOAUTO %s
2 ; RUN: llc -march=amdgcn -mattr=+auto-waitcnt-before-barrier -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=GFX8 -check-prefix=AUTO %s
3 ; RUN: llc -march=amdgcn -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=GFX9 -check-prefix=NOAUTO %s
4 ; RUN: llc -march=amdgcn -mcpu=gfx900 -mattr=+auto-waitcnt-before-barrier -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=GFX9 -check-prefix=AUTO %s
6 ; GCN-LABEL: {{^}}test_barrier:
7 ; GFX8: buffer_store_dword
8 ; GFX9: global_store_dword
12 define amdgpu_kernel void @test_barrier(i32 addrspace(1)* %out, i32 %size) #0 {
14 %tmp = call i32 @llvm.amdgcn.workitem.id.x()
15 %tmp1 = getelementptr i32, i32 addrspace(1)* %out, i32 %tmp
16 store i32 %tmp, i32 addrspace(1)* %tmp1
17 call void @llvm.amdgcn.s.barrier()
18 %tmp3 = sub i32 %size, 1
19 %tmp4 = sub i32 %tmp3, %tmp
20 %tmp5 = getelementptr i32, i32 addrspace(1)* %out, i32 %tmp4
21 %tmp6 = load i32, i32 addrspace(1)* %tmp5
22 store i32 %tmp6, i32 addrspace(1)* %tmp1
26 declare void @llvm.amdgcn.s.barrier() #1
27 declare i32 @llvm.amdgcn.workitem.id.x() #2
29 attributes #0 = { nounwind }
30 attributes #1 = { convergent nounwind }
31 attributes #2 = { nounwind readnone }