TheRock/compiler/amd-llvm/llvm/test/CodeGen/AMDGPU/global-saddr-atomics.gfx1030.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1030 < %s | FileCheck --check-prefix=GCN %s

; --------------------------------------------------------------------------------
; amdgcn atomic csub
; --------------------------------------------------------------------------------

define amdgpu_ps float @global_csub_saddr_i32_rtn(ptr addrspace(1) inreg %sbase, i32 %voffset, i32 %data) {
; GCN-LABEL: global_csub_saddr_i32_rtn:
; GCN:       ; %bb.0:
; GCN-NEXT:    global_atomic_csub v0, v0, v1, s[2:3] glc
; GCN-NEXT:    s_waitcnt vmcnt(0)
; GCN-NEXT:    buffer_gl1_inv
; GCN-NEXT:    buffer_gl0_inv
; GCN-NEXT:    ; return to shader part epilog
  %zext.offset = zext i32 %voffset to i64
  %gep0 = getelementptr inbounds i8, ptr addrspace(1) %sbase, i64 %zext.offset
  %rtn = atomicrmw usub_sat ptr addrspace(1) %gep0, i32 %data seq_cst, !amdgpu.no.remote.memory !0
  %cast.rtn = bitcast i32 %rtn to float
  ret float %cast.rtn
}

define amdgpu_ps float @global_csub_saddr_i32_rtn_neg128(ptr addrspace(1) inreg %sbase, i32 %voffset, i32 %data) {
; GCN-LABEL: global_csub_saddr_i32_rtn_neg128:
; GCN:       ; %bb.0:
; GCN-NEXT:    global_atomic_csub v0, v0, v1, s[2:3] offset:-128 glc
; GCN-NEXT:    s_waitcnt vmcnt(0)
; GCN-NEXT:    buffer_gl1_inv
; GCN-NEXT:    buffer_gl0_inv
; GCN-NEXT:    ; return to shader part epilog
  %zext.offset = zext i32 %voffset to i64
  %gep0 = getelementptr inbounds i8, ptr addrspace(1) %sbase, i64 %zext.offset
  %gep1 = getelementptr inbounds i8, ptr addrspace(1) %gep0, i64 -128
  %rtn = atomicrmw usub_sat ptr addrspace(1) %gep1, i32 %data seq_cst, !amdgpu.no.remote.memory !0
  %cast.rtn = bitcast i32 %rtn to float
  ret float %cast.rtn
}

define amdgpu_ps void @global_csub_saddr_i32_nortn(ptr addrspace(1) inreg %sbase, i32 %voffset, i32 %data) {
; GCN-LABEL: global_csub_saddr_i32_nortn:
; GCN:       ; %bb.0:
; GCN-NEXT:    global_atomic_csub v0, v0, v1, s[2:3] glc
; GCN-NEXT:    s_waitcnt vmcnt(0)
; GCN-NEXT:    buffer_gl1_inv
; GCN-NEXT:    buffer_gl0_inv
; GCN-NEXT:    s_endpgm
  %zext.offset = zext i32 %voffset to i64
  %gep0 = getelementptr inbounds i8, ptr addrspace(1) %sbase, i64 %zext.offset
  %unused = atomicrmw usub_sat ptr addrspace(1) %gep0, i32 %data seq_cst, !amdgpu.no.remote.memory !0
  ret void
}

define amdgpu_ps void @global_csub_saddr_i32_nortn_neg128(ptr addrspace(1) inreg %sbase, i32 %voffset, i32 %data) {
; GCN-LABEL: global_csub_saddr_i32_nortn_neg128:
; GCN:       ; %bb.0:
; GCN-NEXT:    global_atomic_csub v0, v0, v1, s[2:3] offset:-128 glc
; GCN-NEXT:    s_waitcnt vmcnt(0)
; GCN-NEXT:    buffer_gl1_inv
; GCN-NEXT:    buffer_gl0_inv
; GCN-NEXT:    s_endpgm
  %zext.offset = zext i32 %voffset to i64
  %gep0 = getelementptr inbounds i8, ptr addrspace(1) %sbase, i64 %zext.offset
  %gep1 = getelementptr inbounds i8, ptr addrspace(1) %gep0, i64 -128
  %unused = atomicrmw usub_sat ptr addrspace(1) %gep1, i32 %data seq_cst, !amdgpu.no.remote.memory !0
  ret void
}

declare i32 @llvm.amdgcn.global.atomic.csub.p1(ptr addrspace(1) nocapture, i32) #0

attributes #0 = { argmemonly nounwind willreturn }

!0 = !{}