llvm/test/CodeGen/AMDGPU/target-cpu.ll

; RUN: llc -mtriple=amdgcn -mcpu=gfx600 -disable-promote-alloca-to-vector < %s | FileCheck %s

declare ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr() #1

declare i32 @llvm.amdgcn.workitem.id.x() #1

; CI+ intrinsic
declare void @llvm.amdgcn.s.dcache.inv.vol() #0

; VI+ intrinsic
declare void @llvm.amdgcn.s.dcache.wb() #0

; CHECK-LABEL: {{^}}target_none:
; CHECK: s_movk_i32 [[OFFSETREG:s[0-9]+]], 0x400
; CHECK: s_load_dwordx2 s{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, [[OFFSETREG]]
; CHECK: buffer_store_dword v{{[0-9]+}}, v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64
define amdgpu_kernel void @target_none() #0 {
  %kernargs = call ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
  %kernargs.gep = getelementptr inbounds i8, ptr addrspace(4) %kernargs, i64 1024
  %ptr = load ptr addrspace(1), ptr addrspace(4) %kernargs.gep
  %id = call i32 @llvm.amdgcn.workitem.id.x()
  %id.ext = sext i32 %id to i64
  %gep = getelementptr inbounds i32, ptr addrspace(1) %ptr, i64 %id.ext
  store i32 0, ptr addrspace(1) %gep
  ret void
}

; CHECK-LABEL: {{^}}target_tahiti:
; CHECK: s_movk_i32 [[OFFSETREG:s[0-9]+]], 0x400
; CHECK: s_load_dwordx2 s{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, [[OFFSETREG]]
; CHECK: buffer_store_dword v{{[0-9]+}}, v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64
define amdgpu_kernel void @target_tahiti() #1 {
  %kernargs = call ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
  %kernargs.gep = getelementptr inbounds i8, ptr addrspace(4) %kernargs, i64 1024
  %ptr = load ptr addrspace(1), ptr addrspace(4) %kernargs.gep
  %id = call i32 @llvm.amdgcn.workitem.id.x()
  %id.ext = sext i32 %id to i64
  %gep = getelementptr inbounds i32, ptr addrspace(1) %ptr, i64 %id.ext
  store i32 0, ptr addrspace(1) %gep
  ret void
}

; CHECK-LABEL: {{^}}target_bonaire:
; CHECK: s_load_dwordx2 s{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0x100
; CHECK: buffer_store_dword v{{[0-9]+}}, v{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0 addr64
; CHECK: s_dcache_inv_vol
define amdgpu_kernel void @target_bonaire() #3 {
  %kernargs = call ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
  %kernargs.gep = getelementptr inbounds i8, ptr addrspace(4) %kernargs, i64 1024
  %ptr = load ptr addrspace(1), ptr addrspace(4) %kernargs.gep
  %id = call i32 @llvm.amdgcn.workitem.id.x()
  %id.ext = sext i32 %id to i64
  %gep = getelementptr inbounds i32, ptr addrspace(1) %ptr, i64 %id.ext
  store i32 0, ptr addrspace(1) %gep
  call void @llvm.amdgcn.s.dcache.inv.vol()
  ret void
}

; CHECK-LABEL: {{^}}target_fiji:
; CHECK: s_load_dwordx2 s{{\[[0-9]+:[0-9]+\]}}, s{{\[[0-9]+:[0-9]+\]}}, 0x400
; CHECK: flat_store_dword
; CHECK: s_dcache_wb{{$}}
define amdgpu_kernel void @target_fiji() #4 {
  %kernargs = call ptr addrspace(4) @llvm.amdgcn.kernarg.segment.ptr()
  %kernargs.gep = getelementptr inbounds i8, ptr addrspace(4) %kernargs, i64 1024
  %ptr = load ptr addrspace(1), ptr addrspace(4) %kernargs.gep
  %id = call i32 @llvm.amdgcn.workitem.id.x()
  %id.ext = sext i32 %id to i64
  %gep = getelementptr inbounds i32, ptr addrspace(1) %ptr, i64 %id.ext
  store i32 0, ptr addrspace(1) %gep
  call void @llvm.amdgcn.s.dcache.wb()
  ret void
}

attributes #0 = { nounwind }
attributes #1 = { nounwind readnone }
attributes #2 = { nounwind "target-cpu"="tahiti" }
attributes #3 = { nounwind "target-cpu"="bonaire" }
attributes #4 = { nounwind "target-cpu"="fiji" }
attributes #5 = { nounwind "amdgpu-waves-per-eu"="1,3" "amdgpu-flat-work-group-size"="1,256" }