; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -global-isel -new-reg-bank-select -mtriple=amdgcn -mcpu=tonga -amdgpu-dpp-combine=false < %s | FileCheck -check-prefix=GFX8 %s
; RUN: llc -global-isel -new-reg-bank-select -mtriple=amdgcn -mcpu=gfx1010 -amdgpu-dpp-combine=false < %s | FileCheck -check-prefix=GFX10 %s
; RUN: llc -global-isel -new-reg-bank-select -mtriple=amdgcn -mcpu=gfx1100 -amdgpu-enable-delay-alu=0 -amdgpu-dpp-combine=false < %s | FileCheck -check-prefix=GFX11 %s

define amdgpu_kernel void @dpp_test(ptr addrspace(1) %out, i32 %in1, i32 %in2) {
; GFX8-LABEL: dpp_test:
; GFX8:       ; %bb.0:
; GFX8-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
; GFX8-NEXT:    v_mov_b32_e32 v2, s2
; GFX8-NEXT:    v_mov_b32_e32 v0, s3
; GFX8-NEXT:    s_nop 1
; GFX8-NEXT:    v_mov_b32_dpp v2, v0 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX8-NEXT:    v_mov_b32_e32 v0, s0
; GFX8-NEXT:    v_mov_b32_e32 v1, s1
; GFX8-NEXT:    flat_store_dword v[0:1], v2
; GFX8-NEXT:    s_endpgm
;
; GFX10-LABEL: dpp_test:
; GFX10:       ; %bb.0:
; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
; GFX10-NEXT:    v_mov_b32_e32 v0, s2
; GFX10-NEXT:    v_mov_b32_e32 v1, s3
; GFX10-NEXT:    v_mov_b32_dpp v0, v1 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX10-NEXT:    v_mov_b32_e32 v1, 0
; GFX10-NEXT:    global_store_dword v1, v0, s[0:1]
; GFX10-NEXT:    s_endpgm
;
; GFX11-LABEL: dpp_test:
; GFX11:       ; %bb.0:
; GFX11-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
; GFX11-NEXT:    v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3
; GFX11-NEXT:    v_mov_b32_dpp v0, v1 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX11-NEXT:    v_mov_b32_e32 v1, 0
; GFX11-NEXT:    global_store_b32 v1, v0, s[0:1]
; GFX11-NEXT:    s_endpgm
  %tmp0 = call i32 @llvm.amdgcn.update.dpp.i32(i32 %in1, i32 %in2, i32 1, i32 1, i32 1, i1 false)
  store i32 %tmp0, ptr addrspace(1) %out
  ret void
}
define amdgpu_kernel void @update_dppi64_test(ptr addrspace(1) %arg, i64 %in1, i64 %in2) {
; GFX8-LABEL: update_dppi64_test:
; GFX8:       ; %bb.0:
; GFX8-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX8-NEXT:    v_lshlrev_b32_e32 v2, 3, v0
; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
; GFX8-NEXT:    v_mov_b32_e32 v0, s0
; GFX8-NEXT:    v_mov_b32_e32 v1, s1
; GFX8-NEXT:    v_add_u32_e32 v0, vcc, v0, v2
; GFX8-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
; GFX8-NEXT:    flat_load_dwordx2 v[2:3], v[0:1]
; GFX8-NEXT:    v_mov_b32_e32 v4, s2
; GFX8-NEXT:    v_mov_b32_e32 v5, s3
; GFX8-NEXT:    s_waitcnt vmcnt(0)
; GFX8-NEXT:    v_mov_b32_dpp v4, v2 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX8-NEXT:    v_mov_b32_dpp v5, v3 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX8-NEXT:    flat_store_dwordx2 v[0:1], v[4:5]
; GFX8-NEXT:    s_endpgm
;
; GFX10-LABEL: update_dppi64_test:
; GFX10:       ; %bb.0:
; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX10-NEXT:    v_lshlrev_b32_e32 v4, 3, v0
; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
; GFX10-NEXT:    global_load_dwordx2 v[0:1], v4, s[0:1]
; GFX10-NEXT:    v_mov_b32_e32 v2, s2
; GFX10-NEXT:    v_mov_b32_e32 v3, s3
; GFX10-NEXT:    s_waitcnt vmcnt(0)
; GFX10-NEXT:    v_mov_b32_dpp v2, v0 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX10-NEXT:    v_mov_b32_dpp v3, v1 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX10-NEXT:    global_store_dwordx2 v4, v[2:3], s[0:1]
; GFX10-NEXT:    s_endpgm
;
; GFX11-LABEL: update_dppi64_test:
; GFX11:       ; %bb.0:
; GFX11-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
; GFX11-NEXT:    v_dual_mov_b32 v3, s3 :: v_dual_and_b32 v0, 0x3ff, v0
; GFX11-NEXT:    v_lshlrev_b32_e32 v4, 3, v0
; GFX11-NEXT:    v_mov_b32_e32 v2, s2
; GFX11-NEXT:    global_load_b64 v[0:1], v4, s[0:1]
; GFX11-NEXT:    s_waitcnt vmcnt(0)
; GFX11-NEXT:    v_mov_b32_dpp v2, v0 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX11-NEXT:    v_mov_b32_dpp v3, v1 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX11-NEXT:    global_store_b64 v4, v[2:3], s[0:1]
; GFX11-NEXT:    s_endpgm
  %id = tail call i32 @llvm.amdgcn.workitem.id.x()
  %gep = getelementptr inbounds i64, ptr addrspace(1) %arg, i32 %id
  %load = load i64, ptr addrspace(1) %gep
  %tmp0 = call i64 @llvm.amdgcn.update.dpp.i64(i64 %in1, i64 %load, i32 1, i32 1, i32 1, i1 false) #1
  store i64 %tmp0, ptr addrspace(1) %gep
  ret void
}

define amdgpu_kernel void @update_dppf64_test(ptr addrspace(1) %arg, double %in1, double %in2) {
; GFX8-LABEL: update_dppf64_test:
; GFX8:       ; %bb.0:
; GFX8-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX8-NEXT:    v_lshlrev_b32_e32 v2, 3, v0
; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
; GFX8-NEXT:    v_mov_b32_e32 v0, s0
; GFX8-NEXT:    v_mov_b32_e32 v1, s1
; GFX8-NEXT:    v_add_u32_e32 v0, vcc, v0, v2
; GFX8-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
; GFX8-NEXT:    flat_load_dwordx2 v[2:3], v[0:1]
; GFX8-NEXT:    v_mov_b32_e32 v4, s2
; GFX8-NEXT:    v_mov_b32_e32 v5, s3
; GFX8-NEXT:    s_waitcnt vmcnt(0)
; GFX8-NEXT:    v_mov_b32_dpp v4, v2 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX8-NEXT:    v_mov_b32_dpp v5, v3 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX8-NEXT:    flat_store_dwordx2 v[0:1], v[4:5]
; GFX8-NEXT:    s_endpgm
;
; GFX10-LABEL: update_dppf64_test:
; GFX10:       ; %bb.0:
; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX10-NEXT:    v_lshlrev_b32_e32 v4, 3, v0
; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
; GFX10-NEXT:    global_load_dwordx2 v[0:1], v4, s[0:1]
; GFX10-NEXT:    v_mov_b32_e32 v2, s2
; GFX10-NEXT:    v_mov_b32_e32 v3, s3
; GFX10-NEXT:    s_waitcnt vmcnt(0)
; GFX10-NEXT:    v_mov_b32_dpp v2, v0 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX10-NEXT:    v_mov_b32_dpp v3, v1 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX10-NEXT:    global_store_dwordx2 v4, v[2:3], s[0:1]
; GFX10-NEXT:    s_endpgm
;
; GFX11-LABEL: update_dppf64_test:
; GFX11:       ; %bb.0:
; GFX11-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
; GFX11-NEXT:    v_dual_mov_b32 v3, s3 :: v_dual_and_b32 v0, 0x3ff, v0
; GFX11-NEXT:    v_lshlrev_b32_e32 v4, 3, v0
; GFX11-NEXT:    v_mov_b32_e32 v2, s2
; GFX11-NEXT:    global_load_b64 v[0:1], v4, s[0:1]
; GFX11-NEXT:    s_waitcnt vmcnt(0)
; GFX11-NEXT:    v_mov_b32_dpp v2, v0 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX11-NEXT:    v_mov_b32_dpp v3, v1 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX11-NEXT:    global_store_b64 v4, v[2:3], s[0:1]
; GFX11-NEXT:    s_endpgm
  %id = tail call i32 @llvm.amdgcn.workitem.id.x()
  %gep = getelementptr inbounds double, ptr addrspace(1) %arg, i32 %id
  %load = load double, ptr addrspace(1) %gep
  %tmp0 = call double @llvm.amdgcn.update.dpp.f64(double %in1, double %load, i32 1, i32 1, i32 1, i1 false) #1
  store double %tmp0, ptr addrspace(1) %gep
  ret void
}

define amdgpu_kernel void @update_dppv2i32_test(ptr addrspace(1) %arg, <2 x i32> %in1, <2 x i32> %in2) {
; GFX8-LABEL: update_dppv2i32_test:
; GFX8:       ; %bb.0:
; GFX8-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX8-NEXT:    v_lshlrev_b32_e32 v2, 3, v0
; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
; GFX8-NEXT:    v_mov_b32_e32 v0, s0
; GFX8-NEXT:    v_mov_b32_e32 v1, s1
; GFX8-NEXT:    v_add_u32_e32 v0, vcc, v0, v2
; GFX8-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
; GFX8-NEXT:    flat_load_dwordx2 v[2:3], v[0:1]
; GFX8-NEXT:    v_mov_b32_e32 v4, s2
; GFX8-NEXT:    v_mov_b32_e32 v5, s3
; GFX8-NEXT:    s_waitcnt vmcnt(0)
; GFX8-NEXT:    v_mov_b32_dpp v4, v2 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX8-NEXT:    v_mov_b32_dpp v5, v3 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX8-NEXT:    flat_store_dwordx2 v[0:1], v[4:5]
; GFX8-NEXT:    s_endpgm
;
; GFX10-LABEL: update_dppv2i32_test:
; GFX10:       ; %bb.0:
; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX10-NEXT:    v_lshlrev_b32_e32 v4, 3, v0
; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
; GFX10-NEXT:    global_load_dwordx2 v[0:1], v4, s[0:1]
; GFX10-NEXT:    v_mov_b32_e32 v2, s2
; GFX10-NEXT:    v_mov_b32_e32 v3, s3
; GFX10-NEXT:    s_waitcnt vmcnt(0)
; GFX10-NEXT:    v_mov_b32_dpp v2, v0 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX10-NEXT:    v_mov_b32_dpp v3, v1 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX10-NEXT:    global_store_dwordx2 v4, v[2:3], s[0:1]
; GFX10-NEXT:    s_endpgm
;
; GFX11-LABEL: update_dppv2i32_test:
; GFX11:       ; %bb.0:
; GFX11-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
; GFX11-NEXT:    v_dual_mov_b32 v3, s3 :: v_dual_and_b32 v0, 0x3ff, v0
; GFX11-NEXT:    v_lshlrev_b32_e32 v4, 3, v0
; GFX11-NEXT:    v_mov_b32_e32 v2, s2
; GFX11-NEXT:    global_load_b64 v[0:1], v4, s[0:1]
; GFX11-NEXT:    s_waitcnt vmcnt(0)
; GFX11-NEXT:    v_mov_b32_dpp v2, v0 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX11-NEXT:    v_mov_b32_dpp v3, v1 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX11-NEXT:    global_store_b64 v4, v[2:3], s[0:1]
; GFX11-NEXT:    s_endpgm
  %id = tail call i32 @llvm.amdgcn.workitem.id.x()
  %gep = getelementptr inbounds <2 x i32>, ptr addrspace(1) %arg, i32 %id
  %load = load <2 x i32>, ptr addrspace(1) %gep
  %tmp0 = call <2 x i32> @llvm.amdgcn.update.dpp.v2i32(<2 x i32> %in1, <2 x i32> %load, i32 1, i32 1, i32 1, i1 false) #1
  store <2 x i32> %tmp0, ptr addrspace(1) %gep
  ret void
}

define amdgpu_kernel void @update_dppv2f32_test(ptr addrspace(1) %arg, <2 x float> %in1, <2 x float> %in2) {
; GFX8-LABEL: update_dppv2f32_test:
; GFX8:       ; %bb.0:
; GFX8-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX8-NEXT:    v_lshlrev_b32_e32 v2, 3, v0
; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
; GFX8-NEXT:    v_mov_b32_e32 v0, s0
; GFX8-NEXT:    v_mov_b32_e32 v1, s1
; GFX8-NEXT:    v_add_u32_e32 v0, vcc, v0, v2
; GFX8-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
; GFX8-NEXT:    flat_load_dwordx2 v[2:3], v[0:1]
; GFX8-NEXT:    v_mov_b32_e32 v4, s2
; GFX8-NEXT:    v_mov_b32_e32 v5, s3
; GFX8-NEXT:    s_waitcnt vmcnt(0)
; GFX8-NEXT:    v_mov_b32_dpp v4, v2 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX8-NEXT:    v_mov_b32_dpp v5, v3 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX8-NEXT:    flat_store_dwordx2 v[0:1], v[4:5]
; GFX8-NEXT:    s_endpgm
;
; GFX10-LABEL: update_dppv2f32_test:
; GFX10:       ; %bb.0:
; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX10-NEXT:    v_lshlrev_b32_e32 v4, 3, v0
; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
; GFX10-NEXT:    global_load_dwordx2 v[0:1], v4, s[0:1]
; GFX10-NEXT:    v_mov_b32_e32 v2, s2
; GFX10-NEXT:    v_mov_b32_e32 v3, s3
; GFX10-NEXT:    s_waitcnt vmcnt(0)
; GFX10-NEXT:    v_mov_b32_dpp v2, v0 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX10-NEXT:    v_mov_b32_dpp v3, v1 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX10-NEXT:    global_store_dwordx2 v4, v[2:3], s[0:1]
; GFX10-NEXT:    s_endpgm
;
; GFX11-LABEL: update_dppv2f32_test:
; GFX11:       ; %bb.0:
; GFX11-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
; GFX11-NEXT:    v_dual_mov_b32 v3, s3 :: v_dual_and_b32 v0, 0x3ff, v0
; GFX11-NEXT:    v_lshlrev_b32_e32 v4, 3, v0
; GFX11-NEXT:    v_mov_b32_e32 v2, s2
; GFX11-NEXT:    global_load_b64 v[0:1], v4, s[0:1]
; GFX11-NEXT:    s_waitcnt vmcnt(0)
; GFX11-NEXT:    v_mov_b32_dpp v2, v0 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX11-NEXT:    v_mov_b32_dpp v3, v1 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX11-NEXT:    global_store_b64 v4, v[2:3], s[0:1]
; GFX11-NEXT:    s_endpgm
  %id = tail call i32 @llvm.amdgcn.workitem.id.x()
  %gep = getelementptr inbounds <2 x float>, ptr addrspace(1) %arg, i32 %id
  %load = load <2 x float>, ptr addrspace(1) %gep
  %tmp0 = call <2 x float> @llvm.amdgcn.update.dpp.v2f32(<2 x float> %in1, <2 x float> %load, i32 1, i32 1, i32 1, i1 false) #1
  store <2 x float> %tmp0, ptr addrspace(1) %gep
  ret void
}

define amdgpu_kernel void @update_dpp_p0_test(ptr addrspace(1) %arg, ptr %in1, ptr %in2) {
; GFX8-LABEL: update_dpp_p0_test:
; GFX8:       ; %bb.0:
; GFX8-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX8-NEXT:    v_lshlrev_b32_e32 v2, 3, v0
; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
; GFX8-NEXT:    v_mov_b32_e32 v0, s0
; GFX8-NEXT:    v_mov_b32_e32 v1, s1
; GFX8-NEXT:    v_add_u32_e32 v0, vcc, v0, v2
; GFX8-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
; GFX8-NEXT:    flat_load_dwordx2 v[2:3], v[0:1]
; GFX8-NEXT:    v_mov_b32_e32 v4, s2
; GFX8-NEXT:    v_mov_b32_e32 v5, s3
; GFX8-NEXT:    s_waitcnt vmcnt(0)
; GFX8-NEXT:    v_mov_b32_dpp v4, v2 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX8-NEXT:    v_mov_b32_dpp v5, v3 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX8-NEXT:    flat_store_dwordx2 v[0:1], v[4:5]
; GFX8-NEXT:    s_endpgm
;
; GFX10-LABEL: update_dpp_p0_test:
; GFX10:       ; %bb.0:
; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x24
; GFX10-NEXT:    v_lshlrev_b32_e32 v4, 3, v0
; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
; GFX10-NEXT:    global_load_dwordx2 v[0:1], v4, s[0:1]
; GFX10-NEXT:    v_mov_b32_e32 v2, s2
; GFX10-NEXT:    v_mov_b32_e32 v3, s3
; GFX10-NEXT:    s_waitcnt vmcnt(0)
; GFX10-NEXT:    v_mov_b32_dpp v2, v0 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX10-NEXT:    v_mov_b32_dpp v3, v1 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX10-NEXT:    global_store_dwordx2 v4, v[2:3], s[0:1]
; GFX10-NEXT:    s_endpgm
;
; GFX11-LABEL: update_dpp_p0_test:
; GFX11:       ; %bb.0:
; GFX11-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
; GFX11-NEXT:    v_dual_mov_b32 v3, s3 :: v_dual_and_b32 v0, 0x3ff, v0
; GFX11-NEXT:    v_lshlrev_b32_e32 v4, 3, v0
; GFX11-NEXT:    v_mov_b32_e32 v2, s2
; GFX11-NEXT:    global_load_b64 v[0:1], v4, s[0:1]
; GFX11-NEXT:    s_waitcnt vmcnt(0)
; GFX11-NEXT:    v_mov_b32_dpp v2, v0 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX11-NEXT:    v_mov_b32_dpp v3, v1 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX11-NEXT:    global_store_b64 v4, v[2:3], s[0:1]
; GFX11-NEXT:    s_endpgm
  %id = tail call i32 @llvm.amdgcn.workitem.id.x()
  %gep = getelementptr inbounds ptr, ptr addrspace(1) %arg, i32 %id
  %load = load ptr, ptr addrspace(1) %gep
  %tmp0 = call ptr @llvm.amdgcn.update.dpp.v2f32(ptr %in1, ptr %load, i32 1, i32 1, i32 1, i1 false) #1
  store ptr %tmp0, ptr addrspace(1) %gep
  ret void
}

define amdgpu_kernel void @update_dpp_p3_test(ptr addrspace(3) %arg, ptr addrspace(3) %in1, ptr %in2) {
; GFX8-LABEL: update_dpp_p3_test:
; GFX8:       ; %bb.0:
; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX8-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX8-NEXT:    s_mov_b32 m0, -1
; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
; GFX8-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
; GFX8-NEXT:    ds_read_b32 v1, v0
; GFX8-NEXT:    v_mov_b32_e32 v2, s1
; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
; GFX8-NEXT:    s_nop 0
; GFX8-NEXT:    v_mov_b32_dpp v2, v1 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX8-NEXT:    ds_write_b32 v0, v2
; GFX8-NEXT:    s_endpgm
;
; GFX10-LABEL: update_dpp_p3_test:
; GFX10:       ; %bb.0:
; GFX10-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX10-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
; GFX10-NEXT:    v_add_nc_u32_e32 v0, s0, v0
; GFX10-NEXT:    v_mov_b32_e32 v2, s1
; GFX10-NEXT:    ds_read_b32 v1, v0
; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
; GFX10-NEXT:    v_mov_b32_dpp v2, v1 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX10-NEXT:    ds_write_b32 v0, v2
; GFX10-NEXT:    s_endpgm
;
; GFX11-LABEL: update_dpp_p3_test:
; GFX11:       ; %bb.0:
; GFX11-NEXT:    s_load_b64 s[0:1], s[4:5], 0x24
; GFX11-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
; GFX11-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
; GFX11-NEXT:    v_add_nc_u32_e32 v0, s0, v0
; GFX11-NEXT:    v_mov_b32_e32 v2, s1
; GFX11-NEXT:    ds_load_b32 v1, v0
; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
; GFX11-NEXT:    v_mov_b32_dpp v2, v1 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX11-NEXT:    ds_store_b32 v0, v2
; GFX11-NEXT:    s_endpgm
  %id = tail call i32 @llvm.amdgcn.workitem.id.x()
  %gep = getelementptr inbounds ptr addrspace(3), ptr addrspace(3) %arg, i32 %id
  %load = load ptr addrspace(3), ptr addrspace(3) %gep
  %tmp0 = call ptr addrspace(3) @llvm.amdgcn.update.dpp.p3(ptr addrspace(3) %in1, ptr addrspace(3) %load, i32 1, i32 1, i32 1, i1 false) #1
  store ptr addrspace(3) %tmp0, ptr addrspace(3) %gep
  ret void
}

define amdgpu_kernel void @update_dpp_p5_test(ptr addrspace(5) %arg, ptr addrspace(5) %in1, ptr %in2) {
; GFX8-LABEL: update_dpp_p5_test:
; GFX8:       ; %bb.0:
; GFX8-NEXT:    s_mov_b32 s88, SCRATCH_RSRC_DWORD0
; GFX8-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX8-NEXT:    s_mov_b32 s89, SCRATCH_RSRC_DWORD1
; GFX8-NEXT:    s_mov_b32 s90, -1
; GFX8-NEXT:    s_mov_b32 s91, 0xe80000
; GFX8-NEXT:    s_add_u32 s88, s88, s11
; GFX8-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX8-NEXT:    s_addc_u32 s89, s89, 0
; GFX8-NEXT:    s_waitcnt lgkmcnt(0)
; GFX8-NEXT:    v_add_u32_e32 v0, vcc, s0, v0
; GFX8-NEXT:    buffer_load_dword v1, v0, s[88:91], 0 offen
; GFX8-NEXT:    v_mov_b32_e32 v2, s1
; GFX8-NEXT:    s_waitcnt vmcnt(0)
; GFX8-NEXT:    s_nop 0
; GFX8-NEXT:    v_mov_b32_dpp v2, v1 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX8-NEXT:    buffer_store_dword v2, v0, s[88:91], 0 offen
; GFX8-NEXT:    s_endpgm
;
; GFX10-LABEL: update_dpp_p5_test:
; GFX10:       ; %bb.0:
; GFX10-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX10-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX10-NEXT:    s_mov_b32 s12, SCRATCH_RSRC_DWORD0
; GFX10-NEXT:    s_mov_b32 s13, SCRATCH_RSRC_DWORD1
; GFX10-NEXT:    s_mov_b32 s14, -1
; GFX10-NEXT:    s_mov_b32 s15, 0x31c16000
; GFX10-NEXT:    s_add_u32 s12, s12, s11
; GFX10-NEXT:    s_addc_u32 s13, s13, 0
; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
; GFX10-NEXT:    v_add_nc_u32_e32 v0, s0, v0
; GFX10-NEXT:    v_mov_b32_e32 v2, s1
; GFX10-NEXT:    buffer_load_dword v1, v0, s[12:15], 0 offen
; GFX10-NEXT:    s_waitcnt vmcnt(0)
; GFX10-NEXT:    v_mov_b32_dpp v2, v1 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX10-NEXT:    buffer_store_dword v2, v0, s[12:15], 0 offen
; GFX10-NEXT:    s_endpgm
;
; GFX11-LABEL: update_dpp_p5_test:
; GFX11:       ; %bb.0:
; GFX11-NEXT:    s_load_b64 s[0:1], s[4:5], 0x24
; GFX11-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
; GFX11-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
; GFX11-NEXT:    v_add_nc_u32_e32 v0, s0, v0
; GFX11-NEXT:    v_mov_b32_e32 v2, s1
; GFX11-NEXT:    scratch_load_b32 v1, v0, off
; GFX11-NEXT:    s_waitcnt vmcnt(0)
; GFX11-NEXT:    v_mov_b32_dpp v2, v1 quad_perm:[1,0,0,0] row_mask:0x1 bank_mask:0x1
; GFX11-NEXT:    scratch_store_b32 v0, v2, off
; GFX11-NEXT:    s_endpgm
  %id = tail call i32 @llvm.amdgcn.workitem.id.x()
  %gep = getelementptr inbounds ptr addrspace(5), ptr addrspace(5) %arg, i32 %id
  %load = load ptr addrspace(5), ptr addrspace(5) %gep
  %tmp0 = call ptr addrspace(5) @llvm.amdgcn.update.dpp.p5(ptr addrspace(5) %in1, ptr addrspace(5) %load, i32 1, i32 1, i32 1, i1 false) #1
  store ptr addrspace(5) %tmp0, ptr addrspace(5) %gep
  ret void
}

declare i32 @llvm.amdgcn.workitem.id.x() #0
declare i32 @llvm.amdgcn.update.dpp.i32(i32, i32, i32 immarg, i32 immarg, i32 immarg, i1 immarg) #1
declare i64 @llvm.amdgcn.update.dpp.i64(i64, i64, i32 immarg, i32 immarg, i32 immarg, i1 immarg) #1

attributes #0 = { nounwind readnone speculatable }
attributes #1 = { convergent nounwind readnone }