; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=amdgpu7.04--amdpal -mattr=+load-store-opt < %s | FileCheck -enable-var-scope --check-prefix=CI %s
; RUN: llc -mtriple=amdgpu9.00--amdpal -mattr=+load-store-opt,-unaligned-access-mode < %s | FileCheck -enable-var-scope -check-prefixes=GFX9,GFX9-ALIGNED %s
; RUN: llc -mtriple=amdgpu9.00--amdpal -mattr=+load-store-opt,+unaligned-access-mode < %s | FileCheck -enable-var-scope -check-prefixes=GFX9,GFX9-UNALIGNED %s
; RUN: llc -mtriple=amdgpu12.50--amdpal -mattr=+load-store-opt,+unaligned-access-mode < %s | FileCheck -enable-var-scope -check-prefixes=GFX1250,GFX1250-UNALIGNED %s
; RUN: llc -mtriple=amdgpu12.50s--amdpal -mattr=+load-store-opt,+unaligned-access-mode < %s | FileCheck -enable-var-scope -check-prefixes=GFX1250,GFX1250S-UNALIGNED %s

@lds = addrspace(3) global [512 x float] poison, align 4
@lds.f64 = addrspace(3) global [512 x double] poison, align 8

define amdgpu_kernel void @simple_write2_one_val_f32(ptr addrspace(1) %C, ptr addrspace(1) %in) #0 {
; CI-LABEL: simple_write2_one_val_f32:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x2
; CI-NEXT:    s_mov_b32 s3, 0xf000
; CI-NEXT:    s_mov_b32 s2, 0
; CI-NEXT:    v_mov_b32_e32 v1, 0
; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    buffer_load_dword v1, v[0:1], s[0:3], 0 addr64
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    s_waitcnt vmcnt(0)
; CI-NEXT:    ds_write2_b32 v0, v1, v1 offset1:8
; CI-NEXT:    s_endpgm
;
; GFX9-LABEL: simple_write2_one_val_f32:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x8
; GFX9-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    global_load_dword v1, v0, s[0:1]
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    ds_write2_b32 v0, v1, v1 offset1:8
; GFX9-NEXT:    s_endpgm
;
; GFX1250-LABEL: simple_write2_one_val_f32:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-NEXT:    v_nop
; GFX1250-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x8 nv
; GFX1250-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX1250-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250-NEXT:    v_and_b32_e32 v0, 0xffc, v0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    global_load_b32 v1, v0, s[0:1]
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    ds_store_2addr_b32 v0, v1, v1 offset1:8
; GFX1250-NEXT:    s_endpgm
  %x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
  %in.gep = getelementptr float, ptr addrspace(1) %in, i32 %x.i
  %val = load float, ptr addrspace(1) %in.gep, align 4
  %arrayidx0 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %x.i
  store float %val, ptr addrspace(3) %arrayidx0, align 4
  %add.x = add nsw i32 %x.i, 8
  %arrayidx1 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %add.x
  store float %val, ptr addrspace(3) %arrayidx1, align 4
  ret void
}

define amdgpu_kernel void @simple_write2_two_val_f32(ptr addrspace(1) %C, ptr addrspace(1) %in) #0 {
; CI-LABEL: simple_write2_two_val_f32:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x2
; CI-NEXT:    s_mov_b32 s3, 0xf000
; CI-NEXT:    s_mov_b32 s2, 0
; CI-NEXT:    v_mov_b32_e32 v1, 0
; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    buffer_load_dword v2, v[0:1], s[0:3], 0 addr64 glc
; CI-NEXT:    s_waitcnt vmcnt(0)
; CI-NEXT:    buffer_load_dword v1, v[0:1], s[0:3], 0 addr64 offset:4 glc
; CI-NEXT:    s_waitcnt vmcnt(0)
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    ds_write2_b32 v0, v2, v1 offset1:8
; CI-NEXT:    s_endpgm
;
; GFX9-LABEL: simple_write2_two_val_f32:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x8
; GFX9-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    global_load_dword v1, v0, s[0:1] glc
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    global_load_dword v2, v0, s[0:1] offset:4 glc
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    ds_write2_b32 v0, v1, v2 offset1:8
; GFX9-NEXT:    s_endpgm
;
; GFX1250-LABEL: simple_write2_two_val_f32:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-NEXT:    v_nop
; GFX1250-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x8 nv
; GFX1250-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX1250-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250-NEXT:    v_and_b32_e32 v0, 0xffc, v0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    global_load_b32 v1, v0, s[0:1] scope:SCOPE_SYS
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    global_load_b32 v2, v0, s[0:1] offset:4 scope:SCOPE_SYS
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    ds_store_2addr_b32 v0, v1, v2 offset1:8
; GFX1250-NEXT:    s_endpgm
  %x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
  %in.gep.0 = getelementptr float, ptr addrspace(1) %in, i32 %x.i
  %in.gep.1 = getelementptr float, ptr addrspace(1) %in.gep.0, i32 1
  %val0 = load volatile float, ptr addrspace(1) %in.gep.0, align 4
  %val1 = load volatile float, ptr addrspace(1) %in.gep.1, align 4
  %arrayidx0 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %x.i
  store float %val0, ptr addrspace(3) %arrayidx0, align 4
  %add.x = add nsw i32 %x.i, 8
  %arrayidx1 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %add.x
  store float %val1, ptr addrspace(3) %arrayidx1, align 4
  ret void
}

define amdgpu_kernel void @simple_write2_two_val_f32_volatile_0(ptr addrspace(1) %C, ptr addrspace(1) %in0, ptr addrspace(1) %in1) #0 {
; CI-LABEL: simple_write2_two_val_f32_volatile_0:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x2
; CI-NEXT:    s_mov_b32 s7, 0xf000
; CI-NEXT:    s_mov_b32 s6, 0
; CI-NEXT:    v_mov_b32_e32 v1, 0
; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    s_mov_b64 s[4:5], s[0:1]
; CI-NEXT:    s_mov_b64 s[0:1], s[2:3]
; CI-NEXT:    s_mov_b64 s[2:3], s[6:7]
; CI-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64 glc
; CI-NEXT:    s_waitcnt vmcnt(0)
; CI-NEXT:    buffer_load_dword v1, v[0:1], s[0:3], 0 addr64 glc
; CI-NEXT:    s_waitcnt vmcnt(0)
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    ds_write_b32 v0, v2
; CI-NEXT:    ds_write_b32 v0, v1 offset:32
; CI-NEXT:    s_endpgm
;
; GFX9-LABEL: simple_write2_two_val_f32_volatile_0:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x8
; GFX9-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    global_load_dword v1, v0, s[0:1] glc
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    global_load_dword v2, v0, s[2:3] glc
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    ds_write_b32 v0, v1
; GFX9-NEXT:    ds_write_b32 v0, v2 offset:32
; GFX9-NEXT:    s_endpgm
;
; GFX1250-LABEL: simple_write2_two_val_f32_volatile_0:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-NEXT:    v_nop
; GFX1250-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x8 nv
; GFX1250-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX1250-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250-NEXT:    v_and_b32_e32 v0, 0xffc, v0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    global_load_b32 v1, v0, s[0:1] scope:SCOPE_SYS
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    global_load_b32 v2, v0, s[2:3] scope:SCOPE_SYS
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    ds_store_b32 v0, v1
; GFX1250-NEXT:    ds_store_b32 v0, v2 offset:32
; GFX1250-NEXT:    s_endpgm
  %x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
  %in0.gep = getelementptr float, ptr addrspace(1) %in0, i32 %x.i
  %in1.gep = getelementptr float, ptr addrspace(1) %in1, i32 %x.i
  %val0 = load volatile float, ptr addrspace(1) %in0.gep, align 4
  %val1 = load volatile float, ptr addrspace(1) %in1.gep, align 4
  %arrayidx0 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %x.i
  store volatile float %val0, ptr addrspace(3) %arrayidx0, align 4
  %add.x = add nsw i32 %x.i, 8
  %arrayidx1 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %add.x
  store float %val1, ptr addrspace(3) %arrayidx1, align 4
  ret void
}

define amdgpu_kernel void @simple_write2_two_val_f32_volatile_1(ptr addrspace(1) %C, ptr addrspace(1) %in0, ptr addrspace(1) %in1) #0 {
; CI-LABEL: simple_write2_two_val_f32_volatile_1:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x2
; CI-NEXT:    s_mov_b32 s7, 0xf000
; CI-NEXT:    s_mov_b32 s6, 0
; CI-NEXT:    v_mov_b32_e32 v1, 0
; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    s_mov_b64 s[4:5], s[0:1]
; CI-NEXT:    s_mov_b64 s[0:1], s[2:3]
; CI-NEXT:    s_mov_b64 s[2:3], s[6:7]
; CI-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64 glc
; CI-NEXT:    s_waitcnt vmcnt(0)
; CI-NEXT:    buffer_load_dword v1, v[0:1], s[0:3], 0 addr64 glc
; CI-NEXT:    s_waitcnt vmcnt(0)
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    ds_write_b32 v0, v2
; CI-NEXT:    ds_write_b32 v0, v1 offset:32
; CI-NEXT:    s_endpgm
;
; GFX9-LABEL: simple_write2_two_val_f32_volatile_1:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x8
; GFX9-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    global_load_dword v1, v0, s[0:1] glc
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    global_load_dword v2, v0, s[2:3] glc
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    ds_write_b32 v0, v1
; GFX9-NEXT:    ds_write_b32 v0, v2 offset:32
; GFX9-NEXT:    s_endpgm
;
; GFX1250-LABEL: simple_write2_two_val_f32_volatile_1:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-NEXT:    v_nop
; GFX1250-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x8 nv
; GFX1250-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX1250-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250-NEXT:    v_and_b32_e32 v0, 0xffc, v0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    global_load_b32 v1, v0, s[0:1] scope:SCOPE_SYS
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    global_load_b32 v2, v0, s[2:3] scope:SCOPE_SYS
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    ds_store_b32 v0, v1
; GFX1250-NEXT:    ds_store_b32 v0, v2 offset:32
; GFX1250-NEXT:    s_endpgm
  %x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
  %in0.gep = getelementptr float, ptr addrspace(1) %in0, i32 %x.i
  %in1.gep = getelementptr float, ptr addrspace(1) %in1, i32 %x.i
  %val0 = load volatile float, ptr addrspace(1) %in0.gep, align 4
  %val1 = load volatile float, ptr addrspace(1) %in1.gep, align 4
  %arrayidx0 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %x.i
  store float %val0, ptr addrspace(3) %arrayidx0, align 4
  %add.x = add nsw i32 %x.i, 8
  %arrayidx1 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %add.x
  store volatile float %val1, ptr addrspace(3) %arrayidx1, align 4
  ret void
}

; 2 data subregisters from different super registers.
; TODO: GFX9 has v_mov_b32_e32 v2, lds@abs32@lo
;       This should be an s_mov_b32. The v_mov_b32 gets introduced by an
;       early legalization of the constant bus constraint on the v_lshl_add_u32,
;       and then SIFoldOperands folds in an unlucky order.
define amdgpu_kernel void @simple_write2_two_val_subreg2_mixed_f32(ptr addrspace(1) %C, ptr addrspace(1) %in) #0 {
; CI-LABEL: simple_write2_two_val_subreg2_mixed_f32:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x2
; CI-NEXT:    s_mov_b32 s3, 0xf000
; CI-NEXT:    s_mov_b32 s2, 0
; CI-NEXT:    v_lshlrev_b32_e32 v1, 3, v0
; CI-NEXT:    v_mov_b32_e32 v2, 0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    buffer_load_dwordx2 v[3:4], v[1:2], s[0:3], 0 addr64 glc
; CI-NEXT:    s_waitcnt vmcnt(0)
; CI-NEXT:    buffer_load_dwordx2 v[1:2], v[1:2], s[0:3], 0 addr64 offset:8 glc
; CI-NEXT:    s_waitcnt vmcnt(0)
; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    ds_write2_b32 v0, v3, v2 offset1:8
; CI-NEXT:    s_endpgm
;
; GFX9-LABEL: simple_write2_two_val_subreg2_mixed_f32:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x8
; GFX9-NEXT:    v_lshlrev_b32_e32 v4, 3, v0
; GFX9-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX9-NEXT:    ; kill: killed $vgpr4
; GFX9-NEXT:    ; kill: killed $sgpr0_sgpr1
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    global_load_dwordx2 v[1:2], v4, s[0:1] glc
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    global_load_dwordx2 v[2:3], v4, s[0:1] offset:8 glc
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    ds_write2_b32 v0, v1, v3 offset1:8
; GFX9-NEXT:    s_endpgm
;
; GFX1250-LABEL: simple_write2_two_val_subreg2_mixed_f32:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-NEXT:    v_nop
; GFX1250-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x8 nv
; GFX1250-NEXT:    v_and_b32_e32 v4, 0x3ff, v0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    global_load_b64 v[0:1], v4, s[0:1] scale_offset scope:SCOPE_SYS
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    global_load_b64 v[2:3], v4, s[0:1] offset:8 scale_offset scope:SCOPE_SYS
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    v_lshlrev_b32_e32 v1, 2, v4
; GFX1250-NEXT:    ds_store_2addr_b32 v1, v0, v3 offset1:8
; GFX1250-NEXT:    s_endpgm
  %x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
  %in.gep.0 = getelementptr <2 x float>, ptr addrspace(1) %in, i32 %x.i
  %in.gep.1 = getelementptr <2 x float>, ptr addrspace(1) %in.gep.0, i32 1
  %val0 = load volatile <2 x float>, ptr addrspace(1) %in.gep.0, align 8
  %val1 = load volatile <2 x float>, ptr addrspace(1) %in.gep.1, align 8
  %val0.0 = extractelement <2 x float> %val0, i32 0
  %val1.1 = extractelement <2 x float> %val1, i32 1
  %arrayidx0 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %x.i
  store float %val0.0, ptr addrspace(3) %arrayidx0, align 4
  %add.x = add nsw i32 %x.i, 8
  %arrayidx1 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %add.x
  store float %val1.1, ptr addrspace(3) %arrayidx1, align 4
  ret void
}

define amdgpu_kernel void @simple_write2_two_val_subreg2_f32(ptr addrspace(1) %C, ptr addrspace(1) %in) #0 {
; CI-LABEL: simple_write2_two_val_subreg2_f32:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x2
; CI-NEXT:    s_mov_b32 s3, 0xf000
; CI-NEXT:    s_mov_b32 s2, 0
; CI-NEXT:    v_lshlrev_b32_e32 v1, 3, v0
; CI-NEXT:    v_mov_b32_e32 v2, 0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    buffer_load_dwordx2 v[1:2], v[1:2], s[0:3], 0 addr64
; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    s_waitcnt vmcnt(0)
; CI-NEXT:    ds_write2_b32 v0, v1, v2 offset1:8
; CI-NEXT:    s_endpgm
;
; GFX9-LABEL: simple_write2_two_val_subreg2_f32:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x8
; GFX9-NEXT:    v_lshlrev_b32_e32 v1, 3, v0
; GFX9-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    global_load_dwordx2 v[1:2], v1, s[0:1]
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    ds_write2_b32 v0, v1, v2 offset1:8
; GFX9-NEXT:    s_endpgm
;
; GFX1250-LABEL: simple_write2_two_val_subreg2_f32:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-NEXT:    v_nop
; GFX1250-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x8 nv
; GFX1250-NEXT:    v_and_b32_e32 v2, 0x3ff, v0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    global_load_b64 v[0:1], v2, s[0:1] scale_offset
; GFX1250-NEXT:    s_wait_xcnt 0x0
; GFX1250-NEXT:    v_lshlrev_b32_e32 v2, 2, v2
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    ds_store_2addr_b32 v2, v0, v1 offset1:8
; GFX1250-NEXT:    s_endpgm
  %x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
  %in.gep = getelementptr <2 x float>, ptr addrspace(1) %in, i32 %x.i
  %val = load <2 x float>, ptr addrspace(1) %in.gep, align 8
  %val0 = extractelement <2 x float> %val, i32 0
  %val1 = extractelement <2 x float> %val, i32 1
  %arrayidx0 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %x.i
  store float %val0, ptr addrspace(3) %arrayidx0, align 4
  %add.x = add nsw i32 %x.i, 8
  %arrayidx1 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %add.x
  store float %val1, ptr addrspace(3) %arrayidx1, align 4
  ret void
}

define amdgpu_kernel void @simple_write2_two_val_subreg4_f32(ptr addrspace(1) %C, ptr addrspace(1) %in) #0 {
; CI-LABEL: simple_write2_two_val_subreg4_f32:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x2
; CI-NEXT:    s_mov_b32 s3, 0xf000
; CI-NEXT:    s_mov_b32 s2, 0
; CI-NEXT:    v_lshlrev_b32_e32 v1, 4, v0
; CI-NEXT:    v_mov_b32_e32 v2, 0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    buffer_load_dwordx4 v[1:4], v[1:2], s[0:3], 0 addr64
; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    s_waitcnt vmcnt(0)
; CI-NEXT:    ds_write2_b32 v0, v1, v4 offset1:8
; CI-NEXT:    s_endpgm
;
; GFX9-LABEL: simple_write2_two_val_subreg4_f32:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x8
; GFX9-NEXT:    v_lshlrev_b32_e32 v1, 4, v0
; GFX9-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    global_load_dwordx4 v[1:4], v1, s[0:1]
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    ds_write2_b32 v0, v1, v4 offset1:8
; GFX9-NEXT:    s_endpgm
;
; GFX1250-LABEL: simple_write2_two_val_subreg4_f32:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-NEXT:    v_nop
; GFX1250-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x8 nv
; GFX1250-NEXT:    v_and_b32_e32 v4, 0x3ff, v0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    global_load_b128 v[0:3], v4, s[0:1] scale_offset
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    v_lshlrev_b32_e32 v1, 2, v4
; GFX1250-NEXT:    ds_store_2addr_b32 v1, v0, v3 offset1:8
; GFX1250-NEXT:    s_endpgm
  %x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
  %in.gep = getelementptr <4 x float>, ptr addrspace(1) %in, i32 %x.i
  %val = load <4 x float>, ptr addrspace(1) %in.gep, align 16
  %val0 = extractelement <4 x float> %val, i32 0
  %val1 = extractelement <4 x float> %val, i32 3
  %arrayidx0 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %x.i
  store float %val0, ptr addrspace(3) %arrayidx0, align 4
  %add.x = add nsw i32 %x.i, 8
  %arrayidx1 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %add.x
  store float %val1, ptr addrspace(3) %arrayidx1, align 4
  ret void
}

define amdgpu_kernel void @simple_write2_two_val_max_offset_f32(ptr addrspace(1) %C, ptr addrspace(1) %in) #0 {
; CI-LABEL: simple_write2_two_val_max_offset_f32:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x2
; CI-NEXT:    s_mov_b32 s3, 0xf000
; CI-NEXT:    s_mov_b32 s2, 0
; CI-NEXT:    v_mov_b32_e32 v1, 0
; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    buffer_load_dword v2, v[0:1], s[0:3], 0 addr64 glc
; CI-NEXT:    s_waitcnt vmcnt(0)
; CI-NEXT:    buffer_load_dword v1, v[0:1], s[0:3], 0 addr64 offset:4 glc
; CI-NEXT:    s_waitcnt vmcnt(0)
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    ds_write2_b32 v0, v2, v1 offset1:255
; CI-NEXT:    s_endpgm
;
; GFX9-LABEL: simple_write2_two_val_max_offset_f32:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x8
; GFX9-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    global_load_dword v1, v0, s[0:1] glc
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    global_load_dword v2, v0, s[0:1] offset:4 glc
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    ds_write2_b32 v0, v1, v2 offset1:255
; GFX9-NEXT:    s_endpgm
;
; GFX1250-LABEL: simple_write2_two_val_max_offset_f32:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-NEXT:    v_nop
; GFX1250-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x8 nv
; GFX1250-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX1250-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250-NEXT:    v_and_b32_e32 v0, 0xffc, v0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    global_load_b32 v1, v0, s[0:1] scope:SCOPE_SYS
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    global_load_b32 v2, v0, s[0:1] offset:4 scope:SCOPE_SYS
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    ds_store_2addr_b32 v0, v1, v2 offset1:255
; GFX1250-NEXT:    s_endpgm
  %x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
  %in.gep.0 = getelementptr float, ptr addrspace(1) %in, i32 %x.i
  %in.gep.1 = getelementptr float, ptr addrspace(1) %in.gep.0, i32 1
  %val0 = load volatile float, ptr addrspace(1) %in.gep.0, align 4
  %val1 = load volatile float, ptr addrspace(1) %in.gep.1, align 4
  %arrayidx0 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %x.i
  store float %val0, ptr addrspace(3) %arrayidx0, align 4
  %add.x = add nsw i32 %x.i, 255
  %arrayidx1 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %add.x
  store float %val1, ptr addrspace(3) %arrayidx1, align 4
  ret void
}

define amdgpu_kernel void @simple_write2_two_val_too_far_f32(ptr addrspace(1) %C, ptr addrspace(1) %in0, ptr addrspace(1) %in1) #0 {
; CI-LABEL: simple_write2_two_val_too_far_f32:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x2
; CI-NEXT:    s_mov_b32 s7, 0xf000
; CI-NEXT:    s_mov_b32 s6, 0
; CI-NEXT:    v_mov_b32_e32 v1, 0
; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    s_mov_b64 s[4:5], s[0:1]
; CI-NEXT:    s_mov_b64 s[0:1], s[2:3]
; CI-NEXT:    s_mov_b64 s[2:3], s[6:7]
; CI-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
; CI-NEXT:    buffer_load_dword v1, v[0:1], s[0:3], 0 addr64
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    s_waitcnt vmcnt(1)
; CI-NEXT:    ds_write_b32 v0, v2
; CI-NEXT:    s_waitcnt vmcnt(0)
; CI-NEXT:    ds_write_b32 v0, v1 offset:1028
; CI-NEXT:    s_endpgm
;
; GFX9-LABEL: simple_write2_two_val_too_far_f32:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x8
; GFX9-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    global_load_dword v1, v0, s[0:1]
; GFX9-NEXT:    global_load_dword v2, v0, s[2:3]
; GFX9-NEXT:    s_waitcnt vmcnt(1)
; GFX9-NEXT:    ds_write_b32 v0, v1
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    ds_write_b32 v0, v2 offset:1028
; GFX9-NEXT:    s_endpgm
;
; GFX1250-LABEL: simple_write2_two_val_too_far_f32:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-NEXT:    v_nop
; GFX1250-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x8 nv
; GFX1250-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX1250-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250-NEXT:    v_and_b32_e32 v0, 0xffc, v0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_clause 0x1
; GFX1250-NEXT:    global_load_b32 v1, v0, s[0:1]
; GFX1250-NEXT:    global_load_b32 v2, v0, s[2:3]
; GFX1250-NEXT:    s_wait_loadcnt 0x1
; GFX1250-NEXT:    ds_store_b32 v0, v1
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    ds_store_b32 v0, v2 offset:1028
; GFX1250-NEXT:    s_endpgm
  %x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
  %in0.gep = getelementptr float, ptr addrspace(1) %in0, i32 %x.i
  %in1.gep = getelementptr float, ptr addrspace(1) %in1, i32 %x.i
  %val0 = load float, ptr addrspace(1) %in0.gep, align 4
  %val1 = load float, ptr addrspace(1) %in1.gep, align 4
  %arrayidx0 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %x.i
  store float %val0, ptr addrspace(3) %arrayidx0, align 4
  %add.x = add nsw i32 %x.i, 257
  %arrayidx1 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %add.x
  store float %val1, ptr addrspace(3) %arrayidx1, align 4
  ret void
}

define amdgpu_kernel void @simple_write2_two_val_f32_x2(ptr addrspace(1) %C, ptr addrspace(1) %in0, ptr addrspace(1) %in1) #0 {
; CI-LABEL: simple_write2_two_val_f32_x2:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x2
; CI-NEXT:    s_mov_b32 s7, 0xf000
; CI-NEXT:    s_mov_b32 s6, 0
; CI-NEXT:    v_mov_b32_e32 v1, 0
; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    s_mov_b64 s[4:5], s[0:1]
; CI-NEXT:    s_mov_b64 s[0:1], s[2:3]
; CI-NEXT:    s_mov_b64 s[2:3], s[6:7]
; CI-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
; CI-NEXT:    buffer_load_dword v1, v[0:1], s[0:3], 0 addr64
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    s_waitcnt vmcnt(0)
; CI-NEXT:    ds_write2_b32 v0, v2, v1 offset1:8
; CI-NEXT:    ds_write2_b32 v0, v2, v1 offset0:11 offset1:27
; CI-NEXT:    s_endpgm
;
; GFX9-LABEL: simple_write2_two_val_f32_x2:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x8
; GFX9-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    global_load_dword v1, v0, s[0:1]
; GFX9-NEXT:    global_load_dword v2, v0, s[2:3]
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    ds_write2_b32 v0, v1, v2 offset1:8
; GFX9-NEXT:    ds_write2_b32 v0, v1, v2 offset0:11 offset1:27
; GFX9-NEXT:    s_endpgm
;
; GFX1250-LABEL: simple_write2_two_val_f32_x2:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-NEXT:    v_nop
; GFX1250-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x8 nv
; GFX1250-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX1250-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250-NEXT:    v_and_b32_e32 v0, 0xffc, v0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_clause 0x1
; GFX1250-NEXT:    global_load_b32 v1, v0, s[0:1]
; GFX1250-NEXT:    global_load_b32 v2, v0, s[2:3]
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    ds_store_2addr_b32 v0, v1, v2 offset1:8
; GFX1250-NEXT:    ds_store_2addr_b32 v0, v1, v2 offset0:11 offset1:27
; GFX1250-NEXT:    s_endpgm
  %tid.x = tail call i32 @llvm.amdgcn.workitem.id.x() #1
  %in0.gep = getelementptr float, ptr addrspace(1) %in0, i32 %tid.x
  %in1.gep = getelementptr float, ptr addrspace(1) %in1, i32 %tid.x
  %val0 = load float, ptr addrspace(1) %in0.gep, align 4
  %val1 = load float, ptr addrspace(1) %in1.gep, align 4

  %idx.0 = add nsw i32 %tid.x, 0
  %arrayidx0 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %idx.0
  store float %val0, ptr addrspace(3) %arrayidx0, align 4

  %idx.1 = add nsw i32 %tid.x, 8
  %arrayidx1 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %idx.1
  store float %val1, ptr addrspace(3) %arrayidx1, align 4

  %idx.2 = add nsw i32 %tid.x, 11
  %arrayidx2 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %idx.2
  store float %val0, ptr addrspace(3) %arrayidx2, align 4

  %idx.3 = add nsw i32 %tid.x, 27
  %arrayidx3 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %idx.3
  store float %val1, ptr addrspace(3) %arrayidx3, align 4

  ret void
}

define amdgpu_kernel void @simple_write2_two_val_f32_x2_nonzero_base(ptr addrspace(1) %C, ptr addrspace(1) %in0, ptr addrspace(1) %in1) #0 {
; CI-LABEL: simple_write2_two_val_f32_x2_nonzero_base:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x2
; CI-NEXT:    s_mov_b32 s7, 0xf000
; CI-NEXT:    s_mov_b32 s6, 0
; CI-NEXT:    v_mov_b32_e32 v1, 0
; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    s_mov_b64 s[4:5], s[0:1]
; CI-NEXT:    s_mov_b64 s[0:1], s[2:3]
; CI-NEXT:    s_mov_b64 s[2:3], s[6:7]
; CI-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
; CI-NEXT:    buffer_load_dword v1, v[0:1], s[0:3], 0 addr64
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    s_waitcnt vmcnt(0)
; CI-NEXT:    ds_write2_b32 v0, v2, v1 offset0:3 offset1:8
; CI-NEXT:    ds_write2_b32 v0, v2, v1 offset0:11 offset1:27
; CI-NEXT:    s_endpgm
;
; GFX9-LABEL: simple_write2_two_val_f32_x2_nonzero_base:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x8
; GFX9-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    global_load_dword v1, v0, s[0:1]
; GFX9-NEXT:    global_load_dword v2, v0, s[2:3]
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    ds_write2_b32 v0, v1, v2 offset0:3 offset1:8
; GFX9-NEXT:    ds_write2_b32 v0, v1, v2 offset0:11 offset1:27
; GFX9-NEXT:    s_endpgm
;
; GFX1250-LABEL: simple_write2_two_val_f32_x2_nonzero_base:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-NEXT:    v_nop
; GFX1250-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x8 nv
; GFX1250-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX1250-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250-NEXT:    v_and_b32_e32 v0, 0xffc, v0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_clause 0x1
; GFX1250-NEXT:    global_load_b32 v1, v0, s[0:1]
; GFX1250-NEXT:    global_load_b32 v2, v0, s[2:3]
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    ds_store_2addr_b32 v0, v1, v2 offset0:3 offset1:8
; GFX1250-NEXT:    ds_store_2addr_b32 v0, v1, v2 offset0:11 offset1:27
; GFX1250-NEXT:    s_endpgm
  %tid.x = tail call i32 @llvm.amdgcn.workitem.id.x() #1
  %in0.gep = getelementptr float, ptr addrspace(1) %in0, i32 %tid.x
  %in1.gep = getelementptr float, ptr addrspace(1) %in1, i32 %tid.x
  %val0 = load float, ptr addrspace(1) %in0.gep, align 4
  %val1 = load float, ptr addrspace(1) %in1.gep, align 4

  %idx.0 = add nsw i32 %tid.x, 3
  %arrayidx0 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %idx.0
  store float %val0, ptr addrspace(3) %arrayidx0, align 4

  %idx.1 = add nsw i32 %tid.x, 8
  %arrayidx1 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %idx.1
  store float %val1, ptr addrspace(3) %arrayidx1, align 4

  %idx.2 = add nsw i32 %tid.x, 11
  %arrayidx2 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %idx.2
  store float %val0, ptr addrspace(3) %arrayidx2, align 4

  %idx.3 = add nsw i32 %tid.x, 27
  %arrayidx3 = getelementptr inbounds [512 x float], ptr addrspace(3) @lds, i32 0, i32 %idx.3
  store float %val1, ptr addrspace(3) %arrayidx3, align 4

  ret void
}

define amdgpu_kernel void @write2_ptr_subreg_arg_two_val_f32(ptr addrspace(1) %C, ptr addrspace(1) %in0, ptr addrspace(1) %in1, <2 x ptr addrspace(3)> %lds.ptr) #0 {
; CI-LABEL: write2_ptr_subreg_arg_two_val_f32:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x2
; CI-NEXT:    s_load_dwordx2 s[8:9], s[4:5], 0x6
; CI-NEXT:    s_mov_b32 s7, 0xf000
; CI-NEXT:    s_mov_b32 s6, 0
; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    s_mov_b64 s[4:5], s[0:1]
; CI-NEXT:    v_mov_b32_e32 v1, 0
; CI-NEXT:    s_mov_b64 s[0:1], s[2:3]
; CI-NEXT:    s_mov_b64 s[2:3], s[6:7]
; CI-NEXT:    buffer_load_dword v2, v[0:1], s[4:7], 0 addr64
; CI-NEXT:    buffer_load_dword v0, v[0:1], s[0:3], 0 addr64
; CI-NEXT:    v_mov_b32_e32 v1, s8
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    v_mov_b32_e32 v3, s9
; CI-NEXT:    s_waitcnt vmcnt(1)
; CI-NEXT:    ds_write_b32 v1, v2 offset:32
; CI-NEXT:    s_waitcnt vmcnt(0)
; CI-NEXT:    ds_write_b32 v3, v0 offset:32
; CI-NEXT:    s_endpgm
;
; GFX9-LABEL: write2_ptr_subreg_arg_two_val_f32:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x8
; GFX9-NEXT:    s_load_dwordx2 s[6:7], s[4:5], 0x18
; GFX9-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    global_load_dword v1, v0, s[0:1]
; GFX9-NEXT:    global_load_dword v2, v0, s[2:3]
; GFX9-NEXT:    v_mov_b32_e32 v0, s6
; GFX9-NEXT:    v_mov_b32_e32 v3, s7
; GFX9-NEXT:    s_waitcnt vmcnt(1)
; GFX9-NEXT:    ds_write_b32 v0, v1 offset:32
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    ds_write_b32 v3, v2 offset:32
; GFX9-NEXT:    s_endpgm
;
; GFX1250-LABEL: write2_ptr_subreg_arg_two_val_f32:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-NEXT:    v_nop
; GFX1250-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x8 nv
; GFX1250-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
; GFX1250-NEXT:    s_wait_xcnt 0x0
; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x18 nv
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_clause 0x1
; GFX1250-NEXT:    global_load_b32 v1, v0, s[0:1] scale_offset
; GFX1250-NEXT:    global_load_b32 v2, v0, s[2:3] scale_offset
; GFX1250-NEXT:    s_wait_xcnt 0x0
; GFX1250-NEXT:    v_dual_mov_b32 v0, s4 :: v_dual_mov_b32 v3, s5
; GFX1250-NEXT:    s_wait_loadcnt 0x1
; GFX1250-NEXT:    ds_store_b32 v0, v1 offset:32
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    ds_store_b32 v3, v2 offset:32
; GFX1250-NEXT:    s_endpgm
  %x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
  %in0.gep = getelementptr float, ptr addrspace(1) %in0, i32 %x.i
  %in1.gep = getelementptr float, ptr addrspace(1) %in1, i32 %x.i
  %val0 = load float, ptr addrspace(1) %in0.gep, align 4
  %val1 = load float, ptr addrspace(1) %in1.gep, align 4

  %index.0 = insertelement <2 x i32> poison, i32 %x.i, i32 0
  %index.1 = insertelement <2 x i32> %index.0, i32 8, i32 0
  %gep = getelementptr inbounds float, <2 x ptr addrspace(3)> %lds.ptr, <2 x i32> %index.1
  %gep.0 = extractelement <2 x ptr addrspace(3)> %gep, i32 0
  %gep.1 = extractelement <2 x ptr addrspace(3)> %gep, i32 1

  ; Apply an additional offset after the vector that will be more obviously folded.
  %gep.1.offset = getelementptr float, ptr addrspace(3) %gep.1, i32 8
  store float %val0, ptr addrspace(3) %gep.0, align 4

  %add.x = add nsw i32 %x.i, 8
  store float %val1, ptr addrspace(3) %gep.1.offset, align 4
  ret void
}

define amdgpu_kernel void @simple_write2_one_val_f64(ptr addrspace(1) %C, ptr addrspace(1) %in) #0 {
; CI-LABEL: simple_write2_one_val_f64:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x2
; CI-NEXT:    s_mov_b32 s3, 0xf000
; CI-NEXT:    s_mov_b32 s2, 0
; CI-NEXT:    v_mov_b32_e32 v1, 0
; CI-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    buffer_load_dwordx2 v[1:2], v[0:1], s[0:3], 0 addr64
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    s_waitcnt vmcnt(0)
; CI-NEXT:    ds_write2_b64 v0, v[1:2], v[1:2] offset1:8
; CI-NEXT:    s_endpgm
;
; GFX9-LABEL: simple_write2_one_val_f64:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x8
; GFX9-NEXT:    v_lshlrev_b32_e32 v2, 3, v0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    global_load_dwordx2 v[0:1], v2, s[0:1]
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    ds_write2_b64 v2, v[0:1], v[0:1] offset1:8
; GFX9-NEXT:    s_endpgm
;
; GFX1250-LABEL: simple_write2_one_val_f64:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-NEXT:    v_nop
; GFX1250-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x8 nv
; GFX1250-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
; GFX1250-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250-NEXT:    v_and_b32_e32 v2, 0x1ff8, v0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    global_load_b64 v[0:1], v2, s[0:1]
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    ds_store_2addr_b64 v2, v[0:1], v[0:1] offset1:8
; GFX1250-NEXT:    s_endpgm
  %x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
  %in.gep = getelementptr double, ptr addrspace(1) %in, i32 %x.i
  %val = load double, ptr addrspace(1) %in.gep, align 8
  %arrayidx0 = getelementptr inbounds [512 x double], ptr addrspace(3) @lds.f64, i32 0, i32 %x.i
  store double %val, ptr addrspace(3) %arrayidx0, align 8
  %add.x = add nsw i32 %x.i, 8
  %arrayidx1 = getelementptr inbounds [512 x double], ptr addrspace(3) @lds.f64, i32 0, i32 %add.x
  store double %val, ptr addrspace(3) %arrayidx1, align 8
  ret void
}

define amdgpu_kernel void @misaligned_simple_write2_one_val_f64(ptr addrspace(1) %C, ptr addrspace(1) %in, ptr addrspace(3) %lds) #0 {
; CI-LABEL: misaligned_simple_write2_one_val_f64:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x2
; CI-NEXT:    s_load_dword s4, s[4:5], 0x4
; CI-NEXT:    s_mov_b32 s3, 0xf000
; CI-NEXT:    s_mov_b32 s2, 0
; CI-NEXT:    v_mov_b32_e32 v1, 0
; CI-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    buffer_load_dwordx2 v[1:2], v[0:1], s[0:3], 0 addr64
; CI-NEXT:    v_add_i32_e32 v0, vcc, s4, v0
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    s_waitcnt vmcnt(0)
; CI-NEXT:    ds_write2_b32 v0, v1, v2 offset1:1
; CI-NEXT:    ds_write2_b32 v0, v1, v2 offset0:14 offset1:15
; CI-NEXT:    s_endpgm
;
; GFX9-LABEL: misaligned_simple_write2_one_val_f64:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x8
; GFX9-NEXT:    s_load_dword s2, s[4:5], 0x10
; GFX9-NEXT:    v_lshlrev_b32_e32 v2, 3, v0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    global_load_dwordx2 v[0:1], v2, s[0:1]
; GFX9-NEXT:    v_add_u32_e32 v2, s2, v2
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    ds_write2_b32 v2, v0, v1 offset1:1
; GFX9-NEXT:    ds_write2_b32 v2, v0, v1 offset0:14 offset1:15
; GFX9-NEXT:    s_endpgm
;
; GFX1250-UNALIGNED-LABEL: misaligned_simple_write2_one_val_f64:
; GFX1250-UNALIGNED:       ; %bb.0:
; GFX1250-UNALIGNED-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-UNALIGNED-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-UNALIGNED-NEXT:    v_nop
; GFX1250-UNALIGNED-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-UNALIGNED-NEXT:    s_load_b96 s[0:2], s[4:5], 0x8 nv
; GFX1250-UNALIGNED-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
; GFX1250-UNALIGNED-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250-UNALIGNED-NEXT:    v_and_b32_e32 v2, 0x1ff8, v0
; GFX1250-UNALIGNED-NEXT:    s_wait_kmcnt 0x0
; GFX1250-UNALIGNED-NEXT:    global_load_b64 v[0:1], v2, s[0:1]
; GFX1250-UNALIGNED-NEXT:    s_wait_xcnt 0x0
; GFX1250-UNALIGNED-NEXT:    v_add_nc_u32_e32 v2, s2, v2
; GFX1250-UNALIGNED-NEXT:    s_wait_loadcnt 0x0
; GFX1250-UNALIGNED-NEXT:    ds_store_2addr_b32 v2, v0, v1 offset1:1
; GFX1250-UNALIGNED-NEXT:    ds_store_2addr_b32 v2, v0, v1 offset0:14 offset1:15
; GFX1250-UNALIGNED-NEXT:    s_endpgm
;
; GFX1250S-UNALIGNED-LABEL: misaligned_simple_write2_one_val_f64:
; GFX1250S-UNALIGNED:       ; %bb.0:
; GFX1250S-UNALIGNED-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250S-UNALIGNED-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250S-UNALIGNED-NEXT:    v_nop
; GFX1250S-UNALIGNED-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250S-UNALIGNED-NEXT:    s_load_b96 s[0:2], s[4:5], 0x8 nv
; GFX1250S-UNALIGNED-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
; GFX1250S-UNALIGNED-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250S-UNALIGNED-NEXT:    v_and_b32_e32 v2, 0x1ff8, v0
; GFX1250S-UNALIGNED-NEXT:    s_wait_kmcnt 0x0
; GFX1250S-UNALIGNED-NEXT:    global_load_b64 v[0:1], v2, s[0:1]
; GFX1250S-UNALIGNED-NEXT:    s_wait_xcnt 0x0
; GFX1250S-UNALIGNED-NEXT:    v_add_nc_u32_e32 v2, s2, v2
; GFX1250S-UNALIGNED-NEXT:    s_wait_loadcnt 0x0
; GFX1250S-UNALIGNED-NEXT:    ds_store_b64 v2, v[0:1]
; GFX1250S-UNALIGNED-NEXT:    ds_store_b64 v2, v[0:1] offset:56
; GFX1250S-UNALIGNED-NEXT:    s_endpgm
  %x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
  %in.gep = getelementptr double, ptr addrspace(1) %in, i32 %x.i
  %val = load double, ptr addrspace(1) %in.gep, align 8
  %arrayidx0 = getelementptr inbounds double, ptr addrspace(3) %lds, i32 %x.i
  store double %val, ptr addrspace(3) %arrayidx0, align 4
  %add.x = add nsw i32 %x.i, 7
  %arrayidx1 = getelementptr inbounds double, ptr addrspace(3) %lds, i32 %add.x
  store double %val, ptr addrspace(3) %arrayidx1, align 4
  ret void
}

define amdgpu_kernel void @unaligned_offset_simple_write2_one_val_f64(ptr addrspace(1) %C, ptr addrspace(1) %in, ptr addrspace(3) %lds) #0 {
; CI-LABEL: unaligned_offset_simple_write2_one_val_f64:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x2
; CI-NEXT:    s_load_dword s4, s[4:5], 0x4
; CI-NEXT:    s_mov_b32 s3, 0xf000
; CI-NEXT:    s_mov_b32 s2, 0
; CI-NEXT:    v_mov_b32_e32 v1, 0
; CI-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    buffer_load_dwordx2 v[1:2], v[0:1], s[0:3], 0 addr64
; CI-NEXT:    v_add_i32_e32 v0, vcc, s4, v0
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    s_waitcnt vmcnt(0)
; CI-NEXT:    ds_write_b8 v0, v1 offset:9
; CI-NEXT:    ds_write_b8 v0, v2 offset:13
; CI-NEXT:    v_lshrrev_b32_e32 v3, 24, v1
; CI-NEXT:    ds_write_b8 v0, v1 offset:5
; CI-NEXT:    v_lshrrev_b32_e32 v4, 16, v1
; CI-NEXT:    v_lshrrev_b32_e32 v5, 8, v1
; CI-NEXT:    v_lshrrev_b32_e32 v1, 24, v2
; CI-NEXT:    v_lshrrev_b32_e32 v6, 16, v2
; CI-NEXT:    v_lshrrev_b32_e32 v2, 8, v2
; CI-NEXT:    ds_write_b8 v0, v3 offset:8
; CI-NEXT:    ds_write_b8 v0, v4 offset:7
; CI-NEXT:    ds_write_b8 v0, v5 offset:6
; CI-NEXT:    ds_write_b8 v0, v3 offset:12
; CI-NEXT:    ds_write_b8 v0, v4 offset:11
; CI-NEXT:    ds_write_b8 v0, v5 offset:10
; CI-NEXT:    ds_write_b8 v0, v1 offset:16
; CI-NEXT:    ds_write_b8 v0, v6 offset:15
; CI-NEXT:    ds_write_b8 v0, v2 offset:14
; CI-NEXT:    s_endpgm
;
; GFX9-ALIGNED-LABEL: unaligned_offset_simple_write2_one_val_f64:
; GFX9-ALIGNED:       ; %bb.0:
; GFX9-ALIGNED-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x8
; GFX9-ALIGNED-NEXT:    s_load_dword s2, s[4:5], 0x10
; GFX9-ALIGNED-NEXT:    v_lshlrev_b32_e32 v2, 3, v0
; GFX9-ALIGNED-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-ALIGNED-NEXT:    global_load_dwordx2 v[0:1], v2, s[0:1]
; GFX9-ALIGNED-NEXT:    v_add_u32_e32 v2, s2, v2
; GFX9-ALIGNED-NEXT:    s_waitcnt vmcnt(0)
; GFX9-ALIGNED-NEXT:    ds_write_b8_d16_hi v2, v0 offset:7
; GFX9-ALIGNED-NEXT:    ds_write_b8 v2, v0 offset:5
; GFX9-ALIGNED-NEXT:    v_lshrrev_b32_e32 v3, 24, v0
; GFX9-ALIGNED-NEXT:    v_lshrrev_b32_e32 v4, 8, v0
; GFX9-ALIGNED-NEXT:    ds_write_b8_d16_hi v2, v0 offset:11
; GFX9-ALIGNED-NEXT:    ds_write_b8 v2, v0 offset:9
; GFX9-ALIGNED-NEXT:    ds_write_b8_d16_hi v2, v1 offset:15
; GFX9-ALIGNED-NEXT:    ds_write_b8 v2, v1 offset:13
; GFX9-ALIGNED-NEXT:    v_lshrrev_b32_e32 v0, 24, v1
; GFX9-ALIGNED-NEXT:    v_lshrrev_b32_e32 v1, 8, v1
; GFX9-ALIGNED-NEXT:    ds_write_b8 v2, v3 offset:8
; GFX9-ALIGNED-NEXT:    ds_write_b8 v2, v4 offset:6
; GFX9-ALIGNED-NEXT:    ds_write_b8 v2, v3 offset:12
; GFX9-ALIGNED-NEXT:    ds_write_b8 v2, v4 offset:10
; GFX9-ALIGNED-NEXT:    ds_write_b8 v2, v0 offset:16
; GFX9-ALIGNED-NEXT:    ds_write_b8 v2, v1 offset:14
; GFX9-ALIGNED-NEXT:    s_endpgm
;
; GFX9-UNALIGNED-LABEL: unaligned_offset_simple_write2_one_val_f64:
; GFX9-UNALIGNED:       ; %bb.0:
; GFX9-UNALIGNED-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x8
; GFX9-UNALIGNED-NEXT:    s_load_dword s2, s[4:5], 0x10
; GFX9-UNALIGNED-NEXT:    v_lshlrev_b32_e32 v2, 3, v0
; GFX9-UNALIGNED-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-UNALIGNED-NEXT:    global_load_dwordx2 v[0:1], v2, s[0:1]
; GFX9-UNALIGNED-NEXT:    v_add_u32_e32 v2, s2, v2
; GFX9-UNALIGNED-NEXT:    s_waitcnt vmcnt(0)
; GFX9-UNALIGNED-NEXT:    ds_write_b64 v2, v[0:1] offset:5
; GFX9-UNALIGNED-NEXT:    ds_write_b64 v2, v[0:1] offset:9
; GFX9-UNALIGNED-NEXT:    s_endpgm
;
; GFX1250-LABEL: unaligned_offset_simple_write2_one_val_f64:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-NEXT:    v_nop
; GFX1250-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-NEXT:    s_load_b96 s[0:2], s[4:5], 0x8 nv
; GFX1250-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
; GFX1250-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250-NEXT:    v_and_b32_e32 v2, 0x1ff8, v0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    global_load_b64 v[0:1], v2, s[0:1]
; GFX1250-NEXT:    s_wait_xcnt 0x0
; GFX1250-NEXT:    v_add_nc_u32_e32 v2, s2, v2
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    ds_store_b64 v2, v[0:1] offset:5
; GFX1250-NEXT:    ds_store_b64 v2, v[0:1] offset:9
; GFX1250-NEXT:    s_endpgm
  %x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
  %in.gep = getelementptr double, ptr addrspace(1) %in, i32 %x.i
  %val = load double, ptr addrspace(1) %in.gep, align 8
  %base = getelementptr inbounds double, ptr addrspace(3) %lds, i32 %x.i
  %addr0.i8 = getelementptr inbounds i8, ptr addrspace(3) %base, i32 5
  store double %val, ptr addrspace(3) %addr0.i8, align 1
  %addr1.i8 = getelementptr inbounds i8, ptr addrspace(3) %base, i32 9
  store double %val, ptr addrspace(3) %addr1.i8, align 1
  ret void
}

define amdgpu_kernel void @simple_write2_two_val_f64(ptr addrspace(1) %C, ptr addrspace(1) %in) #0 {
; CI-LABEL: simple_write2_two_val_f64:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x2
; CI-NEXT:    s_mov_b32 s3, 0xf000
; CI-NEXT:    s_mov_b32 s2, 0
; CI-NEXT:    v_mov_b32_e32 v1, 0
; CI-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    buffer_load_dwordx2 v[2:3], v[0:1], s[0:3], 0 addr64 glc
; CI-NEXT:    s_waitcnt vmcnt(0)
; CI-NEXT:    buffer_load_dwordx2 v[4:5], v[0:1], s[0:3], 0 addr64 offset:8 glc
; CI-NEXT:    s_waitcnt vmcnt(0)
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    ds_write2_b64 v0, v[2:3], v[4:5] offset1:8
; CI-NEXT:    s_endpgm
;
; GFX9-LABEL: simple_write2_two_val_f64:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x8
; GFX9-NEXT:    v_lshlrev_b32_e32 v4, 3, v0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    global_load_dwordx2 v[0:1], v4, s[0:1] glc
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    global_load_dwordx2 v[2:3], v4, s[0:1] offset:8 glc
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    ds_write2_b64 v4, v[0:1], v[2:3] offset1:8
; GFX9-NEXT:    s_endpgm
;
; GFX1250-LABEL: simple_write2_two_val_f64:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-NEXT:    v_nop
; GFX1250-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x8 nv
; GFX1250-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
; GFX1250-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250-NEXT:    v_and_b32_e32 v4, 0x1ff8, v0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    global_load_b64 v[0:1], v4, s[0:1] scope:SCOPE_SYS
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    global_load_b64 v[2:3], v4, s[0:1] offset:8 scope:SCOPE_SYS
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    ds_store_2addr_b64 v4, v[0:1], v[2:3] offset1:8
; GFX1250-NEXT:    s_endpgm
  %x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
  %in.gep.0 = getelementptr double, ptr addrspace(1) %in, i32 %x.i
  %in.gep.1 = getelementptr double, ptr addrspace(1) %in.gep.0, i32 1
  %val0 = load volatile double, ptr addrspace(1) %in.gep.0, align 8
  %val1 = load volatile double, ptr addrspace(1) %in.gep.1, align 8
  %arrayidx0 = getelementptr inbounds [512 x double], ptr addrspace(3) @lds.f64, i32 0, i32 %x.i
  store double %val0, ptr addrspace(3) %arrayidx0, align 8
  %add.x = add nsw i32 %x.i, 8
  %arrayidx1 = getelementptr inbounds [512 x double], ptr addrspace(3) @lds.f64, i32 0, i32 %add.x
  store double %val1, ptr addrspace(3) %arrayidx1, align 8
  ret void
}

@foo = addrspace(3) global [4 x i32] poison, align 4

define amdgpu_kernel void @store_constant_adjacent_offsets() {
; CI-LABEL: store_constant_adjacent_offsets:
; CI:       ; %bb.0:
; CI-NEXT:    v_mov_b32_e32 v0, 0x7b
; CI-NEXT:    v_mov_b32_e32 v1, v0
; CI-NEXT:    v_mov_b32_e32 v2, 0
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    ds_write_b64 v2, v[0:1]
; CI-NEXT:    s_endpgm
;
; GFX9-LABEL: store_constant_adjacent_offsets:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    v_mov_b32_e32 v0, 0x7b
; GFX9-NEXT:    v_mov_b32_e32 v1, v0
; GFX9-NEXT:    v_mov_b32_e32 v2, 0
; GFX9-NEXT:    ds_write_b64 v2, v[0:1]
; GFX9-NEXT:    s_endpgm
;
; GFX1250-LABEL: store_constant_adjacent_offsets:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-NEXT:    v_nop
; GFX1250-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-NEXT:    v_mov_b64_e32 v[0:1], 0x7b0000007b
; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
; GFX1250-NEXT:    ds_store_b64 v2, v[0:1]
; GFX1250-NEXT:    s_endpgm
  store i32 123, ptr addrspace(3) @foo, align 4
  store i32 123, ptr addrspace(3) getelementptr inbounds ([4 x i32], ptr addrspace(3) @foo, i32 0, i32 1), align 4
  ret void
}

define amdgpu_kernel void @store_constant_disjoint_offsets() {
; CI-LABEL: store_constant_disjoint_offsets:
; CI:       ; %bb.0:
; CI-NEXT:    v_mov_b32_e32 v0, 0x7b
; CI-NEXT:    v_mov_b32_e32 v1, 0
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    ds_write2_b32 v1, v0, v0 offset1:2
; CI-NEXT:    s_endpgm
;
; GFX9-LABEL: store_constant_disjoint_offsets:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    v_mov_b32_e32 v0, 0x7b
; GFX9-NEXT:    v_mov_b32_e32 v1, 0
; GFX9-NEXT:    ds_write2_b32 v1, v0, v0 offset1:2
; GFX9-NEXT:    s_endpgm
;
; GFX1250-LABEL: store_constant_disjoint_offsets:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-NEXT:    v_nop
; GFX1250-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-NEXT:    v_dual_mov_b32 v0, 0x7b :: v_dual_mov_b32 v1, 0
; GFX1250-NEXT:    ds_store_2addr_b32 v1, v0, v0 offset1:2
; GFX1250-NEXT:    s_endpgm
  store i32 123, ptr addrspace(3) @foo, align 4
  store i32 123, ptr addrspace(3) getelementptr inbounds ([4 x i32], ptr addrspace(3) @foo, i32 0, i32 2), align 4
  ret void
}

@bar = addrspace(3) global [4 x i64] poison, align 4

define amdgpu_kernel void @store_misaligned64_constant_offsets() {
; CI-LABEL: store_misaligned64_constant_offsets:
; CI:       ; %bb.0:
; CI-NEXT:    v_mov_b32_e32 v0, 0x7b
; CI-NEXT:    v_mov_b32_e32 v1, 0
; CI-NEXT:    v_mov_b32_e32 v2, v0
; CI-NEXT:    v_mov_b32_e32 v3, v1
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    ds_write_b128 v1, v[0:3]
; CI-NEXT:    s_endpgm
;
; GFX9-LABEL: store_misaligned64_constant_offsets:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    v_mov_b32_e32 v0, 0x7b
; GFX9-NEXT:    v_mov_b32_e32 v1, 0
; GFX9-NEXT:    v_mov_b32_e32 v2, v0
; GFX9-NEXT:    v_mov_b32_e32 v3, v1
; GFX9-NEXT:    ds_write_b128 v1, v[0:3]
; GFX9-NEXT:    s_endpgm
;
; GFX1250-LABEL: store_misaligned64_constant_offsets:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-NEXT:    v_nop
; GFX1250-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-NEXT:    v_dual_mov_b32 v0, 0x7b :: v_dual_mov_b32 v1, 0
; GFX1250-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250-NEXT:    v_dual_mov_b32 v2, v0 :: v_dual_mov_b32 v3, v1
; GFX1250-NEXT:    ds_store_b128 v1, v[0:3]
; GFX1250-NEXT:    s_endpgm
  store i64 123, ptr addrspace(3) @bar, align 4
  store i64 123, ptr addrspace(3) getelementptr inbounds ([4 x i64], ptr addrspace(3) @bar, i32 0, i32 1), align 4
  ret void
}

@bar.large = addrspace(3) global [4096 x i64] poison, align 4

define amdgpu_kernel void @store_misaligned64_constant_large_offsets() {
; CI-LABEL: store_misaligned64_constant_large_offsets:
; CI:       ; %bb.0:
; CI-NEXT:    s_mov_b64 s[0:1], 0x7b
; CI-NEXT:    v_mov_b32_e32 v2, 0
; CI-NEXT:    v_mov_b32_e32 v0, s0
; CI-NEXT:    v_mov_b32_e32 v1, s1
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    ds_write_b64 v2, v[0:1] offset:16384
; CI-NEXT:    ds_write_b64 v2, v[0:1] offset:32760
; CI-NEXT:    s_endpgm
;
; GFX9-LABEL: store_misaligned64_constant_large_offsets:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_mov_b64 s[0:1], 0x7b
; GFX9-NEXT:    v_mov_b32_e32 v2, 0
; GFX9-NEXT:    v_mov_b32_e32 v0, s0
; GFX9-NEXT:    v_mov_b32_e32 v1, s1
; GFX9-NEXT:    ds_write_b64 v2, v[0:1] offset:16384
; GFX9-NEXT:    ds_write_b64 v2, v[0:1] offset:32760
; GFX9-NEXT:    s_endpgm
;
; GFX1250-LABEL: store_misaligned64_constant_large_offsets:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-NEXT:    v_nop
; GFX1250-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-NEXT:    v_mov_b64_e32 v[0:1], 0x7b
; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
; GFX1250-NEXT:    ds_store_b64 v2, v[0:1] offset:16384
; GFX1250-NEXT:    ds_store_b64 v2, v[0:1] offset:32760
; GFX1250-NEXT:    s_endpgm
  store i64 123, ptr addrspace(3) getelementptr inbounds ([4096 x i64], ptr addrspace(3) @bar.large, i32 0, i32 2048), align 4
  store i64 123, ptr addrspace(3) getelementptr inbounds ([4096 x i64], ptr addrspace(3) @bar.large, i32 0, i32 4095), align 4
  ret void
}

@sgemm.lA = internal unnamed_addr addrspace(3) global [264 x float] poison, align 4
@sgemm.lB = internal unnamed_addr addrspace(3) global [776 x float] poison, align 4

define amdgpu_kernel void @write2_sgemm_sequence(ptr addrspace(1) %C, i32 %lda, i32 %ldb, ptr addrspace(1) %in) #0 {
; CI-LABEL: write2_sgemm_sequence:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x4
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    s_load_dword s0, s[0:1], 0x0
; CI-NEXT:    s_lshl_b32 s1, s8, 2
; CI-NEXT:    s_add_i32 s2, s1, 0xc20
; CI-NEXT:    s_addk_i32 s1, 0xc60
; CI-NEXT:    v_mov_b32_e32 v0, s2
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    v_mov_b32_e32 v2, s0
; CI-NEXT:    v_mov_b32_e32 v3, s0
; CI-NEXT:    ds_write2_b32 v0, v2, v3 offset1:1
; CI-NEXT:    v_mov_b32_e32 v0, s1
; CI-NEXT:    ds_write2_b32 v0, v2, v3 offset1:1
; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v1
; CI-NEXT:    ds_write2_b32 v0, v2, v3 offset1:1
; CI-NEXT:    ds_write2_b32 v0, v2, v3 offset0:32 offset1:33
; CI-NEXT:    ds_write2_b32 v0, v2, v3 offset0:64 offset1:65
; CI-NEXT:    s_endpgm
;
; GFX9-LABEL: write2_sgemm_sequence:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x10
; GFX9-NEXT:    s_lshl_b32 s2, s8, 2
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    s_load_dword s0, s[0:1], 0x0
; GFX9-NEXT:    s_add_i32 s1, s2, 0xc20
; GFX9-NEXT:    s_addk_i32 s2, 0xc60
; GFX9-NEXT:    v_mov_b32_e32 v0, s1
; GFX9-NEXT:    v_mov_b32_e32 v2, s2
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    v_mov_b32_e32 v3, s0
; GFX9-NEXT:    v_mov_b32_e32 v4, s0
; GFX9-NEXT:    ds_write2_b32 v0, v3, v4 offset1:1
; GFX9-NEXT:    ds_write2_b32 v2, v3, v4 offset1:1
; GFX9-NEXT:    v_lshlrev_b32_e32 v0, 2, v1
; GFX9-NEXT:    ds_write2_b32 v0, v3, v4 offset1:1
; GFX9-NEXT:    ds_write2_b32 v0, v3, v4 offset0:32 offset1:33
; GFX9-NEXT:    ds_write2_b32 v0, v3, v4 offset0:64 offset1:65
; GFX9-NEXT:    s_endpgm
;
; GFX1250-UNALIGNED-LABEL: write2_sgemm_sequence:
; GFX1250-UNALIGNED:       ; %bb.0:
; GFX1250-UNALIGNED-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-UNALIGNED-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-UNALIGNED-NEXT:    v_nop
; GFX1250-UNALIGNED-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-UNALIGNED-NEXT:    s_load_b64 s[0:1], s[4:5], 0x10 nv
; GFX1250-UNALIGNED-NEXT:    s_and_b32 s2, ttmp6, 15
; GFX1250-UNALIGNED-NEXT:    s_getreg_b32 s3, hwreg(HW_REG_IB_STS2, 6, 4)
; GFX1250-UNALIGNED-NEXT:    s_wait_kmcnt 0x0
; GFX1250-UNALIGNED-NEXT:    s_load_b32 s0, s[0:1], 0x0
; GFX1250-UNALIGNED-NEXT:    s_wait_xcnt 0x0
; GFX1250-UNALIGNED-NEXT:    s_bfe_u32 s1, ttmp6, 0x4000c
; GFX1250-UNALIGNED-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1250-UNALIGNED-NEXT:    s_add_co_i32 s1, s1, 1
; GFX1250-UNALIGNED-NEXT:    s_mul_i32 s1, ttmp9, s1
; GFX1250-UNALIGNED-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1)
; GFX1250-UNALIGNED-NEXT:    s_add_co_i32 s2, s2, s1
; GFX1250-UNALIGNED-NEXT:    s_cmp_eq_u32 s3, 0
; GFX1250-UNALIGNED-NEXT:    s_cselect_b32 s1, ttmp9, s2
; GFX1250-UNALIGNED-NEXT:    s_lshl_b32 s1, s1, 2
; GFX1250-UNALIGNED-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1250-UNALIGNED-NEXT:    s_add_co_i32 s2, s1, 0xc20
; GFX1250-UNALIGNED-NEXT:    v_dual_mov_b32 v1, s2 :: v_dual_lshrrev_b32 v0, 8, v0
; GFX1250-UNALIGNED-NEXT:    s_addk_co_i32 s1, 0xc60
; GFX1250-UNALIGNED-NEXT:    s_wait_kmcnt 0x0
; GFX1250-UNALIGNED-NEXT:    v_dual_mov_b32 v4, s1 :: v_dual_mov_b32 v2, s0
; GFX1250-UNALIGNED-NEXT:    v_mov_b32_e32 v3, s0
; GFX1250-UNALIGNED-NEXT:    v_and_b32_e32 v0, 0xffc, v0
; GFX1250-UNALIGNED-NEXT:    ds_store_2addr_b32 v1, v2, v3 offset1:1
; GFX1250-UNALIGNED-NEXT:    ds_store_2addr_b32 v4, v2, v3 offset1:1
; GFX1250-UNALIGNED-NEXT:    ds_store_2addr_b32 v0, v2, v3 offset1:1
; GFX1250-UNALIGNED-NEXT:    ds_store_2addr_b32 v0, v2, v3 offset0:32 offset1:33
; GFX1250-UNALIGNED-NEXT:    ds_store_2addr_b32 v0, v2, v3 offset0:64 offset1:65
; GFX1250-UNALIGNED-NEXT:    s_endpgm
;
; GFX1250S-UNALIGNED-LABEL: write2_sgemm_sequence:
; GFX1250S-UNALIGNED:       ; %bb.0:
; GFX1250S-UNALIGNED-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250S-UNALIGNED-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250S-UNALIGNED-NEXT:    v_nop
; GFX1250S-UNALIGNED-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250S-UNALIGNED-NEXT:    s_load_b64 s[0:1], s[4:5], 0x10 nv
; GFX1250S-UNALIGNED-NEXT:    s_and_b32 s2, ttmp6, 15
; GFX1250S-UNALIGNED-NEXT:    s_getreg_b32 s3, hwreg(HW_REG_IB_STS2, 6, 4)
; GFX1250S-UNALIGNED-NEXT:    s_wait_kmcnt 0x0
; GFX1250S-UNALIGNED-NEXT:    s_load_b32 s0, s[0:1], 0x0
; GFX1250S-UNALIGNED-NEXT:    s_wait_xcnt 0x0
; GFX1250S-UNALIGNED-NEXT:    s_bfe_u32 s1, ttmp6, 0x4000c
; GFX1250S-UNALIGNED-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1250S-UNALIGNED-NEXT:    s_add_co_i32 s1, s1, 1
; GFX1250S-UNALIGNED-NEXT:    s_mul_i32 s1, ttmp9, s1
; GFX1250S-UNALIGNED-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1)
; GFX1250S-UNALIGNED-NEXT:    s_add_co_i32 s2, s2, s1
; GFX1250S-UNALIGNED-NEXT:    s_cmp_eq_u32 s3, 0
; GFX1250S-UNALIGNED-NEXT:    s_cselect_b32 s1, ttmp9, s2
; GFX1250S-UNALIGNED-NEXT:    s_lshl_b32 s2, s1, 2
; GFX1250S-UNALIGNED-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1)
; GFX1250S-UNALIGNED-NEXT:    v_dual_mov_b32 v3, s2 :: v_dual_lshrrev_b32 v2, 8, v0
; GFX1250S-UNALIGNED-NEXT:    v_and_b32_e32 v2, 0xffc, v2
; GFX1250S-UNALIGNED-NEXT:    s_wait_kmcnt 0x0
; GFX1250S-UNALIGNED-NEXT:    s_mov_b32 s1, s0
; GFX1250S-UNALIGNED-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
; GFX1250S-UNALIGNED-NEXT:    v_mov_b64_e32 v[0:1], s[0:1]
; GFX1250S-UNALIGNED-NEXT:    ds_store_b64 v3, v[0:1] offset:3104
; GFX1250S-UNALIGNED-NEXT:    ds_store_b64 v3, v[0:1] offset:3168
; GFX1250S-UNALIGNED-NEXT:    ds_store_b64 v2, v[0:1]
; GFX1250S-UNALIGNED-NEXT:    ds_store_b64 v2, v[0:1] offset:128
; GFX1250S-UNALIGNED-NEXT:    ds_store_b64 v2, v[0:1] offset:256
; GFX1250S-UNALIGNED-NEXT:    s_endpgm
  %x.i = tail call i32 @llvm.amdgcn.workgroup.id.x() #1
  %y.i = tail call i32 @llvm.amdgcn.workitem.id.y() #1
  %val = load float, ptr addrspace(1) %in
  %arrayidx44 = getelementptr inbounds [264 x float], ptr addrspace(3) @sgemm.lA, i32 0, i32 %x.i
  store float %val, ptr addrspace(3) %arrayidx44, align 4
  %add47 = add nsw i32 %x.i, 1
  %arrayidx48 = getelementptr inbounds [264 x float], ptr addrspace(3) @sgemm.lA, i32 0, i32 %add47
  store float %val, ptr addrspace(3) %arrayidx48, align 4
  %add51 = add nsw i32 %x.i, 16
  %arrayidx52 = getelementptr inbounds [264 x float], ptr addrspace(3) @sgemm.lA, i32 0, i32 %add51
  store float %val, ptr addrspace(3) %arrayidx52, align 4
  %add55 = add nsw i32 %x.i, 17
  %arrayidx56 = getelementptr inbounds [264 x float], ptr addrspace(3) @sgemm.lA, i32 0, i32 %add55
  store float %val, ptr addrspace(3) %arrayidx56, align 4
  %arrayidx60 = getelementptr inbounds [776 x float], ptr addrspace(3) @sgemm.lB, i32 0, i32 %y.i
  store float %val, ptr addrspace(3) %arrayidx60, align 4
  %add63 = add nsw i32 %y.i, 1
  %arrayidx64 = getelementptr inbounds [776 x float], ptr addrspace(3) @sgemm.lB, i32 0, i32 %add63
  store float %val, ptr addrspace(3) %arrayidx64, align 4
  %add67 = add nsw i32 %y.i, 32
  %arrayidx68 = getelementptr inbounds [776 x float], ptr addrspace(3) @sgemm.lB, i32 0, i32 %add67
  store float %val, ptr addrspace(3) %arrayidx68, align 4
  %add71 = add nsw i32 %y.i, 33
  %arrayidx72 = getelementptr inbounds [776 x float], ptr addrspace(3) @sgemm.lB, i32 0, i32 %add71
  store float %val, ptr addrspace(3) %arrayidx72, align 4
  %add75 = add nsw i32 %y.i, 64
  %arrayidx76 = getelementptr inbounds [776 x float], ptr addrspace(3) @sgemm.lB, i32 0, i32 %add75
  store float %val, ptr addrspace(3) %arrayidx76, align 4
  %add79 = add nsw i32 %y.i, 65
  %arrayidx80 = getelementptr inbounds [776 x float], ptr addrspace(3) @sgemm.lB, i32 0, i32 %add79
  store float %val, ptr addrspace(3) %arrayidx80, align 4
  ret void
}

define amdgpu_kernel void @simple_write2_v4f32_superreg_align4(ptr addrspace(3) %out, ptr addrspace(1) %in) #0 {
; CI-LABEL: simple_write2_v4f32_superreg_align4:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x2
; CI-NEXT:    s_load_dword s4, s[4:5], 0x0
; CI-NEXT:    v_lshlrev_b32_e32 v0, 4, v0
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x0
; CI-NEXT:    v_add_i32_e32 v0, vcc, s4, v0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    v_mov_b32_e32 v1, s0
; CI-NEXT:    v_mov_b32_e32 v2, s1
; CI-NEXT:    v_mov_b32_e32 v3, s2
; CI-NEXT:    ds_write2_b32 v0, v1, v2 offset1:1
; CI-NEXT:    v_mov_b32_e32 v1, s3
; CI-NEXT:    ds_write2_b32 v0, v3, v1 offset0:2 offset1:3
; CI-NEXT:    s_endpgm
;
; GFX9-ALIGNED-LABEL: simple_write2_v4f32_superreg_align4:
; GFX9-ALIGNED:       ; %bb.0:
; GFX9-ALIGNED-NEXT:    s_load_dwordx2 s[6:7], s[4:5], 0x8
; GFX9-ALIGNED-NEXT:    s_load_dword s8, s[4:5], 0x0
; GFX9-ALIGNED-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-ALIGNED-NEXT:    v_lshl_add_u32 v0, v0, 4, s8
; GFX9-ALIGNED-NEXT:    s_load_dwordx4 s[0:3], s[6:7], 0x0
; GFX9-ALIGNED-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-ALIGNED-NEXT:    v_mov_b32_e32 v1, s0
; GFX9-ALIGNED-NEXT:    v_mov_b32_e32 v2, s1
; GFX9-ALIGNED-NEXT:    v_mov_b32_e32 v3, s2
; GFX9-ALIGNED-NEXT:    v_mov_b32_e32 v4, s3
; GFX9-ALIGNED-NEXT:    ds_write2_b32 v0, v1, v2 offset1:1
; GFX9-ALIGNED-NEXT:    ds_write2_b32 v0, v3, v4 offset0:2 offset1:3
; GFX9-ALIGNED-NEXT:    s_endpgm
;
; GFX9-UNALIGNED-LABEL: simple_write2_v4f32_superreg_align4:
; GFX9-UNALIGNED:       ; %bb.0:
; GFX9-UNALIGNED-NEXT:    s_load_dwordx2 s[6:7], s[4:5], 0x8
; GFX9-UNALIGNED-NEXT:    s_load_dword s8, s[4:5], 0x0
; GFX9-UNALIGNED-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-UNALIGNED-NEXT:    v_lshl_add_u32 v0, v0, 4, s8
; GFX9-UNALIGNED-NEXT:    s_load_dwordx4 s[0:3], s[6:7], 0x0
; GFX9-UNALIGNED-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-UNALIGNED-NEXT:    v_mov_b32_e32 v1, s2
; GFX9-UNALIGNED-NEXT:    v_mov_b32_e32 v2, s3
; GFX9-UNALIGNED-NEXT:    v_mov_b32_e32 v3, s0
; GFX9-UNALIGNED-NEXT:    v_mov_b32_e32 v4, s1
; GFX9-UNALIGNED-NEXT:    ds_write2_b32 v0, v1, v2 offset0:2 offset1:3
; GFX9-UNALIGNED-NEXT:    ds_write2_b32 v0, v3, v4 offset1:1
; GFX9-UNALIGNED-NEXT:    s_endpgm
;
; GFX1250-UNALIGNED-LABEL: simple_write2_v4f32_superreg_align4:
; GFX1250-UNALIGNED:       ; %bb.0:
; GFX1250-UNALIGNED-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-UNALIGNED-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-UNALIGNED-NEXT:    v_nop
; GFX1250-UNALIGNED-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-UNALIGNED-NEXT:    s_clause 0x1
; GFX1250-UNALIGNED-NEXT:    s_load_b64 s[6:7], s[4:5], 0x8 nv
; GFX1250-UNALIGNED-NEXT:    s_load_b32 s8, s[4:5], 0x0 nv
; GFX1250-UNALIGNED-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
; GFX1250-UNALIGNED-NEXT:    s_wait_kmcnt 0x0
; GFX1250-UNALIGNED-NEXT:    s_load_b128 s[0:3], s[6:7], 0x0
; GFX1250-UNALIGNED-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250-UNALIGNED-NEXT:    v_lshl_add_u32 v0, v0, 4, s8
; GFX1250-UNALIGNED-NEXT:    s_wait_kmcnt 0x0
; GFX1250-UNALIGNED-NEXT:    v_dual_mov_b32 v1, s2 :: v_dual_mov_b32 v2, s3
; GFX1250-UNALIGNED-NEXT:    v_dual_mov_b32 v3, s0 :: v_dual_mov_b32 v4, s1
; GFX1250-UNALIGNED-NEXT:    ds_store_2addr_b32 v0, v1, v2 offset0:2 offset1:3
; GFX1250-UNALIGNED-NEXT:    ds_store_2addr_b32 v0, v3, v4 offset1:1
; GFX1250-UNALIGNED-NEXT:    s_endpgm
;
; GFX1250S-UNALIGNED-LABEL: simple_write2_v4f32_superreg_align4:
; GFX1250S-UNALIGNED:       ; %bb.0:
; GFX1250S-UNALIGNED-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250S-UNALIGNED-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250S-UNALIGNED-NEXT:    v_nop
; GFX1250S-UNALIGNED-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250S-UNALIGNED-NEXT:    s_clause 0x1
; GFX1250S-UNALIGNED-NEXT:    s_load_b64 s[6:7], s[4:5], 0x8 nv
; GFX1250S-UNALIGNED-NEXT:    s_load_b32 s8, s[4:5], 0x0 nv
; GFX1250S-UNALIGNED-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
; GFX1250S-UNALIGNED-NEXT:    s_wait_kmcnt 0x0
; GFX1250S-UNALIGNED-NEXT:    s_load_b128 s[0:3], s[6:7], 0x0
; GFX1250S-UNALIGNED-NEXT:    s_delay_alu instid0(VALU_DEP_1)
; GFX1250S-UNALIGNED-NEXT:    v_lshl_add_u32 v4, v0, 4, s8
; GFX1250S-UNALIGNED-NEXT:    s_wait_kmcnt 0x0
; GFX1250S-UNALIGNED-NEXT:    v_dual_mov_b32 v0, s2 :: v_dual_mov_b32 v1, s3
; GFX1250S-UNALIGNED-NEXT:    v_dual_mov_b32 v2, s0 :: v_dual_mov_b32 v3, s1
; GFX1250S-UNALIGNED-NEXT:    ds_store_b64 v4, v[0:1] offset:8
; GFX1250S-UNALIGNED-NEXT:    ds_store_b64 v4, v[2:3]
; GFX1250S-UNALIGNED-NEXT:    s_endpgm
  %x.i = tail call i32 @llvm.amdgcn.workitem.id.x() #1
  %in.gep = getelementptr inbounds <4 x float>, ptr addrspace(1) %in
  %val0 = load <4 x float>, ptr addrspace(1) %in.gep, align 4
  %out.gep = getelementptr inbounds <4 x float>, ptr addrspace(3) %out, i32 %x.i
  store <4 x float> %val0, ptr addrspace(3) %out.gep, align 4
  ret void
}

@v2i32_align1 = internal addrspace(3) global [100 x <2 x i32>] poison, align 1

define amdgpu_kernel void @write2_v2i32_align1_odd_offset() {
; CI-LABEL: write2_v2i32_align1_odd_offset:
; CI:       ; %bb.0: ; %entry
; CI-NEXT:    v_mov_b32_e32 v0, 0x7b
; CI-NEXT:    v_mov_b32_e32 v1, 0
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    ds_write_b8 v1, v0 offset:65
; CI-NEXT:    v_mov_b32_e32 v0, 1
; CI-NEXT:    ds_write_b8 v1, v0 offset:70
; CI-NEXT:    v_mov_b32_e32 v0, 0xc8
; CI-NEXT:    ds_write_b8 v1, v0 offset:69
; CI-NEXT:    ds_write_b8 v1, v1 offset:68
; CI-NEXT:    ds_write_b8 v1, v1 offset:67
; CI-NEXT:    ds_write_b8 v1, v1 offset:66
; CI-NEXT:    ds_write_b8 v1, v1 offset:72
; CI-NEXT:    ds_write_b8 v1, v1 offset:71
; CI-NEXT:    s_endpgm
;
; GFX9-ALIGNED-LABEL: write2_v2i32_align1_odd_offset:
; GFX9-ALIGNED:       ; %bb.0: ; %entry
; GFX9-ALIGNED-NEXT:    v_mov_b32_e32 v0, 0x7b
; GFX9-ALIGNED-NEXT:    v_mov_b32_e32 v1, 0
; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v0 offset:65
; GFX9-ALIGNED-NEXT:    v_mov_b32_e32 v0, 1
; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v0 offset:70
; GFX9-ALIGNED-NEXT:    v_mov_b32_e32 v0, 0xc8
; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v0 offset:69
; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v1 offset:68
; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v1 offset:67
; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v1 offset:66
; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v1 offset:72
; GFX9-ALIGNED-NEXT:    ds_write_b8 v1, v1 offset:71
; GFX9-ALIGNED-NEXT:    s_endpgm
;
; GFX9-UNALIGNED-LABEL: write2_v2i32_align1_odd_offset:
; GFX9-UNALIGNED:       ; %bb.0: ; %entry
; GFX9-UNALIGNED-NEXT:    v_mov_b32_e32 v0, 0x7b
; GFX9-UNALIGNED-NEXT:    v_mov_b32_e32 v1, 0x1c8
; GFX9-UNALIGNED-NEXT:    v_mov_b32_e32 v2, 0
; GFX9-UNALIGNED-NEXT:    ds_write_b64 v2, v[0:1] offset:65
; GFX9-UNALIGNED-NEXT:    s_endpgm
;
; GFX1250-LABEL: write2_v2i32_align1_odd_offset:
; GFX1250:       ; %bb.0: ; %entry
; GFX1250-NEXT:    s_setreg_imm32_b32 hwreg(HW_REG_WAVE_MODE, 25, 1), 1 ; msbs: dst=0 src0=0 src1=0 src2=0
; GFX1250-NEXT:    s_mov_b64 s[64:65], 0
; GFX1250-NEXT:    v_nop
; GFX1250-NEXT:    global_prefetch_b8 v0, s[64:65] scope:SCOPE_SE
; GFX1250-NEXT:    v_mov_b64_e32 v[0:1], 0x1c80000007b
; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
; GFX1250-NEXT:    ds_store_b64 v2, v[0:1] offset:65
; GFX1250-NEXT:    s_endpgm
entry:
  store <2 x i32> <i32 123, i32 456>, ptr addrspace(3) getelementptr (i8, ptr addrspace(3) @v2i32_align1, i32 65), align 1
  ret void
}

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

attributes #0 = { nounwind }
attributes #1 = { nounwind readnone speculatable }
attributes #2 = { convergent nounwind }