; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 3
; RUN: llc -mtriple=amdgcn -mcpu=pitcairn < %s | FileCheck -enable-var-scope -check-prefix=SI %s
; RUN: llc -mtriple=amdgcn -mcpu=gfx1250 --amdgpu-enable-vopd=0 < %s | FileCheck -enable-var-scope -check-prefix=GFX1250 %s
; RUN: llc -mtriple=r600 -mcpu=cypress < %s | FileCheck -enable-var-scope -check-prefix=EG %s

define amdgpu_kernel void @v_test_imax_sge_i32(ptr addrspace(1) %out, ptr addrspace(1) %aptr, ptr addrspace(1) %bptr) nounwind {
; SI-LABEL: v_test_imax_sge_i32:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
; SI-NEXT:    s_mov_b32 s7, 0xf000
; SI-NEXT:    s_mov_b32 s2, 0
; SI-NEXT:    s_mov_b32 s3, s7
; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; SI-NEXT:    v_mov_b32_e32 v1, 0
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    buffer_load_dword v0, v[0:1], s[0:3], 0 addr64
; SI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
; SI-NEXT:    s_mov_b32 s6, -1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    s_load_dword s2, s[2:3], 0x0
; SI-NEXT:    s_mov_b32 s4, s0
; SI-NEXT:    s_mov_b32 s5, s1
; SI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; SI-NEXT:    v_max_i32_e32 v0, s2, v0
; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: v_test_imax_sge_i32:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x34
; GFX1250-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
; GFX1250-NEXT:    v_mov_b32_e32 v1, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    global_load_b32 v0, v0, s[0:1] scale_offset
; GFX1250-NEXT:    s_wait_xcnt 0x0
; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_load_b32 s2, s[2:3], 0x0
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    v_max_i32_e32 v0, s2, v0
; GFX1250-NEXT:    global_store_b32 v1, v0, s[0:1]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: v_test_imax_sge_i32:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 3, @10, KC0[CB0:0-32], KC1[]
; EG-NEXT:    TEX 1 @6
; EG-NEXT:    ALU 2, @14, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.X, T1.X, 1
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    Fetch clause starting at 6:
; EG-NEXT:     VTX_READ_32 T0.X, T0.X, 0, #1
; EG-NEXT:     VTX_READ_32 T1.X, T1.X, 0, #1
; EG-NEXT:    ALU clause starting at 10:
; EG-NEXT:     MOV T1.X, KC0[2].Z,
; EG-NEXT:     LSHL * T0.W, T0.X, literal.x,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
; EG-NEXT:     ADD_INT * T0.X, KC0[2].W, PV.W,
; EG-NEXT:    ALU clause starting at 14:
; EG-NEXT:     MAX_INT T0.X, T1.X, T0.X,
; EG-NEXT:     LSHR * T1.X, KC0[2].Y, literal.x,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
  %tid = call i32 @llvm.amdgcn.workitem.id.x()
  %gep.in = getelementptr inbounds i32, ptr addrspace(1) %bptr, i32 %tid
  %a = load i32, ptr addrspace(1) %aptr, align 4
  %b = load i32, ptr addrspace(1) %gep.in, align 4
  %cmp = icmp sge i32 %a, %b
  %val = select i1 %cmp, i32 %a, i32 %b
  store i32 %val, ptr addrspace(1) %out, align 4
  ret void
}

; These could be merged into one
define amdgpu_kernel void @v_test_imax_sge_v4i32(ptr addrspace(1) %out, ptr addrspace(1) %aptr, ptr addrspace(1) %bptr) nounwind {
; SI-LABEL: v_test_imax_sge_v4i32:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
; SI-NEXT:    s_mov_b32 s7, 0xf000
; SI-NEXT:    s_mov_b32 s2, 0
; SI-NEXT:    s_mov_b32 s3, s7
; SI-NEXT:    v_lshlrev_b32_e32 v0, 4, v0
; SI-NEXT:    v_mov_b32_e32 v1, 0
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    buffer_load_dwordx4 v[0:3], v[0:1], s[0:3], 0 addr64
; SI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
; SI-NEXT:    s_mov_b32 s6, -1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    s_load_dwordx4 s[8:11], s[2:3], 0x0
; SI-NEXT:    s_mov_b32 s4, s0
; SI-NEXT:    s_mov_b32 s5, s1
; SI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; SI-NEXT:    v_max_i32_e32 v3, s11, v3
; SI-NEXT:    v_max_i32_e32 v2, s10, v2
; SI-NEXT:    v_max_i32_e32 v1, s9, v1
; SI-NEXT:    v_max_i32_e32 v0, s8, v0
; SI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[4:7], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: v_test_imax_sge_v4i32:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x34
; GFX1250-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
; GFX1250-NEXT:    v_mov_b32_e32 v4, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    global_load_b128 v[0:3], v0, s[0:1] scale_offset
; GFX1250-NEXT:    s_wait_xcnt 0x0
; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_load_b128 s[4:7], s[2:3], 0x0
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    v_max_i32_e32 v3, s7, v3
; GFX1250-NEXT:    v_max_i32_e32 v2, s6, v2
; GFX1250-NEXT:    v_max_i32_e32 v1, s5, v1
; GFX1250-NEXT:    v_max_i32_e32 v0, s4, v0
; GFX1250-NEXT:    global_store_b128 v4, v[0:3], s[0:1]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: v_test_imax_sge_v4i32:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 3, @10, KC0[CB0:0-32], KC1[]
; EG-NEXT:    TEX 1 @6
; EG-NEXT:    ALU 5, @14, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.XYZW, T1.X, 1
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    Fetch clause starting at 6:
; EG-NEXT:     VTX_READ_128 T0.XYZW, T0.X, 0, #1
; EG-NEXT:     VTX_READ_128 T1.XYZW, T1.X, 0, #1
; EG-NEXT:    ALU clause starting at 10:
; EG-NEXT:     MOV T1.X, KC0[2].Z,
; EG-NEXT:     LSHL * T0.W, T0.X, literal.x,
; EG-NEXT:    4(5.605194e-45), 0(0.000000e+00)
; EG-NEXT:     ADD_INT * T0.X, KC0[2].W, PV.W,
; EG-NEXT:    ALU clause starting at 14:
; EG-NEXT:     MAX_INT * T0.W, T1.W, T0.W,
; EG-NEXT:     MAX_INT * T0.Z, T1.Z, T0.Z,
; EG-NEXT:     MAX_INT * T0.Y, T1.Y, T0.Y,
; EG-NEXT:     MAX_INT T0.X, T1.X, T0.X,
; EG-NEXT:     LSHR * T1.X, KC0[2].Y, literal.x,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
  %tid = call i32 @llvm.amdgcn.workitem.id.x()
  %gep.in = getelementptr inbounds <4 x i32>, ptr addrspace(1) %bptr, i32 %tid
  %a = load <4 x i32>, ptr addrspace(1) %aptr, align 4
  %b = load <4 x i32>, ptr addrspace(1) %gep.in, align 4
  %cmp = icmp sge <4 x i32> %a, %b
  %val = select <4 x i1> %cmp, <4 x i32> %a, <4 x i32> %b
  store <4 x i32> %val, ptr addrspace(1) %out, align 4
  ret void
}

define amdgpu_kernel void @s_test_imax_sge_i32(ptr addrspace(1) %out, i32 %a, i32 %b) nounwind {
; SI-LABEL: s_test_imax_sge_i32:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
; SI-NEXT:    s_mov_b32 s7, 0xf000
; SI-NEXT:    s_mov_b32 s6, -1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    s_mov_b32 s4, s0
; SI-NEXT:    s_max_i32 s0, s2, s3
; SI-NEXT:    s_mov_b32 s5, s1
; SI-NEXT:    v_mov_b32_e32 v0, s0
; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: s_test_imax_sge_i32:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_max_i32 s2, s2, s3
; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: s_test_imax_sge_i32:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 2, @4, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T1.X, T0.X, 1
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    ALU clause starting at 4:
; EG-NEXT:     LSHR T0.X, KC0[2].Y, literal.x,
; EG-NEXT:     MAX_INT * T1.X, KC0[2].Z, KC0[2].W,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
  %cmp = icmp sge i32 %a, %b
  %val = select i1 %cmp, i32 %a, i32 %b
  store i32 %val, ptr addrspace(1) %out, align 4
  ret void
}

define amdgpu_kernel void @s_test_imax_sge_imm_i32(ptr addrspace(1) %out, i32 %a) nounwind {
; SI-LABEL: s_test_imax_sge_imm_i32:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dword s6, s[4:5], 0xb
; SI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x9
; SI-NEXT:    s_mov_b32 s3, 0xf000
; SI-NEXT:    s_mov_b32 s2, -1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    s_max_i32 s4, s6, 9
; SI-NEXT:    v_mov_b32_e32 v0, s4
; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: s_test_imax_sge_imm_i32:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_load_b96 s[0:2], s[4:5], 0x24
; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_max_i32 s2, s2, 9
; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: s_test_imax_sge_imm_i32:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 2, @4, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T1.X, T0.X, 1
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    ALU clause starting at 4:
; EG-NEXT:     LSHR T0.X, KC0[2].Y, literal.x,
; EG-NEXT:     MAX_INT * T1.X, KC0[2].Z, literal.y,
; EG-NEXT:    2(2.802597e-45), 9(1.261169e-44)
  %cmp = icmp sge i32 %a, 9
  %val = select i1 %cmp, i32 %a, i32 9
  store i32 %val, ptr addrspace(1) %out, align 4
  ret void
}

define amdgpu_kernel void @v_test_imax_sge_i8(ptr addrspace(1) %out, ptr addrspace(1) %aptr, ptr addrspace(1) %bptr) nounwind {
; SI-LABEL: v_test_imax_sge_i8:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
; SI-NEXT:    s_load_dwordx2 s[8:9], s[4:5], 0xd
; SI-NEXT:    s_mov_b32 s7, 0xf000
; SI-NEXT:    s_mov_b32 s6, -1
; SI-NEXT:    s_mov_b32 s14, s6
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    s_mov_b32 s12, s2
; SI-NEXT:    s_mov_b32 s13, s3
; SI-NEXT:    s_mov_b32 s15, s7
; SI-NEXT:    s_mov_b32 s10, s6
; SI-NEXT:    s_mov_b32 s11, s7
; SI-NEXT:    buffer_load_sbyte v0, off, s[12:15], 0
; SI-NEXT:    buffer_load_sbyte v1, off, s[8:11], 0
; SI-NEXT:    s_mov_b32 s4, s0
; SI-NEXT:    s_mov_b32 s5, s1
; SI-NEXT:    s_waitcnt vmcnt(0)
; SI-NEXT:    v_max_i32_e32 v0, v0, v1
; SI-NEXT:    buffer_store_byte v0, off, s[4:7], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: v_test_imax_sge_i8:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_clause 0x1
; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX1250-NEXT:    s_load_b64 s[6:7], s[4:5], 0x34
; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_load_i8 s4, s[2:3], 0x0
; GFX1250-NEXT:    s_load_i8 s5, s[6:7], 0x0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_max_i32 s2, s4, s5
; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
; GFX1250-NEXT:    global_store_b8 v0, v1, s[0:1]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: v_test_imax_sge_i8:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 0, @12, KC0[CB0:0-32], KC1[]
; EG-NEXT:    TEX 0 @8
; EG-NEXT:    ALU 0, @13, KC0[CB0:0-32], KC1[]
; EG-NEXT:    TEX 0 @10
; EG-NEXT:    ALU 14, @14, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT MSKOR T0.XW, T1.X
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    Fetch clause starting at 8:
; EG-NEXT:     VTX_READ_8 T0.X, T0.X, 0, #1
; EG-NEXT:    Fetch clause starting at 10:
; EG-NEXT:     VTX_READ_8 T1.X, T1.X, 0, #1
; EG-NEXT:    ALU clause starting at 12:
; EG-NEXT:     MOV * T0.X, KC0[2].W,
; EG-NEXT:    ALU clause starting at 13:
; EG-NEXT:     MOV * T1.X, KC0[2].Z,
; EG-NEXT:    ALU clause starting at 14:
; EG-NEXT:     BFE_INT T0.Z, T0.X, 0.0, literal.x,
; EG-NEXT:     BFE_INT T0.W, T1.X, 0.0, literal.x, BS:VEC_120/SCL_212
; EG-NEXT:     AND_INT * T1.W, KC0[2].Y, literal.y,
; EG-NEXT:    8(1.121039e-44), 3(4.203895e-45)
; EG-NEXT:     MAX_INT * T0.W, PV.W, PV.Z,
; EG-NEXT:     AND_INT T0.W, PV.W, literal.x,
; EG-NEXT:     LSHL * T1.W, T1.W, literal.y,
; EG-NEXT:    255(3.573311e-43), 3(4.203895e-45)
; EG-NEXT:     LSHL T0.X, PV.W, PS,
; EG-NEXT:     LSHL * T0.W, literal.x, PS,
; EG-NEXT:    255(3.573311e-43), 0(0.000000e+00)
; EG-NEXT:     MOV T0.Y, 0.0,
; EG-NEXT:     MOV * T0.Z, 0.0,
; EG-NEXT:     LSHR * T1.X, KC0[2].Y, literal.x,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
  %a = load i8, ptr addrspace(1) %aptr, align 1
  %b = load i8, ptr addrspace(1) %bptr, align 1
  %cmp = icmp sge i8 %a, %b
  %val = select i1 %cmp, i8 %a, i8 %b
  store i8 %val, ptr addrspace(1) %out, align 1
  ret void
}

define amdgpu_kernel void @s_test_imax_sgt_imm_i32(ptr addrspace(1) %out, i32 %a) nounwind {
; SI-LABEL: s_test_imax_sgt_imm_i32:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dword s6, s[4:5], 0xb
; SI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x9
; SI-NEXT:    s_mov_b32 s3, 0xf000
; SI-NEXT:    s_mov_b32 s2, -1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    s_max_i32 s4, s6, 9
; SI-NEXT:    v_mov_b32_e32 v0, s4
; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: s_test_imax_sgt_imm_i32:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_load_b96 s[0:2], s[4:5], 0x24
; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_max_i32 s2, s2, 9
; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: s_test_imax_sgt_imm_i32:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 2, @4, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T1.X, T0.X, 1
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    ALU clause starting at 4:
; EG-NEXT:     LSHR T0.X, KC0[2].Y, literal.x,
; EG-NEXT:     MAX_INT * T1.X, KC0[2].Z, literal.y,
; EG-NEXT:    2(2.802597e-45), 9(1.261169e-44)
  %cmp = icmp sgt i32 %a, 9
  %val = select i1 %cmp, i32 %a, i32 9
  store i32 %val, ptr addrspace(1) %out, align 4
  ret void
}

define amdgpu_kernel void @s_test_imax_sgt_imm_v2i32(ptr addrspace(1) %out, <2 x i32> %a) nounwind {
; SI-LABEL: s_test_imax_sgt_imm_v2i32:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
; SI-NEXT:    s_mov_b32 s7, 0xf000
; SI-NEXT:    s_mov_b32 s6, -1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    s_mov_b32 s4, s0
; SI-NEXT:    s_mov_b32 s5, s1
; SI-NEXT:    s_max_i32 s0, s3, 9
; SI-NEXT:    s_max_i32 s1, s2, 9
; SI-NEXT:    v_mov_b32_e32 v0, s1
; SI-NEXT:    v_mov_b32_e32 v1, s0
; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: s_test_imax_sgt_imm_v2i32:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_max_i32 s2, s2, 9
; GFX1250-NEXT:    s_max_i32 s3, s3, 9
; GFX1250-NEXT:    v_mov_b32_e32 v0, s2
; GFX1250-NEXT:    v_mov_b32_e32 v1, s3
; GFX1250-NEXT:    global_store_b64 v2, v[0:1], s[0:1]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: s_test_imax_sgt_imm_v2i32:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 4, @4, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.XY, T1.X, 1
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    ALU clause starting at 4:
; EG-NEXT:     MAX_INT * T0.Y, KC0[3].X, literal.x,
; EG-NEXT:    9(1.261169e-44), 0(0.000000e+00)
; EG-NEXT:     MAX_INT T0.X, KC0[2].W, literal.x,
; EG-NEXT:     LSHR * T1.X, KC0[2].Y, literal.y,
; EG-NEXT:    9(1.261169e-44), 2(2.802597e-45)
  %cmp = icmp sgt <2 x i32> %a, <i32 9, i32 9>
  %val = select <2 x i1> %cmp, <2 x i32> %a, <2 x i32> <i32 9, i32 9>
  store <2 x i32> %val, ptr addrspace(1) %out, align 4
  ret void
}

define amdgpu_kernel void @v_test_imax_sgt_i32(ptr addrspace(1) %out, ptr addrspace(1) %aptr, ptr addrspace(1) %bptr) nounwind {
; SI-LABEL: v_test_imax_sgt_i32:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
; SI-NEXT:    s_mov_b32 s7, 0xf000
; SI-NEXT:    s_mov_b32 s2, 0
; SI-NEXT:    s_mov_b32 s3, s7
; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; SI-NEXT:    v_mov_b32_e32 v1, 0
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    buffer_load_dword v0, v[0:1], s[0:3], 0 addr64
; SI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
; SI-NEXT:    s_mov_b32 s6, -1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    s_load_dword s2, s[2:3], 0x0
; SI-NEXT:    s_mov_b32 s4, s0
; SI-NEXT:    s_mov_b32 s5, s1
; SI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; SI-NEXT:    v_max_i32_e32 v0, s2, v0
; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: v_test_imax_sgt_i32:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x34
; GFX1250-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
; GFX1250-NEXT:    v_mov_b32_e32 v1, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    global_load_b32 v0, v0, s[0:1] scale_offset
; GFX1250-NEXT:    s_wait_xcnt 0x0
; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_load_b32 s2, s[2:3], 0x0
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    v_max_i32_e32 v0, s2, v0
; GFX1250-NEXT:    global_store_b32 v1, v0, s[0:1]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: v_test_imax_sgt_i32:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 3, @10, KC0[CB0:0-32], KC1[]
; EG-NEXT:    TEX 1 @6
; EG-NEXT:    ALU 2, @14, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.X, T1.X, 1
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    Fetch clause starting at 6:
; EG-NEXT:     VTX_READ_32 T0.X, T0.X, 0, #1
; EG-NEXT:     VTX_READ_32 T1.X, T1.X, 0, #1
; EG-NEXT:    ALU clause starting at 10:
; EG-NEXT:     MOV T1.X, KC0[2].Z,
; EG-NEXT:     LSHL * T0.W, T0.X, literal.x,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
; EG-NEXT:     ADD_INT * T0.X, KC0[2].W, PV.W,
; EG-NEXT:    ALU clause starting at 14:
; EG-NEXT:     MAX_INT T0.X, T1.X, T0.X,
; EG-NEXT:     LSHR * T1.X, KC0[2].Y, literal.x,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
  %tid = call i32 @llvm.amdgcn.workitem.id.x()
  %gep.in = getelementptr inbounds i32, ptr addrspace(1) %bptr, i32 %tid
  %a = load i32, ptr addrspace(1) %aptr, align 4
  %b = load i32, ptr addrspace(1) %gep.in, align 4
  %cmp = icmp sgt i32 %a, %b
  %val = select i1 %cmp, i32 %a, i32 %b
  store i32 %val, ptr addrspace(1) %out, align 4
  ret void
}

define amdgpu_kernel void @s_test_imax_sgt_i32(ptr addrspace(1) %out, i32 %a, i32 %b) nounwind {
; SI-LABEL: s_test_imax_sgt_i32:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
; SI-NEXT:    s_mov_b32 s7, 0xf000
; SI-NEXT:    s_mov_b32 s6, -1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    s_mov_b32 s4, s0
; SI-NEXT:    s_max_i32 s0, s2, s3
; SI-NEXT:    s_mov_b32 s5, s1
; SI-NEXT:    v_mov_b32_e32 v0, s0
; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: s_test_imax_sgt_i32:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_max_i32 s2, s2, s3
; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: s_test_imax_sgt_i32:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 2, @4, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T1.X, T0.X, 1
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    ALU clause starting at 4:
; EG-NEXT:     LSHR T0.X, KC0[2].Y, literal.x,
; EG-NEXT:     MAX_INT * T1.X, KC0[2].Z, KC0[2].W,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
  %cmp = icmp sgt i32 %a, %b
  %val = select i1 %cmp, i32 %a, i32 %b
  store i32 %val, ptr addrspace(1) %out, align 4
  ret void
}

define amdgpu_kernel void @v_test_umax_uge_i32(ptr addrspace(1) %out, ptr addrspace(1) %aptr, ptr addrspace(1) %bptr) nounwind {
; SI-LABEL: v_test_umax_uge_i32:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
; SI-NEXT:    s_mov_b32 s7, 0xf000
; SI-NEXT:    s_mov_b32 s2, 0
; SI-NEXT:    s_mov_b32 s3, s7
; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; SI-NEXT:    v_mov_b32_e32 v1, 0
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    buffer_load_dword v0, v[0:1], s[0:3], 0 addr64
; SI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
; SI-NEXT:    s_mov_b32 s6, -1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    s_load_dword s2, s[2:3], 0x0
; SI-NEXT:    s_mov_b32 s4, s0
; SI-NEXT:    s_mov_b32 s5, s1
; SI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; SI-NEXT:    v_max_u32_e32 v0, s2, v0
; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: v_test_umax_uge_i32:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x34
; GFX1250-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
; GFX1250-NEXT:    v_mov_b32_e32 v1, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    global_load_b32 v0, v0, s[0:1] scale_offset
; GFX1250-NEXT:    s_wait_xcnt 0x0
; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_load_b32 s2, s[2:3], 0x0
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    v_max_u32_e32 v0, s2, v0
; GFX1250-NEXT:    global_store_b32 v1, v0, s[0:1]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: v_test_umax_uge_i32:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 3, @10, KC0[CB0:0-32], KC1[]
; EG-NEXT:    TEX 1 @6
; EG-NEXT:    ALU 2, @14, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.X, T1.X, 1
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    Fetch clause starting at 6:
; EG-NEXT:     VTX_READ_32 T0.X, T0.X, 0, #1
; EG-NEXT:     VTX_READ_32 T1.X, T1.X, 0, #1
; EG-NEXT:    ALU clause starting at 10:
; EG-NEXT:     MOV T1.X, KC0[2].Z,
; EG-NEXT:     LSHL * T0.W, T0.X, literal.x,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
; EG-NEXT:     ADD_INT * T0.X, KC0[2].W, PV.W,
; EG-NEXT:    ALU clause starting at 14:
; EG-NEXT:     MAX_UINT T0.X, T1.X, T0.X,
; EG-NEXT:     LSHR * T1.X, KC0[2].Y, literal.x,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
  %tid = call i32 @llvm.amdgcn.workitem.id.x()
  %gep.in = getelementptr inbounds i32, ptr addrspace(1) %bptr, i32 %tid
  %a = load i32, ptr addrspace(1) %aptr, align 4
  %b = load i32, ptr addrspace(1) %gep.in, align 4
  %cmp = icmp uge i32 %a, %b
  %val = select i1 %cmp, i32 %a, i32 %b
  store i32 %val, ptr addrspace(1) %out, align 4
  ret void
}

define amdgpu_kernel void @s_test_umax_uge_i32(ptr addrspace(1) %out, i32 %a, i32 %b) nounwind {
; SI-LABEL: s_test_umax_uge_i32:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
; SI-NEXT:    s_mov_b32 s7, 0xf000
; SI-NEXT:    s_mov_b32 s6, -1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    s_mov_b32 s4, s0
; SI-NEXT:    s_max_u32 s0, s2, s3
; SI-NEXT:    s_mov_b32 s5, s1
; SI-NEXT:    v_mov_b32_e32 v0, s0
; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: s_test_umax_uge_i32:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_max_u32 s2, s2, s3
; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: s_test_umax_uge_i32:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 2, @4, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T1.X, T0.X, 1
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    ALU clause starting at 4:
; EG-NEXT:     LSHR T0.X, KC0[2].Y, literal.x,
; EG-NEXT:     MAX_UINT * T1.X, KC0[2].Z, KC0[2].W,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
  %cmp = icmp uge i32 %a, %b
  %val = select i1 %cmp, i32 %a, i32 %b
  store i32 %val, ptr addrspace(1) %out, align 4
  ret void
}

define amdgpu_kernel void @s_test_umax_uge_v3i32(ptr addrspace(1) %out, <3 x i32> %a, <3 x i32> %b) nounwind {
; SI-LABEL: s_test_umax_uge_v3i32:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dwordx8 s[8:15], s[4:5], 0xd
; SI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x9
; SI-NEXT:    s_mov_b32 s3, 0xf000
; SI-NEXT:    s_mov_b32 s2, -1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    s_max_u32 s6, s10, s14
; SI-NEXT:    s_max_u32 s4, s9, s13
; SI-NEXT:    s_max_u32 s5, s8, s12
; SI-NEXT:    v_mov_b32_e32 v0, s6
; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0 offset:8
; SI-NEXT:    s_waitcnt expcnt(0)
; SI-NEXT:    v_mov_b32_e32 v0, s5
; SI-NEXT:    v_mov_b32_e32 v1, s4
; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: s_test_umax_uge_v3i32:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_clause 0x1
; GFX1250-NEXT:    s_load_b256 s[8:15], s[4:5], 0x34
; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x24
; GFX1250-NEXT:    v_mov_b32_e32 v3, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_max_u32 s2, s10, s14
; GFX1250-NEXT:    s_max_u32 s3, s8, s12
; GFX1250-NEXT:    s_max_u32 s4, s9, s13
; GFX1250-NEXT:    v_mov_b32_e32 v0, s3
; GFX1250-NEXT:    v_mov_b32_e32 v1, s4
; GFX1250-NEXT:    v_mov_b32_e32 v2, s2
; GFX1250-NEXT:    global_store_b96 v3, v[0:2], s[0:1]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: s_test_umax_uge_v3i32:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 7, @4, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T3.X, T2.X, 0
; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.XY, T1.X, 1
; EG-NEXT:    CF_END
; EG-NEXT:    ALU clause starting at 4:
; EG-NEXT:     MAX_UINT * T0.Y, KC0[3].Z, KC0[4].Z,
; EG-NEXT:     MAX_UINT * T0.X, KC0[3].Y, KC0[4].Y,
; EG-NEXT:     LSHR T1.X, KC0[2].Y, literal.x,
; EG-NEXT:     ADD_INT * T0.W, KC0[2].Y, literal.y,
; EG-NEXT:    2(2.802597e-45), 8(1.121039e-44)
; EG-NEXT:     LSHR T2.X, PV.W, literal.x,
; EG-NEXT:     MAX_UINT * T3.X, KC0[3].W, KC0[4].W,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
  %cmp = icmp uge <3 x i32> %a, %b
  %val = select <3 x i1> %cmp, <3 x i32> %a, <3 x i32> %b
  store <3 x i32> %val, ptr addrspace(1) %out, align 4
  ret void
}

define amdgpu_kernel void @v_test_umax_uge_i8(ptr addrspace(1) %out, ptr addrspace(1) %aptr, ptr addrspace(1) %bptr) nounwind {
; SI-LABEL: v_test_umax_uge_i8:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
; SI-NEXT:    s_load_dwordx2 s[8:9], s[4:5], 0xd
; SI-NEXT:    s_mov_b32 s7, 0xf000
; SI-NEXT:    s_mov_b32 s6, -1
; SI-NEXT:    s_mov_b32 s14, s6
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    s_mov_b32 s12, s2
; SI-NEXT:    s_mov_b32 s13, s3
; SI-NEXT:    s_mov_b32 s15, s7
; SI-NEXT:    s_mov_b32 s10, s6
; SI-NEXT:    s_mov_b32 s11, s7
; SI-NEXT:    buffer_load_ubyte v0, off, s[12:15], 0
; SI-NEXT:    buffer_load_ubyte v1, off, s[8:11], 0
; SI-NEXT:    s_mov_b32 s4, s0
; SI-NEXT:    s_mov_b32 s5, s1
; SI-NEXT:    s_waitcnt vmcnt(0)
; SI-NEXT:    v_max_u32_e32 v0, v0, v1
; SI-NEXT:    buffer_store_byte v0, off, s[4:7], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: v_test_umax_uge_i8:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_clause 0x1
; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX1250-NEXT:    s_load_b64 s[6:7], s[4:5], 0x34
; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_load_u8 s4, s[2:3], 0x0
; GFX1250-NEXT:    s_load_u8 s5, s[6:7], 0x0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_max_u32 s2, s4, s5
; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
; GFX1250-NEXT:    global_store_b8 v0, v1, s[0:1]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: v_test_umax_uge_i8:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 1, @10, KC0[CB0:0-32], KC1[]
; EG-NEXT:    TEX 1 @6
; EG-NEXT:    ALU 11, @12, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT MSKOR T0.XW, T1.X
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    Fetch clause starting at 6:
; EG-NEXT:     VTX_READ_8 T1.X, T1.X, 0, #1
; EG-NEXT:     VTX_READ_8 T0.X, T0.X, 0, #1
; EG-NEXT:    ALU clause starting at 10:
; EG-NEXT:     MOV T0.X, KC0[2].Z,
; EG-NEXT:     MOV * T1.X, KC0[2].W,
; EG-NEXT:    ALU clause starting at 12:
; EG-NEXT:     AND_INT T0.W, KC0[2].Y, literal.x,
; EG-NEXT:     MAX_UINT * T1.W, T0.X, T1.X,
; EG-NEXT:    3(4.203895e-45), 0(0.000000e+00)
; EG-NEXT:     LSHL * T0.W, PV.W, literal.x,
; EG-NEXT:    3(4.203895e-45), 0(0.000000e+00)
; EG-NEXT:     LSHL T0.X, T1.W, PV.W,
; EG-NEXT:     LSHL * T0.W, literal.x, PV.W,
; EG-NEXT:    255(3.573311e-43), 0(0.000000e+00)
; EG-NEXT:     MOV T0.Y, 0.0,
; EG-NEXT:     MOV * T0.Z, 0.0,
; EG-NEXT:     LSHR * T1.X, KC0[2].Y, literal.x,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
  %a = load i8, ptr addrspace(1) %aptr, align 1
  %b = load i8, ptr addrspace(1) %bptr, align 1
  %cmp = icmp uge i8 %a, %b
  %val = select i1 %cmp, i8 %a, i8 %b
  store i8 %val, ptr addrspace(1) %out, align 1
  ret void
}

define amdgpu_kernel void @v_test_umax_ugt_i32(ptr addrspace(1) %out, ptr addrspace(1) %aptr, ptr addrspace(1) %bptr) nounwind {
; SI-LABEL: v_test_umax_ugt_i32:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0xd
; SI-NEXT:    s_mov_b32 s7, 0xf000
; SI-NEXT:    s_mov_b32 s2, 0
; SI-NEXT:    s_mov_b32 s3, s7
; SI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; SI-NEXT:    v_mov_b32_e32 v1, 0
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    buffer_load_dword v0, v[0:1], s[0:3], 0 addr64
; SI-NEXT:    s_load_dword s0, s[0:1], 0x0
; SI-NEXT:    s_load_dwordx2 s[4:5], s[4:5], 0x9
; SI-NEXT:    s_mov_b32 s6, -1
; SI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; SI-NEXT:    v_max_u32_e32 v0, s0, v0
; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: v_test_umax_ugt_i32:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x34
; GFX1250-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
; GFX1250-NEXT:    v_mov_b32_e32 v1, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    global_load_b32 v0, v0, s[0:1] scale_offset
; GFX1250-NEXT:    s_load_b32 s6, s[0:1], 0x0
; GFX1250-NEXT:    s_load_b64 s[2:3], s[4:5], 0x24
; GFX1250-NEXT:    s_wait_loadcnt 0x0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    v_max_u32_e32 v0, s6, v0
; GFX1250-NEXT:    global_store_b32 v1, v0, s[2:3]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: v_test_umax_ugt_i32:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 3, @10, KC0[CB0:0-32], KC1[]
; EG-NEXT:    TEX 1 @6
; EG-NEXT:    ALU 2, @14, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.X, T1.X, 1
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    Fetch clause starting at 6:
; EG-NEXT:     VTX_READ_32 T0.X, T0.X, 0, #1
; EG-NEXT:     VTX_READ_32 T1.X, T1.X, 0, #1
; EG-NEXT:    ALU clause starting at 10:
; EG-NEXT:     MOV T1.X, KC0[2].W,
; EG-NEXT:     LSHL * T0.W, T0.X, literal.x,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
; EG-NEXT:     ADD_INT * T0.X, KC0[2].W, PV.W,
; EG-NEXT:    ALU clause starting at 14:
; EG-NEXT:     MAX_UINT T0.X, T0.X, T1.X,
; EG-NEXT:     LSHR * T1.X, KC0[2].Y, literal.x,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
  %tid = call i32 @llvm.amdgcn.workitem.id.x()
  %gep.in = getelementptr inbounds i32, ptr addrspace(1) %bptr, i32 %tid
  %a = load i32, ptr addrspace(1) %gep.in, align 4
  %b = load i32, ptr addrspace(1) %bptr, align 4
  %cmp = icmp ugt i32 %a, %b
  %val = select i1 %cmp, i32 %a, i32 %b
  store i32 %val, ptr addrspace(1) %out, align 4
  ret void
}

define amdgpu_kernel void @s_test_umax_ugt_i32(ptr addrspace(1) %out, i32 %a, i32 %b) nounwind {
; SI-LABEL: s_test_umax_ugt_i32:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
; SI-NEXT:    s_mov_b32 s7, 0xf000
; SI-NEXT:    s_mov_b32 s6, -1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    s_mov_b32 s4, s0
; SI-NEXT:    s_max_u32 s0, s2, s3
; SI-NEXT:    s_mov_b32 s5, s1
; SI-NEXT:    v_mov_b32_e32 v0, s0
; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: s_test_umax_ugt_i32:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_max_u32 s2, s2, s3
; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: s_test_umax_ugt_i32:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 2, @4, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T1.X, T0.X, 1
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    ALU clause starting at 4:
; EG-NEXT:     LSHR T0.X, KC0[2].Y, literal.x,
; EG-NEXT:     MAX_UINT * T1.X, KC0[2].Z, KC0[2].W,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
  %cmp = icmp ugt i32 %a, %b
  %val = select i1 %cmp, i32 %a, i32 %b
  store i32 %val, ptr addrspace(1) %out, align 4
  ret void
}

define amdgpu_kernel void @s_test_umax_ugt_imm_v2i32(ptr addrspace(1) %out, <2 x i32> %a) nounwind {
; SI-LABEL: s_test_umax_ugt_imm_v2i32:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
; SI-NEXT:    s_mov_b32 s7, 0xf000
; SI-NEXT:    s_mov_b32 s6, -1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    s_mov_b32 s4, s0
; SI-NEXT:    s_mov_b32 s5, s1
; SI-NEXT:    s_max_u32 s0, s3, 23
; SI-NEXT:    s_max_u32 s1, s2, 15
; SI-NEXT:    v_mov_b32_e32 v0, s1
; SI-NEXT:    v_mov_b32_e32 v1, s0
; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: s_test_umax_ugt_imm_v2i32:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_max_u32 s2, s2, 15
; GFX1250-NEXT:    s_max_u32 s3, s3, 23
; GFX1250-NEXT:    v_mov_b32_e32 v0, s2
; GFX1250-NEXT:    v_mov_b32_e32 v1, s3
; GFX1250-NEXT:    global_store_b64 v2, v[0:1], s[0:1]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: s_test_umax_ugt_imm_v2i32:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 4, @4, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.XY, T1.X, 1
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    ALU clause starting at 4:
; EG-NEXT:     MAX_UINT * T0.Y, KC0[3].X, literal.x,
; EG-NEXT:    23(3.222986e-44), 0(0.000000e+00)
; EG-NEXT:     MAX_UINT T0.X, KC0[2].W, literal.x,
; EG-NEXT:     LSHR * T1.X, KC0[2].Y, literal.y,
; EG-NEXT:    15(2.101948e-44), 2(2.802597e-45)
  %cmp = icmp ugt <2 x i32> %a, <i32 15, i32 23>
  %val = select <2 x i1> %cmp, <2 x i32> %a, <2 x i32> <i32 15, i32 23>
  store <2 x i32> %val, ptr addrspace(1) %out, align 4
  ret void
}

; Make sure redundant and removed

define amdgpu_kernel void @simplify_demanded_bits_test_umax_ugt_i16(ptr addrspace(1) %out, [8 x i32], i16 zeroext %a, [8 x i32], i16 zeroext %b) nounwind {
; SI-LABEL: simplify_demanded_bits_test_umax_ugt_i16:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dword s6, s[4:5], 0x13
; SI-NEXT:    s_load_dword s7, s[4:5], 0x1c
; SI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x9
; SI-NEXT:    s_mov_b32 s3, 0xf000
; SI-NEXT:    s_mov_b32 s2, -1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    s_and_b32 s4, s6, 0xffff
; SI-NEXT:    s_and_b32 s5, s7, 0xffff
; SI-NEXT:    s_max_u32 s4, s4, s5
; SI-NEXT:    v_mov_b32_e32 v0, s4
; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: simplify_demanded_bits_test_umax_ugt_i16:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_clause 0x2
; GFX1250-NEXT:    s_load_b32 s2, s[4:5], 0x4c
; GFX1250-NEXT:    s_load_b32 s3, s[4:5], 0x70
; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x24
; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_and_b32 s2, s2, 0xffff
; GFX1250-NEXT:    s_and_b32 s3, s3, 0xffff
; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1250-NEXT:    s_max_u32 s2, s2, s3
; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: simplify_demanded_bits_test_umax_ugt_i16:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 0, @10, KC0[], KC1[]
; EG-NEXT:    TEX 1 @6
; EG-NEXT:    ALU 5, @11, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.X, T1.X, 1
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    Fetch clause starting at 6:
; EG-NEXT:     VTX_READ_16 T1.X, T0.X, 72, #3
; EG-NEXT:     VTX_READ_16 T0.X, T0.X, 108, #3
; EG-NEXT:    ALU clause starting at 10:
; EG-NEXT:     MOV * T0.X, 0.0,
; EG-NEXT:    ALU clause starting at 11:
; EG-NEXT:     BFE_INT T0.Z, T1.X, 0.0, literal.x,
; EG-NEXT:     BFE_INT * T0.W, T0.X, 0.0, literal.x, BS:VEC_120/SCL_212
; EG-NEXT:    16(2.242078e-44), 0(0.000000e+00)
; EG-NEXT:     MAX_UINT T0.X, PV.Z, PV.W,
; EG-NEXT:     LSHR * T1.X, KC0[2].Y, literal.x,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
  %a.ext = zext i16 %a to i32
  %b.ext = zext i16 %b to i32
  %cmp = icmp ugt i32 %a.ext, %b.ext
  %val = select i1 %cmp, i32 %a.ext, i32 %b.ext
  %mask = and i32 %val, 65535
  store i32 %mask, ptr addrspace(1) %out
  ret void
}

; Make sure redundant sign_extend_inreg removed.

define amdgpu_kernel void @simplify_demanded_bits_test_max_slt_i16(ptr addrspace(1) %out, [8 x i32], i16 signext %a, [8 x i32], i16 signext %b) nounwind {
; SI-LABEL: simplify_demanded_bits_test_max_slt_i16:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dword s6, s[4:5], 0x13
; SI-NEXT:    s_load_dword s7, s[4:5], 0x1c
; SI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x9
; SI-NEXT:    s_mov_b32 s3, 0xf000
; SI-NEXT:    s_mov_b32 s2, -1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    s_sext_i32_i16 s4, s6
; SI-NEXT:    s_sext_i32_i16 s5, s7
; SI-NEXT:    s_max_i32 s4, s4, s5
; SI-NEXT:    v_mov_b32_e32 v0, s4
; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: simplify_demanded_bits_test_max_slt_i16:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_clause 0x2
; GFX1250-NEXT:    s_load_b32 s2, s[4:5], 0x4c
; GFX1250-NEXT:    s_load_b32 s3, s[4:5], 0x70
; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x24
; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_sext_i32_i16 s2, s2
; GFX1250-NEXT:    s_sext_i32_i16 s3, s3
; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1250-NEXT:    s_max_i32 s2, s2, s3
; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: simplify_demanded_bits_test_max_slt_i16:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 0, @10, KC0[], KC1[]
; EG-NEXT:    TEX 1 @6
; EG-NEXT:    ALU 5, @11, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.X, T1.X, 1
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    Fetch clause starting at 6:
; EG-NEXT:     VTX_READ_16 T1.X, T0.X, 72, #3
; EG-NEXT:     VTX_READ_16 T0.X, T0.X, 108, #3
; EG-NEXT:    ALU clause starting at 10:
; EG-NEXT:     MOV * T0.X, 0.0,
; EG-NEXT:    ALU clause starting at 11:
; EG-NEXT:     BFE_INT T0.Z, T1.X, 0.0, literal.x,
; EG-NEXT:     BFE_INT * T0.W, T0.X, 0.0, literal.x, BS:VEC_120/SCL_212
; EG-NEXT:    16(2.242078e-44), 0(0.000000e+00)
; EG-NEXT:     MAX_INT T0.X, PV.Z, PV.W,
; EG-NEXT:     LSHR * T1.X, KC0[2].Y, literal.x,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
  %a.ext = sext i16 %a to i32
  %b.ext = sext i16 %b to i32
  %cmp = icmp sgt i32 %a.ext, %b.ext
  %val = select i1 %cmp, i32 %a.ext, i32 %b.ext
  %shl = shl i32 %val, 16
  %sextinreg = ashr i32 %shl, 16
  store i32 %sextinreg, ptr addrspace(1) %out
  ret void
}

define amdgpu_kernel void @s_test_imax_sge_i16(ptr addrspace(1) %out, [8 x i32], i16 %a, [8 x i32], i16 %b) nounwind {
; SI-LABEL: s_test_imax_sge_i16:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dword s6, s[4:5], 0x13
; SI-NEXT:    s_load_dword s7, s[4:5], 0x1c
; SI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x9
; SI-NEXT:    s_mov_b32 s3, 0xf000
; SI-NEXT:    s_mov_b32 s2, -1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    s_sext_i32_i16 s4, s6
; SI-NEXT:    s_sext_i32_i16 s5, s7
; SI-NEXT:    s_max_i32 s4, s4, s5
; SI-NEXT:    v_mov_b32_e32 v0, s4
; SI-NEXT:    buffer_store_short v0, off, s[0:3], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: s_test_imax_sge_i16:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_clause 0x2
; GFX1250-NEXT:    s_load_b32 s2, s[4:5], 0x70
; GFX1250-NEXT:    s_load_b32 s3, s[4:5], 0x4c
; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x24
; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    s_sext_i32_i16 s2, s2
; GFX1250-NEXT:    s_sext_i32_i16 s3, s3
; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
; GFX1250-NEXT:    s_max_i32 s2, s3, s2
; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
; GFX1250-NEXT:    global_store_b16 v0, v1, s[0:1]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: s_test_imax_sge_i16:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 0, @10, KC0[], KC1[]
; EG-NEXT:    TEX 1 @6
; EG-NEXT:    ALU 14, @11, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT MSKOR T0.XW, T1.X
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    Fetch clause starting at 6:
; EG-NEXT:     VTX_READ_16 T1.X, T0.X, 72, #3
; EG-NEXT:     VTX_READ_16 T0.X, T0.X, 108, #3
; EG-NEXT:    ALU clause starting at 10:
; EG-NEXT:     MOV * T0.X, 0.0,
; EG-NEXT:    ALU clause starting at 11:
; EG-NEXT:     BFE_INT T0.Z, T1.X, 0.0, literal.x,
; EG-NEXT:     BFE_INT T0.W, T0.X, 0.0, literal.x, BS:VEC_120/SCL_212
; EG-NEXT:     AND_INT * T1.W, KC0[2].Y, literal.y,
; EG-NEXT:    16(2.242078e-44), 3(4.203895e-45)
; EG-NEXT:     MAX_INT * T0.W, PV.Z, PV.W,
; EG-NEXT:     AND_INT T0.W, PV.W, literal.x,
; EG-NEXT:     LSHL * T1.W, T1.W, literal.y,
; EG-NEXT:    65535(9.183409e-41), 3(4.203895e-45)
; EG-NEXT:     LSHL T0.X, PV.W, PS,
; EG-NEXT:     LSHL * T0.W, literal.x, PS,
; EG-NEXT:    65535(9.183409e-41), 0(0.000000e+00)
; EG-NEXT:     MOV T0.Y, 0.0,
; EG-NEXT:     MOV * T0.Z, 0.0,
; EG-NEXT:     LSHR * T1.X, KC0[2].Y, literal.x,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
  %cmp = icmp sge i16 %a, %b
  %val = select i1 %cmp, i16 %a, i16 %b
  store i16 %val, ptr addrspace(1) %out
  ret void
}

; 64 bit

define amdgpu_kernel void @test_umax_ugt_i64(ptr addrspace(1) %out, i64 %a, i64 %b) nounwind {
; SI-LABEL: test_umax_ugt_i64:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
; SI-NEXT:    s_load_dwordx2 s[8:9], s[4:5], 0xd
; SI-NEXT:    s_mov_b32 s7, 0xf000
; SI-NEXT:    s_mov_b32 s6, -1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    s_mov_b32 s4, s0
; SI-NEXT:    v_mov_b32_e32 v0, s8
; SI-NEXT:    v_mov_b32_e32 v1, s9
; SI-NEXT:    v_cmp_gt_u64_e32 vcc, s[2:3], v[0:1]
; SI-NEXT:    s_mov_b32 s5, s1
; SI-NEXT:    s_and_b64 s[0:1], vcc, exec
; SI-NEXT:    s_cselect_b32 s0, s3, s9
; SI-NEXT:    s_cselect_b32 s1, s2, s8
; SI-NEXT:    v_mov_b32_e32 v0, s1
; SI-NEXT:    v_mov_b32_e32 v1, s0
; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: test_umax_ugt_i64:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_clause 0x1
; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX1250-NEXT:    s_load_b64 s[6:7], s[4:5], 0x34
; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    v_max_u64 v[0:1], s[2:3], s[6:7]
; GFX1250-NEXT:    global_store_b64 v2, v[0:1], s[0:1]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: test_umax_ugt_i64:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 7, @4, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.XY, T1.X, 1
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    ALU clause starting at 4:
; EG-NEXT:     SETE_INT T0.Z, KC0[3].X, KC0[3].Z,
; EG-NEXT:     SETGT_UINT * T0.W, KC0[3].X, KC0[3].Z,
; EG-NEXT:     SETGT_UINT * T1.W, KC0[2].W, KC0[3].Y,
; EG-NEXT:     CNDE_INT * T0.W, T0.Z, T0.W, PV.W,
; EG-NEXT:     CNDE_INT * T0.Y, PV.W, KC0[3].Z, KC0[3].X,
; EG-NEXT:     CNDE_INT * T0.X, T0.W, KC0[3].Y, KC0[2].W,
; EG-NEXT:     LSHR * T1.X, KC0[2].Y, literal.x,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
  %tmp = icmp ugt i64 %a, %b
  %val = select i1 %tmp, i64 %a, i64 %b
  store i64 %val, ptr addrspace(1) %out, align 8
  ret void
}

define amdgpu_kernel void @test_umax_uge_i64(ptr addrspace(1) %out, i64 %a, i64 %b) nounwind {
; SI-LABEL: test_umax_uge_i64:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
; SI-NEXT:    s_load_dwordx2 s[8:9], s[4:5], 0xd
; SI-NEXT:    s_mov_b32 s7, 0xf000
; SI-NEXT:    s_mov_b32 s6, -1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    s_mov_b32 s4, s0
; SI-NEXT:    v_mov_b32_e32 v0, s8
; SI-NEXT:    v_mov_b32_e32 v1, s9
; SI-NEXT:    v_cmp_ge_u64_e32 vcc, s[2:3], v[0:1]
; SI-NEXT:    s_mov_b32 s5, s1
; SI-NEXT:    s_and_b64 s[0:1], vcc, exec
; SI-NEXT:    s_cselect_b32 s0, s3, s9
; SI-NEXT:    s_cselect_b32 s1, s2, s8
; SI-NEXT:    v_mov_b32_e32 v0, s1
; SI-NEXT:    v_mov_b32_e32 v1, s0
; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: test_umax_uge_i64:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_clause 0x1
; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX1250-NEXT:    s_load_b64 s[6:7], s[4:5], 0x34
; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    v_max_u64 v[0:1], s[2:3], s[6:7]
; GFX1250-NEXT:    global_store_b64 v2, v[0:1], s[0:1]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: test_umax_uge_i64:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 7, @4, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.XY, T1.X, 1
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    ALU clause starting at 4:
; EG-NEXT:     SETE_INT T0.Z, KC0[3].X, KC0[3].Z,
; EG-NEXT:     SETGT_UINT * T0.W, KC0[3].X, KC0[3].Z,
; EG-NEXT:     SETGT_UINT * T1.W, KC0[2].W, KC0[3].Y,
; EG-NEXT:     CNDE_INT * T0.W, T0.Z, T0.W, PV.W,
; EG-NEXT:     CNDE_INT * T0.Y, PV.W, KC0[3].Z, KC0[3].X,
; EG-NEXT:     CNDE_INT * T0.X, T0.W, KC0[3].Y, KC0[2].W,
; EG-NEXT:     LSHR * T1.X, KC0[2].Y, literal.x,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
  %tmp = icmp uge i64 %a, %b
  %val = select i1 %tmp, i64 %a, i64 %b
  store i64 %val, ptr addrspace(1) %out, align 8
  ret void
}

define amdgpu_kernel void @test_imax_sgt_i64(ptr addrspace(1) %out, i64 %a, i64 %b) nounwind {
; SI-LABEL: test_imax_sgt_i64:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
; SI-NEXT:    s_load_dwordx2 s[8:9], s[4:5], 0xd
; SI-NEXT:    s_mov_b32 s7, 0xf000
; SI-NEXT:    s_mov_b32 s6, -1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    s_mov_b32 s4, s0
; SI-NEXT:    v_mov_b32_e32 v0, s8
; SI-NEXT:    v_mov_b32_e32 v1, s9
; SI-NEXT:    v_cmp_gt_i64_e32 vcc, s[2:3], v[0:1]
; SI-NEXT:    s_mov_b32 s5, s1
; SI-NEXT:    s_and_b64 s[0:1], vcc, exec
; SI-NEXT:    s_cselect_b32 s0, s3, s9
; SI-NEXT:    s_cselect_b32 s1, s2, s8
; SI-NEXT:    v_mov_b32_e32 v0, s1
; SI-NEXT:    v_mov_b32_e32 v1, s0
; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: test_imax_sgt_i64:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_clause 0x1
; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX1250-NEXT:    s_load_b64 s[6:7], s[4:5], 0x34
; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    v_max_i64 v[0:1], s[2:3], s[6:7]
; GFX1250-NEXT:    global_store_b64 v2, v[0:1], s[0:1]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: test_imax_sgt_i64:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 7, @4, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.XY, T1.X, 1
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    ALU clause starting at 4:
; EG-NEXT:     SETE_INT T0.Z, KC0[3].X, KC0[3].Z,
; EG-NEXT:     SETGT_INT * T0.W, KC0[3].X, KC0[3].Z,
; EG-NEXT:     SETGT_UINT * T1.W, KC0[2].W, KC0[3].Y,
; EG-NEXT:     CNDE_INT * T0.W, T0.Z, T0.W, PV.W,
; EG-NEXT:     CNDE_INT * T0.Y, PV.W, KC0[3].Z, KC0[3].X,
; EG-NEXT:     CNDE_INT * T0.X, T0.W, KC0[3].Y, KC0[2].W,
; EG-NEXT:     LSHR * T1.X, KC0[2].Y, literal.x,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
  %tmp = icmp sgt i64 %a, %b
  %val = select i1 %tmp, i64 %a, i64 %b
  store i64 %val, ptr addrspace(1) %out, align 8
  ret void
}

define amdgpu_kernel void @test_imax_sge_i64(ptr addrspace(1) %out, i64 %a, i64 %b) nounwind {
; SI-LABEL: test_imax_sge_i64:
; SI:       ; %bb.0:
; SI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x9
; SI-NEXT:    s_load_dwordx2 s[8:9], s[4:5], 0xd
; SI-NEXT:    s_mov_b32 s7, 0xf000
; SI-NEXT:    s_mov_b32 s6, -1
; SI-NEXT:    s_waitcnt lgkmcnt(0)
; SI-NEXT:    s_mov_b32 s4, s0
; SI-NEXT:    v_mov_b32_e32 v0, s8
; SI-NEXT:    v_mov_b32_e32 v1, s9
; SI-NEXT:    v_cmp_ge_i64_e32 vcc, s[2:3], v[0:1]
; SI-NEXT:    s_mov_b32 s5, s1
; SI-NEXT:    s_and_b64 s[0:1], vcc, exec
; SI-NEXT:    s_cselect_b32 s0, s3, s9
; SI-NEXT:    s_cselect_b32 s1, s2, s8
; SI-NEXT:    v_mov_b32_e32 v0, s1
; SI-NEXT:    v_mov_b32_e32 v1, s0
; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
; SI-NEXT:    s_endpgm
;
; GFX1250-LABEL: test_imax_sge_i64:
; GFX1250:       ; %bb.0:
; GFX1250-NEXT:    s_clause 0x1
; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
; GFX1250-NEXT:    s_load_b64 s[6:7], s[4:5], 0x34
; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
; GFX1250-NEXT:    s_wait_kmcnt 0x0
; GFX1250-NEXT:    v_max_i64 v[0:1], s[2:3], s[6:7]
; GFX1250-NEXT:    global_store_b64 v2, v[0:1], s[0:1]
; GFX1250-NEXT:    s_endpgm
;
; EG-LABEL: test_imax_sge_i64:
; EG:       ; %bb.0:
; EG-NEXT:    ALU 7, @4, KC0[CB0:0-32], KC1[]
; EG-NEXT:    MEM_RAT_CACHELESS STORE_RAW T0.XY, T1.X, 1
; EG-NEXT:    CF_END
; EG-NEXT:    PAD
; EG-NEXT:    ALU clause starting at 4:
; EG-NEXT:     SETE_INT T0.Z, KC0[3].X, KC0[3].Z,
; EG-NEXT:     SETGT_INT * T0.W, KC0[3].X, KC0[3].Z,
; EG-NEXT:     SETGT_UINT * T1.W, KC0[2].W, KC0[3].Y,
; EG-NEXT:     CNDE_INT * T0.W, T0.Z, T0.W, PV.W,
; EG-NEXT:     CNDE_INT * T0.Y, PV.W, KC0[3].Z, KC0[3].X,
; EG-NEXT:     CNDE_INT * T0.X, T0.W, KC0[3].Y, KC0[2].W,
; EG-NEXT:     LSHR * T1.X, KC0[2].Y, literal.x,
; EG-NEXT:    2(2.802597e-45), 0(0.000000e+00)
  %tmp = icmp sge i64 %a, %b
  %val = select i1 %tmp, i64 %a, i64 %b
  store i64 %val, ptr addrspace(1) %out, align 8
  ret void
}

declare i32 @llvm.amdgcn.workitem.id.x() #0

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