llvm/test/CodeGen/PowerPC/ppc64-P9-vabsd.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=powerpc64le-unknown-linux-gnu -mcpu=pwr9 -ppc-vsr-nums-as-vr -ppc-asm-full-reg-names -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK,CHECK-PWR9,CHECK-PWR9-LE
; RUN: llc < %s -mtriple=powerpc64-unknown-linux-gnu -mcpu=pwr9 -ppc-vsr-nums-as-vr -ppc-asm-full-reg-names -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK,CHECK-PWR9,CHECK-PWR9-BE
; RUN: llc < %s -mtriple=powerpc64le-unknown-linux-gnu -mcpu=pwr8 -ppc-vsr-nums-as-vr -ppc-asm-full-reg-names -verify-machineinstrs | FileCheck %s --check-prefixes=CHECK,CHECK-PWR78,CHECK-PWR8 -implicit-check-not vabsdu
; RUN: llc < %s -mtriple=powerpc64-unknown-linux-gnu -mcpu=pwr7 -ppc-vsr-nums-as-vr -ppc-asm-full-reg-names -verify-machineinstrs | FileCheck %s -check-prefixes=CHECK,CHECK-PWR78,CHECK-PWR7 -implicit-check-not vmaxsd

define <4 x i32> @simple_absv_32(<4 x i32> %a) local_unnamed_addr {
; CHECK-PWR9-LABEL: simple_absv_32:
; CHECK-PWR9:       # %bb.0: # %entry
; CHECK-PWR9-NEXT:    vnegw v3, v2
; CHECK-PWR9-NEXT:    vmaxsw v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: simple_absv_32:
; CHECK-PWR78:       # %bb.0: # %entry
; CHECK-PWR78-NEXT:    xxlxor v3, v3, v3
; CHECK-PWR78-NEXT:    vsubuwm v3, v3, v2
; CHECK-PWR78-NEXT:    vmaxsw v2, v2, v3
; CHECK-PWR78-NEXT:    blr
entry:
  %sub.i = sub <4 x i32> zeroinitializer, %a
  %0 = tail call <4 x i32> @llvm.ppc.altivec.vmaxsw(<4 x i32> %a, <4 x i32> %sub.i)
  ret <4 x i32> %0
}

define <4 x i32> @simple_absv_32_swap(<4 x i32> %a) local_unnamed_addr {
; CHECK-PWR9-LABEL: simple_absv_32_swap:
; CHECK-PWR9:       # %bb.0: # %entry
; CHECK-PWR9-NEXT:    vnegw v3, v2
; CHECK-PWR9-NEXT:    vmaxsw v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: simple_absv_32_swap:
; CHECK-PWR78:       # %bb.0: # %entry
; CHECK-PWR78-NEXT:    xxlxor v3, v3, v3
; CHECK-PWR78-NEXT:    vsubuwm v3, v3, v2
; CHECK-PWR78-NEXT:    vmaxsw v2, v3, v2
; CHECK-PWR78-NEXT:    blr
entry:
  %sub.i = sub <4 x i32> zeroinitializer, %a
  %0 = tail call <4 x i32> @llvm.ppc.altivec.vmaxsw(<4 x i32> %sub.i, <4 x i32> %a)
  ret <4 x i32> %0
}

define <8 x i16> @simple_absv_16(<8 x i16> %a) local_unnamed_addr {
; CHECK-LABEL: simple_absv_16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    xxlxor v3, v3, v3
; CHECK-NEXT:    vsubuhm v3, v3, v2
; CHECK-NEXT:    vmaxsh v2, v2, v3
; CHECK-NEXT:    blr
entry:
  %sub.i = sub <8 x i16> zeroinitializer, %a
  %0 = tail call <8 x i16> @llvm.ppc.altivec.vmaxsh(<8 x i16> %a, <8 x i16> %sub.i)
  ret <8 x i16> %0
}

define <16 x i8> @simple_absv_8(<16 x i8> %a) local_unnamed_addr {
; CHECK-LABEL: simple_absv_8:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    xxlxor v3, v3, v3
; CHECK-NEXT:    vsububm v3, v3, v2
; CHECK-NEXT:    vmaxsb v2, v2, v3
; CHECK-NEXT:    blr
entry:
  %sub.i = sub <16 x i8> zeroinitializer, %a
  %0 = tail call <16 x i8> @llvm.ppc.altivec.vmaxsb(<16 x i8> %a, <16 x i8> %sub.i)
  ret <16 x i8> %0
}

; v2i64 vmax isn't avaiable on pwr7
define <2 x i64> @sub_absv_64(<2 x i64> %a, <2 x i64> %b) local_unnamed_addr {
; CHECK-PWR9-LABEL: sub_absv_64:
; CHECK-PWR9:       # %bb.0: # %entry
; CHECK-PWR9-NEXT:    vsubudm v2, v2, v3
; CHECK-PWR9-NEXT:    vnegd v3, v2
; CHECK-PWR9-NEXT:    vmaxsd v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR8-LABEL: sub_absv_64:
; CHECK-PWR8:       # %bb.0: # %entry
; CHECK-PWR8-NEXT:    vsubudm v2, v2, v3
; CHECK-PWR8-NEXT:    xxlxor v3, v3, v3
; CHECK-PWR8-NEXT:    vsubudm v3, v3, v2
; CHECK-PWR8-NEXT:    vmaxsd v2, v2, v3
; CHECK-PWR8-NEXT:    blr
;
; CHECK-PWR7-LABEL: sub_absv_64:
; CHECK-PWR7:       # %bb.0: # %entry
; CHECK-PWR7-NEXT:    addi r3, r1, -48
; CHECK-PWR7-NEXT:    stxvd2x v2, 0, r3
; CHECK-PWR7-NEXT:    addi r3, r1, -32
; CHECK-PWR7-NEXT:    stxvd2x v3, 0, r3
; CHECK-PWR7-NEXT:    ld r4, -40(r1)
; CHECK-PWR7-NEXT:    ld r5, -24(r1)
; CHECK-PWR7-NEXT:    ld r3, -48(r1)
; CHECK-PWR7-NEXT:    sub r4, r4, r5
; CHECK-PWR7-NEXT:    sradi r5, r4, 63
; CHECK-PWR7-NEXT:    xor r4, r4, r5
; CHECK-PWR7-NEXT:    sub r4, r4, r5
; CHECK-PWR7-NEXT:    ld r5, -32(r1)
; CHECK-PWR7-NEXT:    std r4, -8(r1)
; CHECK-PWR7-NEXT:    sub r3, r3, r5
; CHECK-PWR7-NEXT:    sradi r4, r3, 63
; CHECK-PWR7-NEXT:    xor r3, r3, r4
; CHECK-PWR7-NEXT:    sub r3, r3, r4
; CHECK-PWR7-NEXT:    std r3, -16(r1)
; CHECK-PWR7-NEXT:    addi r3, r1, -16
; CHECK-PWR7-NEXT:    lxvd2x v2, 0, r3
; CHECK-PWR7-NEXT:    blr
entry:
  %0 = sub nsw <2 x i64> %a, %b
  %1 = icmp sgt <2 x i64> %0, <i64 -1, i64 -1>
  %2 = sub <2 x i64> zeroinitializer, %0
  %3 = select <2 x i1> %1, <2 x i64> %0, <2 x i64> %2
  ret <2 x i64> %3
}

; The select pattern can only be detected for v4i32.
define <4 x i32> @sub_absv_32(<4 x i32> %a, <4 x i32> %b) local_unnamed_addr {
; CHECK-PWR9-LABEL: sub_absv_32:
; CHECK-PWR9:       # %bb.0: # %entry
; CHECK-PWR9-NEXT:    xvnegsp v3, v3
; CHECK-PWR9-NEXT:    xvnegsp v2, v2
; CHECK-PWR9-NEXT:    vabsduw v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: sub_absv_32:
; CHECK-PWR78:       # %bb.0: # %entry
; CHECK-PWR78-NEXT:    vsubuwm v2, v2, v3
; CHECK-PWR78-NEXT:    xxlxor v3, v3, v3
; CHECK-PWR78-NEXT:    vsubuwm v3, v3, v2
; CHECK-PWR78-NEXT:    vmaxsw v2, v2, v3
; CHECK-PWR78-NEXT:    blr
entry:
  %0 = sub nsw <4 x i32> %a, %b
  %1 = icmp sgt <4 x i32> %0, <i32 -1, i32 -1, i32 -1, i32 -1>
  %2 = sub <4 x i32> zeroinitializer, %0
  %3 = select <4 x i1> %1, <4 x i32> %0, <4 x i32> %2
  ret <4 x i32> %3
}

define <8 x i16> @sub_absv_16(<8 x i16> %a, <8 x i16> %b) local_unnamed_addr {
; CHECK-LABEL: sub_absv_16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsubuhm v2, v2, v3
; CHECK-NEXT:    xxlxor v3, v3, v3
; CHECK-NEXT:    vsubuhm v3, v3, v2
; CHECK-NEXT:    vmaxsh v2, v2, v3
; CHECK-NEXT:    blr
entry:
  %0 = sub nsw <8 x i16> %a, %b
  %1 = icmp sgt <8 x i16> %0, <i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1>
  %2 = sub <8 x i16> zeroinitializer, %0
  %3 = select <8 x i1> %1, <8 x i16> %0, <8 x i16> %2
  ret <8 x i16> %3
}

define <16 x i8> @sub_absv_8(<16 x i8> %a, <16 x i8> %b) local_unnamed_addr {
; CHECK-LABEL: sub_absv_8:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsububm v2, v2, v3
; CHECK-NEXT:    xxlxor v3, v3, v3
; CHECK-NEXT:    vsububm v3, v3, v2
; CHECK-NEXT:    vmaxsb v2, v2, v3
; CHECK-NEXT:    blr
entry:
  %0 = sub nsw <16 x i8> %a, %b
  %1 = icmp sgt <16 x i8> %0, <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>
  %2 = sub <16 x i8> zeroinitializer, %0
  %3 = select <16 x i1> %1, <16 x i8> %0, <16 x i8> %2
  ret <16 x i8> %3
}

define <8 x i16> @sub_absv_16_ext(<8 x i16> %a, <8 x i16> %b) local_unnamed_addr {
; CHECK-LABEL: sub_absv_16_ext:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vminsh v4, v2, v3
; CHECK-NEXT:    vmaxsh v2, v2, v3
; CHECK-NEXT:    vsubuhm v2, v2, v4
; CHECK-NEXT:    blr
entry:
  %0 = sext <8 x i16> %a to <8 x i32>
  %1 = sext <8 x i16> %b to <8 x i32>
  %2 = sub nsw <8 x i32> %0, %1
  %3 = icmp sgt <8 x i32> %2, <i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1>
  %4 = sub nsw <8 x i32> zeroinitializer, %2
  %5 = select <8 x i1> %3, <8 x i32> %2, <8 x i32> %4
  %6 = trunc <8 x i32> %5 to <8 x i16>
  ret <8 x i16> %6
}

define <16 x i8> @sub_absv_8_ext(<16 x i8> %a, <16 x i8> %b) local_unnamed_addr {
; CHECK-PWR9-LABEL: sub_absv_8_ext:
; CHECK-PWR9:       # %bb.0: # %entry
; CHECK-PWR9-NEXT:    vabsdub v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: sub_absv_8_ext:
; CHECK-PWR78:       # %bb.0: # %entry
; CHECK-PWR78-NEXT:    vminub v4, v2, v3
; CHECK-PWR78-NEXT:    vmaxub v2, v2, v3
; CHECK-PWR78-NEXT:    vsububm v2, v2, v4
; CHECK-PWR78-NEXT:    blr
entry:
  %0 = zext <16 x i8> %a to <16 x i32>
  %1 = zext <16 x i8> %b to <16 x i32>
  %2 = sub nsw <16 x i32> %0, %1
  %3 = tail call <16 x i32> @llvm.abs.v16i32(<16 x i32> %2, i1 true)
  %4 = trunc <16 x i32> %3 to <16 x i8>
  ret <16 x i8> %4
}

define <4 x i32> @sub_absv_vec_32(<4 x i32> %a, <4 x i32> %b) local_unnamed_addr {
; CHECK-PWR9-LABEL: sub_absv_vec_32:
; CHECK-PWR9:       # %bb.0: # %entry
; CHECK-PWR9-NEXT:    xvnegsp v3, v3
; CHECK-PWR9-NEXT:    xvnegsp v2, v2
; CHECK-PWR9-NEXT:    vabsduw v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: sub_absv_vec_32:
; CHECK-PWR78:       # %bb.0: # %entry
; CHECK-PWR78-NEXT:    vsubuwm v2, v2, v3
; CHECK-PWR78-NEXT:    xxlxor v3, v3, v3
; CHECK-PWR78-NEXT:    vsubuwm v3, v3, v2
; CHECK-PWR78-NEXT:    vmaxsw v2, v2, v3
; CHECK-PWR78-NEXT:    blr
entry:
  %sub = sub nsw <4 x i32> %a, %b
  %sub.i = sub <4 x i32> zeroinitializer, %sub
  %0 = tail call <4 x i32> @llvm.ppc.altivec.vmaxsw(<4 x i32> %sub, <4 x i32> %sub.i)
  ret <4 x i32> %0
}

define <8 x i16> @sub_absv_vec_16(<8 x i16> %a, <8 x i16> %b) local_unnamed_addr {
; CHECK-LABEL: sub_absv_vec_16:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsubuhm v2, v2, v3
; CHECK-NEXT:    xxlxor v3, v3, v3
; CHECK-NEXT:    vsubuhm v3, v3, v2
; CHECK-NEXT:    vmaxsh v2, v2, v3
; CHECK-NEXT:    blr
entry:
  %sub = sub nsw <8 x i16> %a, %b
  %sub.i = sub <8 x i16> zeroinitializer, %sub
  %0 = tail call <8 x i16> @llvm.ppc.altivec.vmaxsh(<8 x i16> %sub, <8 x i16> %sub.i)
  ret <8 x i16> %0
}

define <16 x i8> @sub_absv_vec_8(<16 x i8> %a, <16 x i8> %b) local_unnamed_addr {
; CHECK-LABEL: sub_absv_vec_8:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vsububm v2, v2, v3
; CHECK-NEXT:    xxlxor v3, v3, v3
; CHECK-NEXT:    vsububm v3, v3, v2
; CHECK-NEXT:    vmaxsb v2, v2, v3
; CHECK-NEXT:    blr
entry:
  %sub = sub nsw <16 x i8> %a, %b
  %sub.i = sub <16 x i8> zeroinitializer, %sub
  %0 = tail call <16 x i8> @llvm.ppc.altivec.vmaxsb(<16 x i8> %sub, <16 x i8> %sub.i)
  ret <16 x i8> %0
}

define <4 x i32> @zext_sub_absd32(<4 x i16>, <4 x i16>) local_unnamed_addr {
; CHECK-PWR9-LE-LABEL: zext_sub_absd32:
; CHECK-PWR9-LE:       # %bb.0:
; CHECK-PWR9-LE-NEXT:    vabsduh v2, v2, v3
; CHECK-PWR9-LE-NEXT:    xxlxor v3, v3, v3
; CHECK-PWR9-LE-NEXT:    vmrglh v2, v3, v2
; CHECK-PWR9-LE-NEXT:    blr
;
; CHECK-PWR9-BE-LABEL: zext_sub_absd32:
; CHECK-PWR9-BE:       # %bb.0:
; CHECK-PWR9-BE-NEXT:    vabsduh v2, v2, v3
; CHECK-PWR9-BE-NEXT:    xxlxor v3, v3, v3
; CHECK-PWR9-BE-NEXT:    vmrghh v2, v3, v2
; CHECK-PWR9-BE-NEXT:    blr
;
; CHECK-PWR8-LABEL: zext_sub_absd32:
; CHECK-PWR8:       # %bb.0:
; CHECK-PWR8-NEXT:    xxlxor v4, v4, v4
; CHECK-PWR8-NEXT:    vmrglh v2, v4, v2
; CHECK-PWR8-NEXT:    vmrglh v3, v4, v3
; CHECK-PWR8-NEXT:    vsubuwm v2, v2, v3
; CHECK-PWR8-NEXT:    vsubuwm v3, v4, v2
; CHECK-PWR8-NEXT:    vmaxsw v2, v2, v3
; CHECK-PWR8-NEXT:    blr
;
; CHECK-PWR7-LABEL: zext_sub_absd32:
; CHECK-PWR7:       # %bb.0:
; CHECK-PWR7-NEXT:    addis r3, r2, .LCPI13_0@toc@ha
; CHECK-PWR7-NEXT:    xxlxor v5, v5, v5
; CHECK-PWR7-NEXT:    addi r3, r3, .LCPI13_0@toc@l
; CHECK-PWR7-NEXT:    lxvw4x v4, 0, r3
; CHECK-PWR7-NEXT:    vperm v2, v5, v2, v4
; CHECK-PWR7-NEXT:    vperm v3, v5, v3, v4
; CHECK-PWR7-NEXT:    vsubuwm v2, v2, v3
; CHECK-PWR7-NEXT:    vsubuwm v3, v5, v2
; CHECK-PWR7-NEXT:    vmaxsw v2, v2, v3
; CHECK-PWR7-NEXT:    blr
    %3 = zext <4 x i16> %0 to <4 x i32>
    %4 = zext <4 x i16> %1 to <4 x i32>
    %5 = sub <4 x i32> %3, %4
    %6 = sub <4 x i32> zeroinitializer, %5
    %7 = tail call <4 x i32> @llvm.ppc.altivec.vmaxsw(<4 x i32> %5, <4 x i32> %6)
    ret <4 x i32> %7
}

define <8 x i16> @zext_sub_absd16(<8 x i8>, <8 x i8>) local_unnamed_addr {
; CHECK-PWR9-LE-LABEL: zext_sub_absd16:
; CHECK-PWR9-LE:       # %bb.0:
; CHECK-PWR9-LE-NEXT:    vabsdub v2, v2, v3
; CHECK-PWR9-LE-NEXT:    xxlxor v3, v3, v3
; CHECK-PWR9-LE-NEXT:    vmrglb v2, v3, v2
; CHECK-PWR9-LE-NEXT:    blr
;
; CHECK-PWR9-BE-LABEL: zext_sub_absd16:
; CHECK-PWR9-BE:       # %bb.0:
; CHECK-PWR9-BE-NEXT:    vabsdub v2, v2, v3
; CHECK-PWR9-BE-NEXT:    xxlxor v3, v3, v3
; CHECK-PWR9-BE-NEXT:    vmrghb v2, v3, v2
; CHECK-PWR9-BE-NEXT:    blr
;
; CHECK-PWR8-LABEL: zext_sub_absd16:
; CHECK-PWR8:       # %bb.0:
; CHECK-PWR8-NEXT:    xxlxor v4, v4, v4
; CHECK-PWR8-NEXT:    vmrglb v2, v4, v2
; CHECK-PWR8-NEXT:    vmrglb v3, v4, v3
; CHECK-PWR8-NEXT:    vsubuhm v2, v2, v3
; CHECK-PWR8-NEXT:    vsubuhm v3, v4, v2
; CHECK-PWR8-NEXT:    vmaxsh v2, v2, v3
; CHECK-PWR8-NEXT:    blr
;
; CHECK-PWR7-LABEL: zext_sub_absd16:
; CHECK-PWR7:       # %bb.0:
; CHECK-PWR7-NEXT:    addis r3, r2, .LCPI14_0@toc@ha
; CHECK-PWR7-NEXT:    xxlxor v5, v5, v5
; CHECK-PWR7-NEXT:    addi r3, r3, .LCPI14_0@toc@l
; CHECK-PWR7-NEXT:    lxvw4x v4, 0, r3
; CHECK-PWR7-NEXT:    vperm v2, v5, v2, v4
; CHECK-PWR7-NEXT:    vperm v3, v5, v3, v4
; CHECK-PWR7-NEXT:    vsubuhm v2, v2, v3
; CHECK-PWR7-NEXT:    vsubuhm v3, v5, v2
; CHECK-PWR7-NEXT:    vmaxsh v2, v2, v3
; CHECK-PWR7-NEXT:    blr
    %3 = zext <8 x i8> %0 to <8 x i16>
    %4 = zext <8 x i8> %1 to <8 x i16>
    %5 = sub <8 x i16> %3, %4
    %6 = sub <8 x i16> zeroinitializer, %5
    %7 = tail call <8 x i16> @llvm.ppc.altivec.vmaxsh(<8 x i16> %5, <8 x i16> %6)
    ret <8 x i16> %7
}

define <16 x i8> @zext_sub_absd8(<16 x i4>, <16 x i4>) local_unnamed_addr {
; CHECK-PWR9-LABEL: zext_sub_absd8:
; CHECK-PWR9:       # %bb.0:
; CHECK-PWR9-NEXT:    xxspltib vs0, 15
; CHECK-PWR9-NEXT:    xxland v3, v3, vs0
; CHECK-PWR9-NEXT:    xxland v2, v2, vs0
; CHECK-PWR9-NEXT:    vabsdub v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: zext_sub_absd8:
; CHECK-PWR78:       # %bb.0:
; CHECK-PWR78-NEXT:    vspltisb v4, 15
; CHECK-PWR78-NEXT:    xxland v2, v2, v4
; CHECK-PWR78-NEXT:    xxland v3, v3, v4
; CHECK-PWR78-NEXT:    vsububm v2, v2, v3
; CHECK-PWR78-NEXT:    xxlxor v3, v3, v3
; CHECK-PWR78-NEXT:    vsububm v3, v3, v2
; CHECK-PWR78-NEXT:    vmaxsb v2, v2, v3
; CHECK-PWR78-NEXT:    blr
    %3 = zext <16 x i4> %0 to <16 x i8>
    %4 = zext <16 x i4> %1 to <16 x i8>
    %5 = sub <16 x i8> %3, %4
    %6 = sub <16 x i8> zeroinitializer, %5
    %7 = tail call <16 x i8> @llvm.ppc.altivec.vmaxsb(<16 x i8> %5, <16 x i8> %6)
    ret <16 x i8> %7
}

define <4 x i32> @sext_sub_absd32(<4 x i16>, <4 x i16>) local_unnamed_addr {
; CHECK-PWR9-LE-LABEL: sext_sub_absd32:
; CHECK-PWR9-LE:       # %bb.0:
; CHECK-PWR9-LE-NEXT:    vminsh v4, v2, v3
; CHECK-PWR9-LE-NEXT:    vmaxsh v2, v2, v3
; CHECK-PWR9-LE-NEXT:    xxlxor v3, v3, v3
; CHECK-PWR9-LE-NEXT:    vsubuhm v2, v2, v4
; CHECK-PWR9-LE-NEXT:    vmrglh v2, v3, v2
; CHECK-PWR9-LE-NEXT:    blr
;
; CHECK-PWR9-BE-LABEL: sext_sub_absd32:
; CHECK-PWR9-BE:       # %bb.0:
; CHECK-PWR9-BE-NEXT:    vminsh v4, v2, v3
; CHECK-PWR9-BE-NEXT:    vmaxsh v2, v2, v3
; CHECK-PWR9-BE-NEXT:    xxlxor v3, v3, v3
; CHECK-PWR9-BE-NEXT:    vsubuhm v2, v2, v4
; CHECK-PWR9-BE-NEXT:    vmrghh v2, v3, v2
; CHECK-PWR9-BE-NEXT:    blr
;
; CHECK-PWR8-LABEL: sext_sub_absd32:
; CHECK-PWR8:       # %bb.0:
; CHECK-PWR8-NEXT:    vspltisw v4, 8
; CHECK-PWR8-NEXT:    vmrglh v2, v2, v2
; CHECK-PWR8-NEXT:    vadduwm v4, v4, v4
; CHECK-PWR8-NEXT:    vmrglh v3, v3, v3
; CHECK-PWR8-NEXT:    vslw v2, v2, v4
; CHECK-PWR8-NEXT:    vslw v3, v3, v4
; CHECK-PWR8-NEXT:    vsraw v2, v2, v4
; CHECK-PWR8-NEXT:    vsraw v3, v3, v4
; CHECK-PWR8-NEXT:    vsubuwm v2, v2, v3
; CHECK-PWR8-NEXT:    xxlxor v3, v3, v3
; CHECK-PWR8-NEXT:    vsubuwm v3, v3, v2
; CHECK-PWR8-NEXT:    vmaxsw v2, v2, v3
; CHECK-PWR8-NEXT:    blr
;
; CHECK-PWR7-LABEL: sext_sub_absd32:
; CHECK-PWR7:       # %bb.0:
; CHECK-PWR7-NEXT:    vspltisw v4, 8
; CHECK-PWR7-NEXT:    vmrghh v2, v2, v2
; CHECK-PWR7-NEXT:    vmrghh v3, v3, v3
; CHECK-PWR7-NEXT:    vadduwm v4, v4, v4
; CHECK-PWR7-NEXT:    vslw v2, v2, v4
; CHECK-PWR7-NEXT:    vslw v3, v3, v4
; CHECK-PWR7-NEXT:    vsraw v2, v2, v4
; CHECK-PWR7-NEXT:    vsraw v3, v3, v4
; CHECK-PWR7-NEXT:    vsubuwm v2, v2, v3
; CHECK-PWR7-NEXT:    xxlxor v3, v3, v3
; CHECK-PWR7-NEXT:    vsubuwm v3, v3, v2
; CHECK-PWR7-NEXT:    vmaxsw v2, v2, v3
; CHECK-PWR7-NEXT:    blr
    %3 = sext <4 x i16> %0 to <4 x i32>
    %4 = sext <4 x i16> %1 to <4 x i32>
    %5 = sub <4 x i32> %3, %4
    %6 = sub <4 x i32> zeroinitializer, %5
    %7 = tail call <4 x i32> @llvm.ppc.altivec.vmaxsw(<4 x i32> %5, <4 x i32> %6)
    ret <4 x i32> %7
}

define <8 x i16> @sext_sub_absd16(<8 x i8>, <8 x i8>) local_unnamed_addr {
; CHECK-PWR9-LE-LABEL: sext_sub_absd16:
; CHECK-PWR9-LE:       # %bb.0:
; CHECK-PWR9-LE-NEXT:    vminsb v4, v2, v3
; CHECK-PWR9-LE-NEXT:    vmaxsb v2, v2, v3
; CHECK-PWR9-LE-NEXT:    xxlxor v3, v3, v3
; CHECK-PWR9-LE-NEXT:    vsububm v2, v2, v4
; CHECK-PWR9-LE-NEXT:    vmrglb v2, v3, v2
; CHECK-PWR9-LE-NEXT:    blr
;
; CHECK-PWR9-BE-LABEL: sext_sub_absd16:
; CHECK-PWR9-BE:       # %bb.0:
; CHECK-PWR9-BE-NEXT:    vminsb v4, v2, v3
; CHECK-PWR9-BE-NEXT:    vmaxsb v2, v2, v3
; CHECK-PWR9-BE-NEXT:    xxlxor v3, v3, v3
; CHECK-PWR9-BE-NEXT:    vsububm v2, v2, v4
; CHECK-PWR9-BE-NEXT:    vmrghb v2, v3, v2
; CHECK-PWR9-BE-NEXT:    blr
;
; CHECK-PWR8-LABEL: sext_sub_absd16:
; CHECK-PWR8:       # %bb.0:
; CHECK-PWR8-NEXT:    vmrglb v2, v2, v2
; CHECK-PWR8-NEXT:    vspltish v4, 8
; CHECK-PWR8-NEXT:    vslh v2, v2, v4
; CHECK-PWR8-NEXT:    vmrglb v3, v3, v3
; CHECK-PWR8-NEXT:    vslh v3, v3, v4
; CHECK-PWR8-NEXT:    vsrah v2, v2, v4
; CHECK-PWR8-NEXT:    vsrah v3, v3, v4
; CHECK-PWR8-NEXT:    vsubuhm v2, v2, v3
; CHECK-PWR8-NEXT:    xxlxor v3, v3, v3
; CHECK-PWR8-NEXT:    vsubuhm v3, v3, v2
; CHECK-PWR8-NEXT:    vmaxsh v2, v2, v3
; CHECK-PWR8-NEXT:    blr
;
; CHECK-PWR7-LABEL: sext_sub_absd16:
; CHECK-PWR7:       # %bb.0:
; CHECK-PWR7-NEXT:    vspltish v4, 8
; CHECK-PWR7-NEXT:    vmrghb v2, v2, v2
; CHECK-PWR7-NEXT:    vmrghb v3, v3, v3
; CHECK-PWR7-NEXT:    vslh v2, v2, v4
; CHECK-PWR7-NEXT:    vslh v3, v3, v4
; CHECK-PWR7-NEXT:    vsrah v2, v2, v4
; CHECK-PWR7-NEXT:    vsrah v3, v3, v4
; CHECK-PWR7-NEXT:    vsubuhm v2, v2, v3
; CHECK-PWR7-NEXT:    xxlxor v3, v3, v3
; CHECK-PWR7-NEXT:    vsubuhm v3, v3, v2
; CHECK-PWR7-NEXT:    vmaxsh v2, v2, v3
; CHECK-PWR7-NEXT:    blr
    %3 = sext <8 x i8> %0 to <8 x i16>
    %4 = sext <8 x i8> %1 to <8 x i16>
    %5 = sub <8 x i16> %3, %4
    %6 = sub <8 x i16> zeroinitializer, %5
    %7 = tail call <8 x i16> @llvm.ppc.altivec.vmaxsh(<8 x i16> %5, <8 x i16> %6)
    ret <8 x i16> %7
}

define <16 x i8> @sext_sub_absd8(<16 x i4>, <16 x i4>) local_unnamed_addr {
; CHECK-PWR9-LABEL: sext_sub_absd8:
; CHECK-PWR9:       # %bb.0:
; CHECK-PWR9-NEXT:    xxspltib v4, 4
; CHECK-PWR9-NEXT:    vslb v3, v3, v4
; CHECK-PWR9-NEXT:    vslb v2, v2, v4
; CHECK-PWR9-NEXT:    vsrab v3, v3, v4
; CHECK-PWR9-NEXT:    vsrab v2, v2, v4
; CHECK-PWR9-NEXT:    vminsb v4, v2, v3
; CHECK-PWR9-NEXT:    vmaxsb v2, v2, v3
; CHECK-PWR9-NEXT:    vsububm v2, v2, v4
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: sext_sub_absd8:
; CHECK-PWR78:       # %bb.0:
; CHECK-PWR78-NEXT:    vspltisb v4, 4
; CHECK-PWR78-NEXT:    vslb v2, v2, v4
; CHECK-PWR78-NEXT:    vslb v3, v3, v4
; CHECK-PWR78-NEXT:    vsrab v2, v2, v4
; CHECK-PWR78-NEXT:    vsrab v3, v3, v4
; CHECK-PWR78-NEXT:    vsububm v2, v2, v3
; CHECK-PWR78-NEXT:    xxlxor v3, v3, v3
; CHECK-PWR78-NEXT:    vsububm v3, v3, v2
; CHECK-PWR78-NEXT:    vmaxsb v2, v2, v3
; CHECK-PWR78-NEXT:    blr
    %3 = sext <16 x i4> %0 to <16 x i8>
    %4 = sext <16 x i4> %1 to <16 x i8>
    %5 = sub <16 x i8> %3, %4
    %6 = sub <16 x i8> zeroinitializer, %5
    %7 = tail call <16 x i8> @llvm.ppc.altivec.vmaxsb(<16 x i8> %5, <16 x i8> %6)
    ret <16 x i8> %7
}

; To verify vabsdu* exploitation for ucmp + sub + select sequence

define <4 x i32> @absd_int32_ugt(<4 x i32>, <4 x i32>) {
; CHECK-PWR9-LABEL: absd_int32_ugt:
; CHECK-PWR9:       # %bb.0:
; CHECK-PWR9-NEXT:    vabsduw v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: absd_int32_ugt:
; CHECK-PWR78:       # %bb.0:
; CHECK-PWR78-NEXT:    vminuw v4, v2, v3
; CHECK-PWR78-NEXT:    vmaxuw v2, v2, v3
; CHECK-PWR78-NEXT:    vsubuwm v2, v2, v4
; CHECK-PWR78-NEXT:    blr
  %3 = icmp ugt <4 x i32> %0, %1
  %4 = sub <4 x i32> %0, %1
  %5 = sub <4 x i32> %1, %0
  %6 = select <4 x i1> %3, <4 x i32> %4, <4 x i32> %5
  ret <4 x i32> %6
}

define <4 x i32> @absd_int32_uge(<4 x i32>, <4 x i32>) {
; CHECK-PWR9-LABEL: absd_int32_uge:
; CHECK-PWR9:       # %bb.0:
; CHECK-PWR9-NEXT:    vabsduw v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: absd_int32_uge:
; CHECK-PWR78:       # %bb.0:
; CHECK-PWR78-NEXT:    vminuw v4, v2, v3
; CHECK-PWR78-NEXT:    vmaxuw v2, v2, v3
; CHECK-PWR78-NEXT:    vsubuwm v2, v2, v4
; CHECK-PWR78-NEXT:    blr
  %3 = icmp uge <4 x i32> %0, %1
  %4 = sub <4 x i32> %0, %1
  %5 = sub <4 x i32> %1, %0
  %6 = select <4 x i1> %3, <4 x i32> %4, <4 x i32> %5
  ret <4 x i32> %6
}

define <4 x i32> @absd_int32_ult(<4 x i32>, <4 x i32>) {
; CHECK-PWR9-LABEL: absd_int32_ult:
; CHECK-PWR9:       # %bb.0:
; CHECK-PWR9-NEXT:    vabsduw v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: absd_int32_ult:
; CHECK-PWR78:       # %bb.0:
; CHECK-PWR78-NEXT:    vminuw v4, v2, v3
; CHECK-PWR78-NEXT:    vmaxuw v2, v2, v3
; CHECK-PWR78-NEXT:    vsubuwm v2, v2, v4
; CHECK-PWR78-NEXT:    blr
  %3 = icmp ult <4 x i32> %0, %1
  %4 = sub <4 x i32> %0, %1
  %5 = sub <4 x i32> %1, %0
  %6 = select <4 x i1> %3, <4 x i32> %5, <4 x i32> %4
  ret <4 x i32> %6
}

define <4 x i32> @absd_int32_ule(<4 x i32>, <4 x i32>) {
; CHECK-PWR9-LABEL: absd_int32_ule:
; CHECK-PWR9:       # %bb.0:
; CHECK-PWR9-NEXT:    vabsduw v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: absd_int32_ule:
; CHECK-PWR78:       # %bb.0:
; CHECK-PWR78-NEXT:    vminuw v4, v2, v3
; CHECK-PWR78-NEXT:    vmaxuw v2, v2, v3
; CHECK-PWR78-NEXT:    vsubuwm v2, v2, v4
; CHECK-PWR78-NEXT:    blr
  %3 = icmp ule <4 x i32> %0, %1
  %4 = sub <4 x i32> %0, %1
  %5 = sub <4 x i32> %1, %0
  %6 = select <4 x i1> %3, <4 x i32> %5, <4 x i32> %4
  ret <4 x i32> %6
}

define <8 x i16> @absd_int16_ugt(<8 x i16>, <8 x i16>) {
; CHECK-PWR9-LABEL: absd_int16_ugt:
; CHECK-PWR9:       # %bb.0:
; CHECK-PWR9-NEXT:    vabsduh v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: absd_int16_ugt:
; CHECK-PWR78:       # %bb.0:
; CHECK-PWR78-NEXT:    vminuh v4, v2, v3
; CHECK-PWR78-NEXT:    vmaxuh v2, v2, v3
; CHECK-PWR78-NEXT:    vsubuhm v2, v2, v4
; CHECK-PWR78-NEXT:    blr
  %3 = icmp ugt <8 x i16> %0, %1
  %4 = sub <8 x i16> %0, %1
  %5 = sub <8 x i16> %1, %0
  %6 = select <8 x i1> %3, <8 x i16> %4, <8 x i16> %5
  ret <8 x i16> %6
}

define <8 x i16> @absd_int16_uge(<8 x i16>, <8 x i16>) {
; CHECK-PWR9-LABEL: absd_int16_uge:
; CHECK-PWR9:       # %bb.0:
; CHECK-PWR9-NEXT:    vabsduh v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: absd_int16_uge:
; CHECK-PWR78:       # %bb.0:
; CHECK-PWR78-NEXT:    vminuh v4, v2, v3
; CHECK-PWR78-NEXT:    vmaxuh v2, v2, v3
; CHECK-PWR78-NEXT:    vsubuhm v2, v2, v4
; CHECK-PWR78-NEXT:    blr
  %3 = icmp uge <8 x i16> %0, %1
  %4 = sub <8 x i16> %0, %1
  %5 = sub <8 x i16> %1, %0
  %6 = select <8 x i1> %3, <8 x i16> %4, <8 x i16> %5
  ret <8 x i16> %6
}

define <8 x i16> @absd_int16_ult(<8 x i16>, <8 x i16>) {
; CHECK-PWR9-LABEL: absd_int16_ult:
; CHECK-PWR9:       # %bb.0:
; CHECK-PWR9-NEXT:    vabsduh v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: absd_int16_ult:
; CHECK-PWR78:       # %bb.0:
; CHECK-PWR78-NEXT:    vminuh v4, v2, v3
; CHECK-PWR78-NEXT:    vmaxuh v2, v2, v3
; CHECK-PWR78-NEXT:    vsubuhm v2, v2, v4
; CHECK-PWR78-NEXT:    blr
  %3 = icmp ult <8 x i16> %0, %1
  %4 = sub <8 x i16> %0, %1
  %5 = sub <8 x i16> %1, %0
  %6 = select <8 x i1> %3, <8 x i16> %5, <8 x i16> %4
  ret <8 x i16> %6
}

define <8 x i16> @absd_int16_ule(<8 x i16>, <8 x i16>) {
; CHECK-PWR9-LABEL: absd_int16_ule:
; CHECK-PWR9:       # %bb.0:
; CHECK-PWR9-NEXT:    vabsduh v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: absd_int16_ule:
; CHECK-PWR78:       # %bb.0:
; CHECK-PWR78-NEXT:    vminuh v4, v2, v3
; CHECK-PWR78-NEXT:    vmaxuh v2, v2, v3
; CHECK-PWR78-NEXT:    vsubuhm v2, v2, v4
; CHECK-PWR78-NEXT:    blr
  %3 = icmp ule <8 x i16> %0, %1
  %4 = sub <8 x i16> %0, %1
  %5 = sub <8 x i16> %1, %0
  %6 = select <8 x i1> %3, <8 x i16> %5, <8 x i16> %4
  ret <8 x i16> %6
}

define <16 x i8> @absd_int8_ugt(<16 x i8>, <16 x i8>) {
; CHECK-PWR9-LABEL: absd_int8_ugt:
; CHECK-PWR9:       # %bb.0:
; CHECK-PWR9-NEXT:    vabsdub v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: absd_int8_ugt:
; CHECK-PWR78:       # %bb.0:
; CHECK-PWR78-NEXT:    vminub v4, v2, v3
; CHECK-PWR78-NEXT:    vmaxub v2, v2, v3
; CHECK-PWR78-NEXT:    vsububm v2, v2, v4
; CHECK-PWR78-NEXT:    blr
  %3 = icmp ugt <16 x i8> %0, %1
  %4 = sub <16 x i8> %0, %1
  %5 = sub <16 x i8> %1, %0
  %6 = select <16 x i1> %3, <16 x i8> %4, <16 x i8> %5
  ret <16 x i8> %6
}

define <16 x i8> @absd_int8_uge(<16 x i8>, <16 x i8>) {
; CHECK-PWR9-LABEL: absd_int8_uge:
; CHECK-PWR9:       # %bb.0:
; CHECK-PWR9-NEXT:    vabsdub v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: absd_int8_uge:
; CHECK-PWR78:       # %bb.0:
; CHECK-PWR78-NEXT:    vminub v4, v2, v3
; CHECK-PWR78-NEXT:    vmaxub v2, v2, v3
; CHECK-PWR78-NEXT:    vsububm v2, v2, v4
; CHECK-PWR78-NEXT:    blr
  %3 = icmp uge <16 x i8> %0, %1
  %4 = sub <16 x i8> %0, %1
  %5 = sub <16 x i8> %1, %0
  %6 = select <16 x i1> %3, <16 x i8> %4, <16 x i8> %5
  ret <16 x i8> %6
}

define <16 x i8> @absd_int8_ult(<16 x i8>, <16 x i8>) {
; CHECK-PWR9-LABEL: absd_int8_ult:
; CHECK-PWR9:       # %bb.0:
; CHECK-PWR9-NEXT:    vabsdub v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: absd_int8_ult:
; CHECK-PWR78:       # %bb.0:
; CHECK-PWR78-NEXT:    vminub v4, v2, v3
; CHECK-PWR78-NEXT:    vmaxub v2, v2, v3
; CHECK-PWR78-NEXT:    vsububm v2, v2, v4
; CHECK-PWR78-NEXT:    blr
  %3 = icmp ult <16 x i8> %0, %1
  %4 = sub <16 x i8> %0, %1
  %5 = sub <16 x i8> %1, %0
  %6 = select <16 x i1> %3, <16 x i8> %5, <16 x i8> %4
  ret <16 x i8> %6
}

define <16 x i8> @absd_int8_ule(<16 x i8>, <16 x i8>) {
; CHECK-PWR9-LABEL: absd_int8_ule:
; CHECK-PWR9:       # %bb.0:
; CHECK-PWR9-NEXT:    vabsdub v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: absd_int8_ule:
; CHECK-PWR78:       # %bb.0:
; CHECK-PWR78-NEXT:    vminub v4, v2, v3
; CHECK-PWR78-NEXT:    vmaxub v2, v2, v3
; CHECK-PWR78-NEXT:    vsububm v2, v2, v4
; CHECK-PWR78-NEXT:    blr
  %3 = icmp ule <16 x i8> %0, %1
  %4 = sub <16 x i8> %0, %1
  %5 = sub <16 x i8> %1, %0
  %6 = select <16 x i1> %3, <16 x i8> %5, <16 x i8> %4
  ret <16 x i8> %6
}

; Tests for ABDS icmp + sub + select sequence

define <4 x i32> @absd_int32_sgt(<4 x i32>, <4 x i32>) {
; CHECK-PWR9-LABEL: absd_int32_sgt:
; CHECK-PWR9:       # %bb.0:
; CHECK-PWR9-NEXT:    xvnegsp v3, v3
; CHECK-PWR9-NEXT:    xvnegsp v2, v2
; CHECK-PWR9-NEXT:    vabsduw v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: absd_int32_sgt:
; CHECK-PWR78:       # %bb.0:
; CHECK-PWR78-NEXT:    vminsw v4, v2, v3
; CHECK-PWR78-NEXT:    vmaxsw v2, v2, v3
; CHECK-PWR78-NEXT:    vsubuwm v2, v2, v4
; CHECK-PWR78-NEXT:    blr
  %3 = icmp sgt <4 x i32> %0, %1
  %4 = sub <4 x i32> %0, %1
  %5 = sub <4 x i32> %1, %0
  %6 = select <4 x i1> %3, <4 x i32> %4, <4 x i32> %5
  ret <4 x i32> %6
}

define <4 x i32> @absd_int32_sge(<4 x i32>, <4 x i32>) {
; CHECK-PWR9-LABEL: absd_int32_sge:
; CHECK-PWR9:       # %bb.0:
; CHECK-PWR9-NEXT:    xvnegsp v3, v3
; CHECK-PWR9-NEXT:    xvnegsp v2, v2
; CHECK-PWR9-NEXT:    vabsduw v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: absd_int32_sge:
; CHECK-PWR78:       # %bb.0:
; CHECK-PWR78-NEXT:    vminsw v4, v2, v3
; CHECK-PWR78-NEXT:    vmaxsw v2, v2, v3
; CHECK-PWR78-NEXT:    vsubuwm v2, v2, v4
; CHECK-PWR78-NEXT:    blr
  %3 = icmp sge <4 x i32> %0, %1
  %4 = sub <4 x i32> %0, %1
  %5 = sub <4 x i32> %1, %0
  %6 = select <4 x i1> %3, <4 x i32> %4, <4 x i32> %5
  ret <4 x i32> %6
}

define <4 x i32> @absd_int32_slt(<4 x i32>, <4 x i32>) {
; CHECK-PWR9-LABEL: absd_int32_slt:
; CHECK-PWR9:       # %bb.0:
; CHECK-PWR9-NEXT:    xvnegsp v3, v3
; CHECK-PWR9-NEXT:    xvnegsp v2, v2
; CHECK-PWR9-NEXT:    vabsduw v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: absd_int32_slt:
; CHECK-PWR78:       # %bb.0:
; CHECK-PWR78-NEXT:    vminsw v4, v2, v3
; CHECK-PWR78-NEXT:    vmaxsw v2, v2, v3
; CHECK-PWR78-NEXT:    vsubuwm v2, v2, v4
; CHECK-PWR78-NEXT:    blr
  %3 = icmp slt <4 x i32> %0, %1
  %4 = sub <4 x i32> %0, %1
  %5 = sub <4 x i32> %1, %0
  %6 = select <4 x i1> %3, <4 x i32> %5, <4 x i32> %4
  ret <4 x i32> %6
}

define <4 x i32> @absd_int32_sle(<4 x i32>, <4 x i32>) {
; CHECK-PWR9-LABEL: absd_int32_sle:
; CHECK-PWR9:       # %bb.0:
; CHECK-PWR9-NEXT:    xvnegsp v3, v3
; CHECK-PWR9-NEXT:    xvnegsp v2, v2
; CHECK-PWR9-NEXT:    vabsduw v2, v2, v3
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: absd_int32_sle:
; CHECK-PWR78:       # %bb.0:
; CHECK-PWR78-NEXT:    vminsw v4, v2, v3
; CHECK-PWR78-NEXT:    vmaxsw v2, v2, v3
; CHECK-PWR78-NEXT:    vsubuwm v2, v2, v4
; CHECK-PWR78-NEXT:    blr
  %3 = icmp sle <4 x i32> %0, %1
  %4 = sub <4 x i32> %0, %1
  %5 = sub <4 x i32> %1, %0
  %6 = select <4 x i1> %3, <4 x i32> %5, <4 x i32> %4
  ret <4 x i32> %6
}

define <8 x i16> @absd_int16_sgt(<8 x i16>, <8 x i16>) {
; CHECK-LABEL: absd_int16_sgt:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vminsh v4, v2, v3
; CHECK-NEXT:    vmaxsh v2, v2, v3
; CHECK-NEXT:    vsubuhm v2, v2, v4
; CHECK-NEXT:    blr
  %3 = icmp sgt <8 x i16> %0, %1
  %4 = sub <8 x i16> %0, %1
  %5 = sub <8 x i16> %1, %0
  %6 = select <8 x i1> %3, <8 x i16> %4, <8 x i16> %5
  ret <8 x i16> %6
}

define <8 x i16> @absd_int16_sge(<8 x i16>, <8 x i16>) {
; CHECK-LABEL: absd_int16_sge:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vminsh v4, v2, v3
; CHECK-NEXT:    vmaxsh v2, v2, v3
; CHECK-NEXT:    vsubuhm v2, v2, v4
; CHECK-NEXT:    blr
  %3 = icmp sge <8 x i16> %0, %1
  %4 = sub <8 x i16> %0, %1
  %5 = sub <8 x i16> %1, %0
  %6 = select <8 x i1> %3, <8 x i16> %4, <8 x i16> %5
  ret <8 x i16> %6
}

define <8 x i16> @absd_int16_slt(<8 x i16>, <8 x i16>) {
; CHECK-LABEL: absd_int16_slt:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vminsh v4, v2, v3
; CHECK-NEXT:    vmaxsh v2, v2, v3
; CHECK-NEXT:    vsubuhm v2, v2, v4
; CHECK-NEXT:    blr
  %3 = icmp slt <8 x i16> %0, %1
  %4 = sub <8 x i16> %0, %1
  %5 = sub <8 x i16> %1, %0
  %6 = select <8 x i1> %3, <8 x i16> %5, <8 x i16> %4
  ret <8 x i16> %6
}

define <8 x i16> @absd_int16_sle(<8 x i16>, <8 x i16>) {
; CHECK-LABEL: absd_int16_sle:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vminsh v4, v2, v3
; CHECK-NEXT:    vmaxsh v2, v2, v3
; CHECK-NEXT:    vsubuhm v2, v2, v4
; CHECK-NEXT:    blr
  %3 = icmp sle <8 x i16> %0, %1
  %4 = sub <8 x i16> %0, %1
  %5 = sub <8 x i16> %1, %0
  %6 = select <8 x i1> %3, <8 x i16> %5, <8 x i16> %4
  ret <8 x i16> %6
}

define <16 x i8> @absd_int8_sgt(<16 x i8>, <16 x i8>) {
; CHECK-LABEL: absd_int8_sgt:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vminsb v4, v2, v3
; CHECK-NEXT:    vmaxsb v2, v2, v3
; CHECK-NEXT:    vsububm v2, v2, v4
; CHECK-NEXT:    blr
  %3 = icmp sgt <16 x i8> %0, %1
  %4 = sub <16 x i8> %0, %1
  %5 = sub <16 x i8> %1, %0
  %6 = select <16 x i1> %3, <16 x i8> %4, <16 x i8> %5
  ret <16 x i8> %6
}

define <16 x i8> @absd_int8_sge(<16 x i8>, <16 x i8>) {
; CHECK-LABEL: absd_int8_sge:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vminsb v4, v2, v3
; CHECK-NEXT:    vmaxsb v2, v2, v3
; CHECK-NEXT:    vsububm v2, v2, v4
; CHECK-NEXT:    blr
  %3 = icmp sge <16 x i8> %0, %1
  %4 = sub <16 x i8> %0, %1
  %5 = sub <16 x i8> %1, %0
  %6 = select <16 x i1> %3, <16 x i8> %4, <16 x i8> %5
  ret <16 x i8> %6
}

define <16 x i8> @absd_int8_slt(<16 x i8>, <16 x i8>) {
; CHECK-LABEL: absd_int8_slt:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vminsb v4, v2, v3
; CHECK-NEXT:    vmaxsb v2, v2, v3
; CHECK-NEXT:    vsububm v2, v2, v4
; CHECK-NEXT:    blr
  %3 = icmp slt <16 x i8> %0, %1
  %4 = sub <16 x i8> %0, %1
  %5 = sub <16 x i8> %1, %0
  %6 = select <16 x i1> %3, <16 x i8> %5, <16 x i8> %4
  ret <16 x i8> %6
}

define <16 x i8> @absd_int8_sle(<16 x i8>, <16 x i8>) {
; CHECK-LABEL: absd_int8_sle:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vminsb v4, v2, v3
; CHECK-NEXT:    vmaxsb v2, v2, v3
; CHECK-NEXT:    vsububm v2, v2, v4
; CHECK-NEXT:    blr
  %3 = icmp sle <16 x i8> %0, %1
  %4 = sub <16 x i8> %0, %1
  %5 = sub <16 x i8> %1, %0
  %6 = select <16 x i1> %3, <16 x i8> %5, <16 x i8> %4
  ret <16 x i8> %6
}

; some cases we are unable to optimize
; check whether goes beyond the scope
define <4 x i32> @absd_int32_ugt_opp(<4 x i32>, <4 x i32>) {
; CHECK-PWR9-LABEL: absd_int32_ugt_opp:
; CHECK-PWR9:       # %bb.0:
; CHECK-PWR9-NEXT:    vabsduw v2, v2, v3
; CHECK-PWR9-NEXT:    vnegw v2, v2
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR78-LABEL: absd_int32_ugt_opp:
; CHECK-PWR78:       # %bb.0:
; CHECK-PWR78-NEXT:    vcmpgtuw v4, v2, v3
; CHECK-PWR78-NEXT:    vsubuwm v5, v2, v3
; CHECK-PWR78-NEXT:    vsubuwm v2, v3, v2
; CHECK-PWR78-NEXT:    xxsel v2, v5, v2, v4
; CHECK-PWR78-NEXT:    blr
  %3 = icmp ugt <4 x i32> %0, %1
  %4 = sub <4 x i32> %0, %1
  %5 = sub <4 x i32> %1, %0
  %6 = select <4 x i1> %3, <4 x i32> %5, <4 x i32> %4
  ret <4 x i32> %6
}

define <2 x i64> @absd_int64_ugt(<2 x i64>, <2 x i64>) {
; CHECK-PWR9-LABEL: absd_int64_ugt:
; CHECK-PWR9:       # %bb.0:
; CHECK-PWR9-NEXT:    vminud v4, v2, v3
; CHECK-PWR9-NEXT:    vmaxud v2, v2, v3
; CHECK-PWR9-NEXT:    vsubudm v2, v2, v4
; CHECK-PWR9-NEXT:    blr
;
; CHECK-PWR8-LABEL: absd_int64_ugt:
; CHECK-PWR8:       # %bb.0:
; CHECK-PWR8-NEXT:    vminud v4, v2, v3
; CHECK-PWR8-NEXT:    vmaxud v2, v2, v3
; CHECK-PWR8-NEXT:    vsubudm v2, v2, v4
; CHECK-PWR8-NEXT:    blr
;
; CHECK-PWR7-LABEL: absd_int64_ugt:
; CHECK-PWR7:       # %bb.0:
; CHECK-PWR7-NEXT:    addi r3, r1, -96
; CHECK-PWR7-NEXT:    stxvd2x v2, 0, r3
; CHECK-PWR7-NEXT:    addi r3, r1, -80
; CHECK-PWR7-NEXT:    stxvd2x v3, 0, r3
; CHECK-PWR7-NEXT:    ld r3, -88(r1)
; CHECK-PWR7-NEXT:    ld r4, -72(r1)
; CHECK-PWR7-NEXT:    ld r6, -80(r1)
; CHECK-PWR7-NEXT:    sub r5, r3, r4
; CHECK-PWR7-NEXT:    cmpld r3, r4
; CHECK-PWR7-NEXT:    li r3, 0
; CHECK-PWR7-NEXT:    li r4, -1
; CHECK-PWR7-NEXT:    std r5, -56(r1)
; CHECK-PWR7-NEXT:    ld r5, -96(r1)
; CHECK-PWR7-NEXT:    sub r7, r5, r6
; CHECK-PWR7-NEXT:    std r7, -64(r1)
; CHECK-PWR7-NEXT:    iselgt r7, r4, r3
; CHECK-PWR7-NEXT:    cmpld r5, r6
; CHECK-PWR7-NEXT:    std r7, -40(r1)
; CHECK-PWR7-NEXT:    iselgt r3, r4, r3
; CHECK-PWR7-NEXT:    addi r4, r1, -64
; CHECK-PWR7-NEXT:    std r3, -48(r1)
; CHECK-PWR7-NEXT:    lxvw4x vs0, 0, r4
; CHECK-PWR7-NEXT:    addi r4, r1, -48
; CHECK-PWR7-NEXT:    lxvw4x vs1, 0, r4
; CHECK-PWR7-NEXT:    addi r4, r1, -32
; CHECK-PWR7-NEXT:    xxlxor vs0, vs0, vs1
; CHECK-PWR7-NEXT:    stxvw4x vs0, 0, r4
; CHECK-PWR7-NEXT:    ld r4, -24(r1)
; CHECK-PWR7-NEXT:    sub r4, r7, r4
; CHECK-PWR7-NEXT:    std r4, -8(r1)
; CHECK-PWR7-NEXT:    ld r4, -32(r1)
; CHECK-PWR7-NEXT:    sub r3, r3, r4
; CHECK-PWR7-NEXT:    std r3, -16(r1)
; CHECK-PWR7-NEXT:    addi r3, r1, -16
; CHECK-PWR7-NEXT:    lxvd2x v2, 0, r3
; CHECK-PWR7-NEXT:    blr
  %3 = icmp ugt <2 x i64> %0, %1
  %4 = sub <2 x i64> %0, %1
  %5 = sub <2 x i64> %1, %0
  %6 = select <2 x i1> %3, <2 x i64> %4, <2 x i64> %5
  ret <2 x i64> %6
}

declare <4 x i32> @llvm.ppc.altivec.vmaxsw(<4 x i32>, <4 x i32>)

declare <8 x i16> @llvm.ppc.altivec.vmaxsh(<8 x i16>, <8 x i16>)

declare <16 x i8> @llvm.ppc.altivec.vmaxsb(<16 x i8>, <16 x i8>)