llvm/test/CodeGen/ARM/vminmax.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
; RUN: llc -mtriple=arm-eabi -mattr=+neon %s -o - | FileCheck %s

define <8 x i8> @vmins8(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vmins8:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vldr d16, [r1]
; CHECK-NEXT:    vldr d17, [r0]
; CHECK-NEXT:    vmin.s8 d16, d17, d16
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <8 x i8>, ptr %A
  %tmp2 = load <8 x i8>, ptr %B
  %tmp3 = call <8 x i8> @llvm.arm.neon.vmins.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
  ret <8 x i8> %tmp3
}

define <4 x i16> @vmins16(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vmins16:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vldr d16, [r1]
; CHECK-NEXT:    vldr d17, [r0]
; CHECK-NEXT:    vmin.s16 d16, d17, d16
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <4 x i16>, ptr %A
  %tmp2 = load <4 x i16>, ptr %B
  %tmp3 = call <4 x i16> @llvm.arm.neon.vmins.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
  ret <4 x i16> %tmp3
}

define <2 x i32> @vmins32(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vmins32:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vldr d16, [r1]
; CHECK-NEXT:    vldr d17, [r0]
; CHECK-NEXT:    vmin.s32 d16, d17, d16
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <2 x i32>, ptr %A
  %tmp2 = load <2 x i32>, ptr %B
  %tmp3 = call <2 x i32> @llvm.arm.neon.vmins.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
  ret <2 x i32> %tmp3
}

define <8 x i8> @vminu8(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vminu8:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vldr d16, [r1]
; CHECK-NEXT:    vldr d17, [r0]
; CHECK-NEXT:    vmin.u8 d16, d17, d16
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <8 x i8>, ptr %A
  %tmp2 = load <8 x i8>, ptr %B
  %tmp3 = call <8 x i8> @llvm.arm.neon.vminu.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
  ret <8 x i8> %tmp3
}

define <4 x i16> @vminu16(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vminu16:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vldr d16, [r1]
; CHECK-NEXT:    vldr d17, [r0]
; CHECK-NEXT:    vmin.u16 d16, d17, d16
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <4 x i16>, ptr %A
  %tmp2 = load <4 x i16>, ptr %B
  %tmp3 = call <4 x i16> @llvm.arm.neon.vminu.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
  ret <4 x i16> %tmp3
}

define <2 x i32> @vminu32(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vminu32:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vldr d16, [r1]
; CHECK-NEXT:    vldr d17, [r0]
; CHECK-NEXT:    vmin.u32 d16, d17, d16
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <2 x i32>, ptr %A
  %tmp2 = load <2 x i32>, ptr %B
  %tmp3 = call <2 x i32> @llvm.arm.neon.vminu.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
  ret <2 x i32> %tmp3
}

define <2 x float> @vminf32(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vminf32:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vldr d16, [r1]
; CHECK-NEXT:    vldr d17, [r0]
; CHECK-NEXT:    vmin.f32 d16, d17, d16
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <2 x float>, ptr %A
  %tmp2 = load <2 x float>, ptr %B
  %tmp3 = call <2 x float> @llvm.arm.neon.vmins.v2f32(<2 x float> %tmp1, <2 x float> %tmp2)
  ret <2 x float> %tmp3
}

define <16 x i8> @vminQs8(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vminQs8:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vld1.64 {d16, d17}, [r1]
; CHECK-NEXT:    vld1.64 {d18, d19}, [r0]
; CHECK-NEXT:    vmin.s8 q8, q9, q8
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    vmov r2, r3, d17
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <16 x i8>, ptr %A
  %tmp2 = load <16 x i8>, ptr %B
  %tmp3 = call <16 x i8> @llvm.arm.neon.vmins.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
  ret <16 x i8> %tmp3
}

define <8 x i16> @vminQs16(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vminQs16:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vld1.64 {d16, d17}, [r1]
; CHECK-NEXT:    vld1.64 {d18, d19}, [r0]
; CHECK-NEXT:    vmin.s16 q8, q9, q8
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    vmov r2, r3, d17
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <8 x i16>, ptr %A
  %tmp2 = load <8 x i16>, ptr %B
  %tmp3 = call <8 x i16> @llvm.arm.neon.vmins.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
  ret <8 x i16> %tmp3
}

define <4 x i32> @vminQs32(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vminQs32:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vld1.64 {d16, d17}, [r1]
; CHECK-NEXT:    vld1.64 {d18, d19}, [r0]
; CHECK-NEXT:    vmin.s32 q8, q9, q8
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    vmov r2, r3, d17
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <4 x i32>, ptr %A
  %tmp2 = load <4 x i32>, ptr %B
  %tmp3 = call <4 x i32> @llvm.arm.neon.vmins.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
  ret <4 x i32> %tmp3
}

define <16 x i8> @vminQu8(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vminQu8:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vld1.64 {d16, d17}, [r1]
; CHECK-NEXT:    vld1.64 {d18, d19}, [r0]
; CHECK-NEXT:    vmin.u8 q8, q9, q8
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    vmov r2, r3, d17
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <16 x i8>, ptr %A
  %tmp2 = load <16 x i8>, ptr %B
  %tmp3 = call <16 x i8> @llvm.arm.neon.vminu.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
  ret <16 x i8> %tmp3
}

define <8 x i16> @vminQu16(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vminQu16:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vld1.64 {d16, d17}, [r1]
; CHECK-NEXT:    vld1.64 {d18, d19}, [r0]
; CHECK-NEXT:    vmin.u16 q8, q9, q8
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    vmov r2, r3, d17
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <8 x i16>, ptr %A
  %tmp2 = load <8 x i16>, ptr %B
  %tmp3 = call <8 x i16> @llvm.arm.neon.vminu.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
  ret <8 x i16> %tmp3
}

define <4 x i32> @vminQu32(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vminQu32:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vld1.64 {d16, d17}, [r1]
; CHECK-NEXT:    vld1.64 {d18, d19}, [r0]
; CHECK-NEXT:    vmin.u32 q8, q9, q8
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    vmov r2, r3, d17
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <4 x i32>, ptr %A
  %tmp2 = load <4 x i32>, ptr %B
  %tmp3 = call <4 x i32> @llvm.arm.neon.vminu.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
  ret <4 x i32> %tmp3
}

define <4 x float> @vminQf32(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vminQf32:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vld1.64 {d16, d17}, [r1]
; CHECK-NEXT:    vld1.64 {d18, d19}, [r0]
; CHECK-NEXT:    vmin.f32 q8, q9, q8
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    vmov r2, r3, d17
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <4 x float>, ptr %A
  %tmp2 = load <4 x float>, ptr %B
  %tmp3 = call <4 x float> @llvm.arm.neon.vmins.v4f32(<4 x float> %tmp1, <4 x float> %tmp2)
  ret <4 x float> %tmp3
}

declare <8 x i8>  @llvm.arm.neon.vmins.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
declare <4 x i16> @llvm.arm.neon.vmins.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
declare <2 x i32> @llvm.arm.neon.vmins.v2i32(<2 x i32>, <2 x i32>) nounwind readnone

declare <8 x i8>  @llvm.arm.neon.vminu.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
declare <4 x i16> @llvm.arm.neon.vminu.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
declare <2 x i32> @llvm.arm.neon.vminu.v2i32(<2 x i32>, <2 x i32>) nounwind readnone

declare <2 x float> @llvm.arm.neon.vmins.v2f32(<2 x float>, <2 x float>) nounwind readnone

declare <16 x i8> @llvm.arm.neon.vmins.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
declare <8 x i16> @llvm.arm.neon.vmins.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
declare <4 x i32> @llvm.arm.neon.vmins.v4i32(<4 x i32>, <4 x i32>) nounwind readnone

declare <16 x i8> @llvm.arm.neon.vminu.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
declare <8 x i16> @llvm.arm.neon.vminu.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
declare <4 x i32> @llvm.arm.neon.vminu.v4i32(<4 x i32>, <4 x i32>) nounwind readnone

declare <4 x float> @llvm.arm.neon.vmins.v4f32(<4 x float>, <4 x float>) nounwind readnone

define <8 x i8> @vmaxs8(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vmaxs8:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vldr d16, [r1]
; CHECK-NEXT:    vldr d17, [r0]
; CHECK-NEXT:    vmax.s8 d16, d17, d16
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <8 x i8>, ptr %A
  %tmp2 = load <8 x i8>, ptr %B
  %tmp3 = call <8 x i8> @llvm.arm.neon.vmaxs.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
  ret <8 x i8> %tmp3
}

define <4 x i16> @vmaxs16(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vmaxs16:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vldr d16, [r1]
; CHECK-NEXT:    vldr d17, [r0]
; CHECK-NEXT:    vmax.s16 d16, d17, d16
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <4 x i16>, ptr %A
  %tmp2 = load <4 x i16>, ptr %B
  %tmp3 = call <4 x i16> @llvm.arm.neon.vmaxs.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
  ret <4 x i16> %tmp3
}

define <2 x i32> @vmaxs32(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vmaxs32:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vldr d16, [r1]
; CHECK-NEXT:    vldr d17, [r0]
; CHECK-NEXT:    vmax.s32 d16, d17, d16
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <2 x i32>, ptr %A
  %tmp2 = load <2 x i32>, ptr %B
  %tmp3 = call <2 x i32> @llvm.arm.neon.vmaxs.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
  ret <2 x i32> %tmp3
}

define <8 x i8> @vmaxu8(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vmaxu8:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vldr d16, [r1]
; CHECK-NEXT:    vldr d17, [r0]
; CHECK-NEXT:    vmax.u8 d16, d17, d16
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <8 x i8>, ptr %A
  %tmp2 = load <8 x i8>, ptr %B
  %tmp3 = call <8 x i8> @llvm.arm.neon.vmaxu.v8i8(<8 x i8> %tmp1, <8 x i8> %tmp2)
  ret <8 x i8> %tmp3
}

define <4 x i16> @vmaxu16(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vmaxu16:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vldr d16, [r1]
; CHECK-NEXT:    vldr d17, [r0]
; CHECK-NEXT:    vmax.u16 d16, d17, d16
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <4 x i16>, ptr %A
  %tmp2 = load <4 x i16>, ptr %B
  %tmp3 = call <4 x i16> @llvm.arm.neon.vmaxu.v4i16(<4 x i16> %tmp1, <4 x i16> %tmp2)
  ret <4 x i16> %tmp3
}

define <2 x i32> @vmaxu32(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vmaxu32:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vldr d16, [r1]
; CHECK-NEXT:    vldr d17, [r0]
; CHECK-NEXT:    vmax.u32 d16, d17, d16
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <2 x i32>, ptr %A
  %tmp2 = load <2 x i32>, ptr %B
  %tmp3 = call <2 x i32> @llvm.arm.neon.vmaxu.v2i32(<2 x i32> %tmp1, <2 x i32> %tmp2)
  ret <2 x i32> %tmp3
}

define <2 x float> @vmaxf32(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vmaxf32:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vldr d16, [r1]
; CHECK-NEXT:    vldr d17, [r0]
; CHECK-NEXT:    vmax.f32 d16, d17, d16
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <2 x float>, ptr %A
  %tmp2 = load <2 x float>, ptr %B
  %tmp3 = call <2 x float> @llvm.arm.neon.vmaxs.v2f32(<2 x float> %tmp1, <2 x float> %tmp2)
  ret <2 x float> %tmp3
}

define <16 x i8> @vmaxQs8(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vmaxQs8:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vld1.64 {d16, d17}, [r1]
; CHECK-NEXT:    vld1.64 {d18, d19}, [r0]
; CHECK-NEXT:    vmax.s8 q8, q9, q8
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    vmov r2, r3, d17
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <16 x i8>, ptr %A
  %tmp2 = load <16 x i8>, ptr %B
  %tmp3 = call <16 x i8> @llvm.arm.neon.vmaxs.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
  ret <16 x i8> %tmp3
}

define <8 x i16> @vmaxQs16(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vmaxQs16:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vld1.64 {d16, d17}, [r1]
; CHECK-NEXT:    vld1.64 {d18, d19}, [r0]
; CHECK-NEXT:    vmax.s16 q8, q9, q8
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    vmov r2, r3, d17
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <8 x i16>, ptr %A
  %tmp2 = load <8 x i16>, ptr %B
  %tmp3 = call <8 x i16> @llvm.arm.neon.vmaxs.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
  ret <8 x i16> %tmp3
}

define <4 x i32> @vmaxQs32(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vmaxQs32:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vld1.64 {d16, d17}, [r1]
; CHECK-NEXT:    vld1.64 {d18, d19}, [r0]
; CHECK-NEXT:    vmax.s32 q8, q9, q8
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    vmov r2, r3, d17
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <4 x i32>, ptr %A
  %tmp2 = load <4 x i32>, ptr %B
  %tmp3 = call <4 x i32> @llvm.arm.neon.vmaxs.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
  ret <4 x i32> %tmp3
}

define <16 x i8> @vmaxQu8(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vmaxQu8:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vld1.64 {d16, d17}, [r1]
; CHECK-NEXT:    vld1.64 {d18, d19}, [r0]
; CHECK-NEXT:    vmax.u8 q8, q9, q8
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    vmov r2, r3, d17
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <16 x i8>, ptr %A
  %tmp2 = load <16 x i8>, ptr %B
  %tmp3 = call <16 x i8> @llvm.arm.neon.vmaxu.v16i8(<16 x i8> %tmp1, <16 x i8> %tmp2)
  ret <16 x i8> %tmp3
}

define <8 x i16> @vmaxQu16(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vmaxQu16:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vld1.64 {d16, d17}, [r1]
; CHECK-NEXT:    vld1.64 {d18, d19}, [r0]
; CHECK-NEXT:    vmax.u16 q8, q9, q8
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    vmov r2, r3, d17
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <8 x i16>, ptr %A
  %tmp2 = load <8 x i16>, ptr %B
  %tmp3 = call <8 x i16> @llvm.arm.neon.vmaxu.v8i16(<8 x i16> %tmp1, <8 x i16> %tmp2)
  ret <8 x i16> %tmp3
}

define <4 x i32> @vmaxQu32(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vmaxQu32:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vld1.64 {d16, d17}, [r1]
; CHECK-NEXT:    vld1.64 {d18, d19}, [r0]
; CHECK-NEXT:    vmax.u32 q8, q9, q8
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    vmov r2, r3, d17
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <4 x i32>, ptr %A
  %tmp2 = load <4 x i32>, ptr %B
  %tmp3 = call <4 x i32> @llvm.arm.neon.vmaxu.v4i32(<4 x i32> %tmp1, <4 x i32> %tmp2)
  ret <4 x i32> %tmp3
}

define <4 x float> @vmaxQf32(ptr %A, ptr %B) nounwind {
; CHECK-LABEL: vmaxQf32:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vld1.64 {d16, d17}, [r1]
; CHECK-NEXT:    vld1.64 {d18, d19}, [r0]
; CHECK-NEXT:    vmax.f32 q8, q9, q8
; CHECK-NEXT:    vmov r0, r1, d16
; CHECK-NEXT:    vmov r2, r3, d17
; CHECK-NEXT:    mov pc, lr
  %tmp1 = load <4 x float>, ptr %A
  %tmp2 = load <4 x float>, ptr %B
  %tmp3 = call <4 x float> @llvm.arm.neon.vmaxs.v4f32(<4 x float> %tmp1, <4 x float> %tmp2)
  ret <4 x float> %tmp3
}

declare <8 x i8>  @llvm.arm.neon.vmaxs.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
declare <4 x i16> @llvm.arm.neon.vmaxs.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
declare <2 x i32> @llvm.arm.neon.vmaxs.v2i32(<2 x i32>, <2 x i32>) nounwind readnone

declare <8 x i8>  @llvm.arm.neon.vmaxu.v8i8(<8 x i8>, <8 x i8>) nounwind readnone
declare <4 x i16> @llvm.arm.neon.vmaxu.v4i16(<4 x i16>, <4 x i16>) nounwind readnone
declare <2 x i32> @llvm.arm.neon.vmaxu.v2i32(<2 x i32>, <2 x i32>) nounwind readnone

declare <2 x float> @llvm.arm.neon.vmaxs.v2f32(<2 x float>, <2 x float>) nounwind readnone

declare <16 x i8> @llvm.arm.neon.vmaxs.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
declare <8 x i16> @llvm.arm.neon.vmaxs.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
declare <4 x i32> @llvm.arm.neon.vmaxs.v4i32(<4 x i32>, <4 x i32>) nounwind readnone

declare <16 x i8> @llvm.arm.neon.vmaxu.v16i8(<16 x i8>, <16 x i8>) nounwind readnone
declare <8 x i16> @llvm.arm.neon.vmaxu.v8i16(<8 x i16>, <8 x i16>) nounwind readnone
declare <4 x i32> @llvm.arm.neon.vmaxu.v4i32(<4 x i32>, <4 x i32>) nounwind readnone

declare <4 x float> @llvm.arm.neon.vmaxs.v4f32(<4 x float>, <4 x float>) nounwind readnone

declare float @llvm.maxnum.f32(float %a, float %b)
declare float @llvm.minnum.f32(float %a, float %b)

define float @maxnum(float %a, float %b) {
; CHECK-LABEL: maxnum:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vmov s0, r1
; CHECK-NEXT:    vmov s2, r0
; CHECK-NEXT:    vmax.f32 d0, d1, d0
; CHECK-NEXT:    vmov r0, s0
; CHECK-NEXT:    mov pc, lr
  %r = call nnan float @llvm.maxnum.f32(float %a, float %b)
  ret float %r
}

define float @minnum(float %a, float %b) {
; CHECK-LABEL: minnum:
; CHECK:       @ %bb.0:
; CHECK-NEXT:    vmov s0, r1
; CHECK-NEXT:    vmov s2, r0
; CHECK-NEXT:    vmin.f32 d0, d1, d0
; CHECK-NEXT:    vmov r0, s0
; CHECK-NEXT:    mov pc, lr
  %r = call nnan float @llvm.minnum.f32(float %a, float %b)
  ret float %r
}