; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -aarch64-sve-vector-bits-min=256  < %s | FileCheck %s -check-prefixes=CHECK,VBITS_GE_256
; RUN: llc -aarch64-sve-vector-bits-min=512  < %s | FileCheck %s -check-prefixes=CHECK,VBITS_GE_512
; RUN: llc -aarch64-sve-vector-bits-min=2048 < %s | FileCheck %s -check-prefixes=CHECK,VBITS_GE_512

target triple = "aarch64-unknown-linux-gnu"

;
; SMAX
;

; Don't use SVE for 64-bit vectors.
define <8 x i8> @smax_v8i8(<8 x i8> %op1, <8 x i8> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: smax_v8i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smax v0.8b, v0.8b, v1.8b
; CHECK-NEXT:    ret
  %res = call <8 x i8> @llvm.smax.v8i8(<8 x i8> %op1, <8 x i8> %op2)
  ret <8 x i8> %res
}

; Don't use SVE for 128-bit vectors.
define <16 x i8> @smax_v16i8(<16 x i8> %op1, <16 x i8> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: smax_v16i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smax v0.16b, v0.16b, v1.16b
; CHECK-NEXT:    ret
  %res = call <16 x i8> @llvm.smax.v16i8(<16 x i8> %op1, <16 x i8> %op2)
  ret <16 x i8> %res
}

define void @smax_v32i8(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: smax_v32i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.b, vl32
; CHECK-NEXT:    ld1b { z0.b }, p0/z, [x0]
; CHECK-NEXT:    ld1b { z1.b }, p0/z, [x1]
; CHECK-NEXT:    smax z0.b, p0/m, z0.b, z1.b
; CHECK-NEXT:    st1b { z0.b }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <32 x i8>, ptr %a
  %op2 = load <32 x i8>, ptr %b
  %res = call <32 x i8> @llvm.smax.v32i8(<32 x i8> %op1, <32 x i8> %op2)
  store <32 x i8> %res, ptr %a
  ret void
}

define void @smax_v64i8(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: smax_v64i8:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.b, vl32
; VBITS_GE_256-NEXT:    mov w8, #32 // =0x20
; VBITS_GE_256-NEXT:    ld1b { z0.b }, p0/z, [x0, x8]
; VBITS_GE_256-NEXT:    ld1b { z1.b }, p0/z, [x1, x8]
; VBITS_GE_256-NEXT:    ld1b { z2.b }, p0/z, [x0]
; VBITS_GE_256-NEXT:    ld1b { z3.b }, p0/z, [x1]
; VBITS_GE_256-NEXT:    smax z0.b, p0/m, z0.b, z1.b
; VBITS_GE_256-NEXT:    movprfx z1, z2
; VBITS_GE_256-NEXT:    smax z1.b, p0/m, z1.b, z3.b
; VBITS_GE_256-NEXT:    st1b { z0.b }, p0, [x0, x8]
; VBITS_GE_256-NEXT:    st1b { z1.b }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: smax_v64i8:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.b, vl64
; VBITS_GE_512-NEXT:    ld1b { z0.b }, p0/z, [x0]
; VBITS_GE_512-NEXT:    ld1b { z1.b }, p0/z, [x1]
; VBITS_GE_512-NEXT:    smax z0.b, p0/m, z0.b, z1.b
; VBITS_GE_512-NEXT:    st1b { z0.b }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <64 x i8>, ptr %a
  %op2 = load <64 x i8>, ptr %b
  %res = call <64 x i8> @llvm.smax.v64i8(<64 x i8> %op1, <64 x i8> %op2)
  store <64 x i8> %res, ptr %a
  ret void
}

define void @smax_v128i8(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: smax_v128i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.b, vl128
; CHECK-NEXT:    ld1b { z0.b }, p0/z, [x0]
; CHECK-NEXT:    ld1b { z1.b }, p0/z, [x1]
; CHECK-NEXT:    smax z0.b, p0/m, z0.b, z1.b
; CHECK-NEXT:    st1b { z0.b }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <128 x i8>, ptr %a
  %op2 = load <128 x i8>, ptr %b
  %res = call <128 x i8> @llvm.smax.v128i8(<128 x i8> %op1, <128 x i8> %op2)
  store <128 x i8> %res, ptr %a
  ret void
}

define void @smax_v256i8(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: smax_v256i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.b, vl256
; CHECK-NEXT:    ld1b { z0.b }, p0/z, [x0]
; CHECK-NEXT:    ld1b { z1.b }, p0/z, [x1]
; CHECK-NEXT:    smax z0.b, p0/m, z0.b, z1.b
; CHECK-NEXT:    st1b { z0.b }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <256 x i8>, ptr %a
  %op2 = load <256 x i8>, ptr %b
  %res = call <256 x i8> @llvm.smax.v256i8(<256 x i8> %op1, <256 x i8> %op2)
  store <256 x i8> %res, ptr %a
  ret void
}

; Don't use SVE for 64-bit vectors.
define <4 x i16> @smax_v4i16(<4 x i16> %op1, <4 x i16> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: smax_v4i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smax v0.4h, v0.4h, v1.4h
; CHECK-NEXT:    ret
  %res = call <4 x i16> @llvm.smax.v4i16(<4 x i16> %op1, <4 x i16> %op2)
  ret <4 x i16> %res
}

; Don't use SVE for 128-bit vectors.
define <8 x i16> @smax_v8i16(<8 x i16> %op1, <8 x i16> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: smax_v8i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smax v0.8h, v0.8h, v1.8h
; CHECK-NEXT:    ret
  %res = call <8 x i16> @llvm.smax.v8i16(<8 x i16> %op1, <8 x i16> %op2)
  ret <8 x i16> %res
}

define void @smax_v16i16(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: smax_v16i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl16
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    ld1h { z1.h }, p0/z, [x1]
; CHECK-NEXT:    smax z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <16 x i16>, ptr %a
  %op2 = load <16 x i16>, ptr %b
  %res = call <16 x i16> @llvm.smax.v16i16(<16 x i16> %op1, <16 x i16> %op2)
  store <16 x i16> %res, ptr %a
  ret void
}

define void @smax_v32i16(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: smax_v32i16:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.h, vl16
; VBITS_GE_256-NEXT:    mov x8, #16 // =0x10
; VBITS_GE_256-NEXT:    ld1h { z0.h }, p0/z, [x0, x8, lsl #1]
; VBITS_GE_256-NEXT:    ld1h { z1.h }, p0/z, [x1, x8, lsl #1]
; VBITS_GE_256-NEXT:    ld1h { z2.h }, p0/z, [x0]
; VBITS_GE_256-NEXT:    ld1h { z3.h }, p0/z, [x1]
; VBITS_GE_256-NEXT:    smax z0.h, p0/m, z0.h, z1.h
; VBITS_GE_256-NEXT:    movprfx z1, z2
; VBITS_GE_256-NEXT:    smax z1.h, p0/m, z1.h, z3.h
; VBITS_GE_256-NEXT:    st1h { z0.h }, p0, [x0, x8, lsl #1]
; VBITS_GE_256-NEXT:    st1h { z1.h }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: smax_v32i16:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.h, vl32
; VBITS_GE_512-NEXT:    ld1h { z0.h }, p0/z, [x0]
; VBITS_GE_512-NEXT:    ld1h { z1.h }, p0/z, [x1]
; VBITS_GE_512-NEXT:    smax z0.h, p0/m, z0.h, z1.h
; VBITS_GE_512-NEXT:    st1h { z0.h }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <32 x i16>, ptr %a
  %op2 = load <32 x i16>, ptr %b
  %res = call <32 x i16> @llvm.smax.v32i16(<32 x i16> %op1, <32 x i16> %op2)
  store <32 x i16> %res, ptr %a
  ret void
}

define void @smax_v64i16(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: smax_v64i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl64
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    ld1h { z1.h }, p0/z, [x1]
; CHECK-NEXT:    smax z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <64 x i16>, ptr %a
  %op2 = load <64 x i16>, ptr %b
  %res = call <64 x i16> @llvm.smax.v64i16(<64 x i16> %op1, <64 x i16> %op2)
  store <64 x i16> %res, ptr %a
  ret void
}

define void @smax_v128i16(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: smax_v128i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl128
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    ld1h { z1.h }, p0/z, [x1]
; CHECK-NEXT:    smax z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <128 x i16>, ptr %a
  %op2 = load <128 x i16>, ptr %b
  %res = call <128 x i16> @llvm.smax.v128i16(<128 x i16> %op1, <128 x i16> %op2)
  store <128 x i16> %res, ptr %a
  ret void
}

; Don't use SVE for 64-bit vectors.
define <2 x i32> @smax_v2i32(<2 x i32> %op1, <2 x i32> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: smax_v2i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smax v0.2s, v0.2s, v1.2s
; CHECK-NEXT:    ret
  %res = call <2 x i32> @llvm.smax.v2i32(<2 x i32> %op1, <2 x i32> %op2)
  ret <2 x i32> %res
}

; Don't use SVE for 128-bit vectors.
define <4 x i32> @smax_v4i32(<4 x i32> %op1, <4 x i32> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: smax_v4i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smax v0.4s, v0.4s, v1.4s
; CHECK-NEXT:    ret
  %res = call <4 x i32> @llvm.smax.v4i32(<4 x i32> %op1, <4 x i32> %op2)
  ret <4 x i32> %res
}

define void @smax_v8i32(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: smax_v8i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl8
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    ld1w { z1.s }, p0/z, [x1]
; CHECK-NEXT:    smax z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <8 x i32>, ptr %a
  %op2 = load <8 x i32>, ptr %b
  %res = call <8 x i32> @llvm.smax.v8i32(<8 x i32> %op1, <8 x i32> %op2)
  store <8 x i32> %res, ptr %a
  ret void
}

define void @smax_v16i32(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: smax_v16i32:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.s, vl8
; VBITS_GE_256-NEXT:    mov x8, #8 // =0x8
; VBITS_GE_256-NEXT:    ld1w { z0.s }, p0/z, [x0, x8, lsl #2]
; VBITS_GE_256-NEXT:    ld1w { z1.s }, p0/z, [x1, x8, lsl #2]
; VBITS_GE_256-NEXT:    ld1w { z2.s }, p0/z, [x0]
; VBITS_GE_256-NEXT:    ld1w { z3.s }, p0/z, [x1]
; VBITS_GE_256-NEXT:    smax z0.s, p0/m, z0.s, z1.s
; VBITS_GE_256-NEXT:    movprfx z1, z2
; VBITS_GE_256-NEXT:    smax z1.s, p0/m, z1.s, z3.s
; VBITS_GE_256-NEXT:    st1w { z0.s }, p0, [x0, x8, lsl #2]
; VBITS_GE_256-NEXT:    st1w { z1.s }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: smax_v16i32:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.s, vl16
; VBITS_GE_512-NEXT:    ld1w { z0.s }, p0/z, [x0]
; VBITS_GE_512-NEXT:    ld1w { z1.s }, p0/z, [x1]
; VBITS_GE_512-NEXT:    smax z0.s, p0/m, z0.s, z1.s
; VBITS_GE_512-NEXT:    st1w { z0.s }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <16 x i32>, ptr %a
  %op2 = load <16 x i32>, ptr %b
  %res = call <16 x i32> @llvm.smax.v16i32(<16 x i32> %op1, <16 x i32> %op2)
  store <16 x i32> %res, ptr %a
  ret void
}

define void @smax_v32i32(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: smax_v32i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl32
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    ld1w { z1.s }, p0/z, [x1]
; CHECK-NEXT:    smax z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <32 x i32>, ptr %a
  %op2 = load <32 x i32>, ptr %b
  %res = call <32 x i32> @llvm.smax.v32i32(<32 x i32> %op1, <32 x i32> %op2)
  store <32 x i32> %res, ptr %a
  ret void
}

define void @smax_v64i32(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: smax_v64i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl64
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    ld1w { z1.s }, p0/z, [x1]
; CHECK-NEXT:    smax z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <64 x i32>, ptr %a
  %op2 = load <64 x i32>, ptr %b
  %res = call <64 x i32> @llvm.smax.v64i32(<64 x i32> %op1, <64 x i32> %op2)
  store <64 x i32> %res, ptr %a
  ret void
}

; Vector i64 max are not legal for NEON so use SVE when available.
define <1 x i64> @smax_v1i64(<1 x i64> %op1, <1 x i64> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: smax_v1i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl1
; CHECK-NEXT:    // kill: def $d0 killed $d0 def $z0
; CHECK-NEXT:    // kill: def $d1 killed $d1 def $z1
; CHECK-NEXT:    smax z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $z0
; CHECK-NEXT:    ret
  %res = call <1 x i64> @llvm.smax.v1i64(<1 x i64> %op1, <1 x i64> %op2)
  ret <1 x i64> %res
}

; Vector i64 max are not legal for NEON so use SVE when available.
define <2 x i64> @smax_v2i64(<2 x i64> %op1, <2 x i64> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: smax_v2i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl2
; CHECK-NEXT:    // kill: def $q0 killed $q0 def $z0
; CHECK-NEXT:    // kill: def $q1 killed $q1 def $z1
; CHECK-NEXT:    smax z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    // kill: def $q0 killed $q0 killed $z0
; CHECK-NEXT:    ret
  %res = call <2 x i64> @llvm.smax.v2i64(<2 x i64> %op1, <2 x i64> %op2)
  ret <2 x i64> %res
}

define void @smax_v4i64(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: smax_v4i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    ld1d { z1.d }, p0/z, [x1]
; CHECK-NEXT:    smax z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <4 x i64>, ptr %a
  %op2 = load <4 x i64>, ptr %b
  %res = call <4 x i64> @llvm.smax.v4i64(<4 x i64> %op1, <4 x i64> %op2)
  store <4 x i64> %res, ptr %a
  ret void
}

define void @smax_v8i64(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: smax_v8i64:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.d, vl4
; VBITS_GE_256-NEXT:    mov x8, #4 // =0x4
; VBITS_GE_256-NEXT:    ld1d { z0.d }, p0/z, [x0, x8, lsl #3]
; VBITS_GE_256-NEXT:    ld1d { z1.d }, p0/z, [x1, x8, lsl #3]
; VBITS_GE_256-NEXT:    ld1d { z2.d }, p0/z, [x0]
; VBITS_GE_256-NEXT:    ld1d { z3.d }, p0/z, [x1]
; VBITS_GE_256-NEXT:    smax z0.d, p0/m, z0.d, z1.d
; VBITS_GE_256-NEXT:    movprfx z1, z2
; VBITS_GE_256-NEXT:    smax z1.d, p0/m, z1.d, z3.d
; VBITS_GE_256-NEXT:    st1d { z0.d }, p0, [x0, x8, lsl #3]
; VBITS_GE_256-NEXT:    st1d { z1.d }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: smax_v8i64:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.d, vl8
; VBITS_GE_512-NEXT:    ld1d { z0.d }, p0/z, [x0]
; VBITS_GE_512-NEXT:    ld1d { z1.d }, p0/z, [x1]
; VBITS_GE_512-NEXT:    smax z0.d, p0/m, z0.d, z1.d
; VBITS_GE_512-NEXT:    st1d { z0.d }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <8 x i64>, ptr %a
  %op2 = load <8 x i64>, ptr %b
  %res = call <8 x i64> @llvm.smax.v8i64(<8 x i64> %op1, <8 x i64> %op2)
  store <8 x i64> %res, ptr %a
  ret void
}

define void @smax_v16i64(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: smax_v16i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl16
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    ld1d { z1.d }, p0/z, [x1]
; CHECK-NEXT:    smax z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <16 x i64>, ptr %a
  %op2 = load <16 x i64>, ptr %b
  %res = call <16 x i64> @llvm.smax.v16i64(<16 x i64> %op1, <16 x i64> %op2)
  store <16 x i64> %res, ptr %a
  ret void
}

define void @smax_v32i64(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: smax_v32i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl32
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    ld1d { z1.d }, p0/z, [x1]
; CHECK-NEXT:    smax z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <32 x i64>, ptr %a
  %op2 = load <32 x i64>, ptr %b
  %res = call <32 x i64> @llvm.smax.v32i64(<32 x i64> %op1, <32 x i64> %op2)
  store <32 x i64> %res, ptr %a
  ret void
}

;
; SMIN
;

; Don't use SVE for 64-bit vectors.
define <8 x i8> @smin_v8i8(<8 x i8> %op1, <8 x i8> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: smin_v8i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smin v0.8b, v0.8b, v1.8b
; CHECK-NEXT:    ret
  %res = call <8 x i8> @llvm.smin.v8i8(<8 x i8> %op1, <8 x i8> %op2)
  ret <8 x i8> %res
}

; Don't use SVE for 128-bit vectors.
define <16 x i8> @smin_v16i8(<16 x i8> %op1, <16 x i8> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: smin_v16i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smin v0.16b, v0.16b, v1.16b
; CHECK-NEXT:    ret
  %res = call <16 x i8> @llvm.smin.v16i8(<16 x i8> %op1, <16 x i8> %op2)
  ret <16 x i8> %res
}

define void @smin_v32i8(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: smin_v32i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.b, vl32
; CHECK-NEXT:    ld1b { z0.b }, p0/z, [x0]
; CHECK-NEXT:    ld1b { z1.b }, p0/z, [x1]
; CHECK-NEXT:    smin z0.b, p0/m, z0.b, z1.b
; CHECK-NEXT:    st1b { z0.b }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <32 x i8>, ptr %a
  %op2 = load <32 x i8>, ptr %b
  %res = call <32 x i8> @llvm.smin.v32i8(<32 x i8> %op1, <32 x i8> %op2)
  store <32 x i8> %res, ptr %a
  ret void
}

define void @smin_v64i8(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: smin_v64i8:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.b, vl32
; VBITS_GE_256-NEXT:    mov w8, #32 // =0x20
; VBITS_GE_256-NEXT:    ld1b { z0.b }, p0/z, [x0, x8]
; VBITS_GE_256-NEXT:    ld1b { z1.b }, p0/z, [x1, x8]
; VBITS_GE_256-NEXT:    ld1b { z2.b }, p0/z, [x0]
; VBITS_GE_256-NEXT:    ld1b { z3.b }, p0/z, [x1]
; VBITS_GE_256-NEXT:    smin z0.b, p0/m, z0.b, z1.b
; VBITS_GE_256-NEXT:    movprfx z1, z2
; VBITS_GE_256-NEXT:    smin z1.b, p0/m, z1.b, z3.b
; VBITS_GE_256-NEXT:    st1b { z0.b }, p0, [x0, x8]
; VBITS_GE_256-NEXT:    st1b { z1.b }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: smin_v64i8:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.b, vl64
; VBITS_GE_512-NEXT:    ld1b { z0.b }, p0/z, [x0]
; VBITS_GE_512-NEXT:    ld1b { z1.b }, p0/z, [x1]
; VBITS_GE_512-NEXT:    smin z0.b, p0/m, z0.b, z1.b
; VBITS_GE_512-NEXT:    st1b { z0.b }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <64 x i8>, ptr %a
  %op2 = load <64 x i8>, ptr %b
  %res = call <64 x i8> @llvm.smin.v64i8(<64 x i8> %op1, <64 x i8> %op2)
  store <64 x i8> %res, ptr %a
  ret void
}

define void @smin_v128i8(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: smin_v128i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.b, vl128
; CHECK-NEXT:    ld1b { z0.b }, p0/z, [x0]
; CHECK-NEXT:    ld1b { z1.b }, p0/z, [x1]
; CHECK-NEXT:    smin z0.b, p0/m, z0.b, z1.b
; CHECK-NEXT:    st1b { z0.b }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <128 x i8>, ptr %a
  %op2 = load <128 x i8>, ptr %b
  %res = call <128 x i8> @llvm.smin.v128i8(<128 x i8> %op1, <128 x i8> %op2)
  store <128 x i8> %res, ptr %a
  ret void
}

define void @smin_v256i8(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: smin_v256i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.b, vl256
; CHECK-NEXT:    ld1b { z0.b }, p0/z, [x0]
; CHECK-NEXT:    ld1b { z1.b }, p0/z, [x1]
; CHECK-NEXT:    smin z0.b, p0/m, z0.b, z1.b
; CHECK-NEXT:    st1b { z0.b }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <256 x i8>, ptr %a
  %op2 = load <256 x i8>, ptr %b
  %res = call <256 x i8> @llvm.smin.v256i8(<256 x i8> %op1, <256 x i8> %op2)
  store <256 x i8> %res, ptr %a
  ret void
}

; Don't use SVE for 64-bit vectors.
define <4 x i16> @smin_v4i16(<4 x i16> %op1, <4 x i16> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: smin_v4i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smin v0.4h, v0.4h, v1.4h
; CHECK-NEXT:    ret
  %res = call <4 x i16> @llvm.smin.v4i16(<4 x i16> %op1, <4 x i16> %op2)
  ret <4 x i16> %res
}

; Don't use SVE for 128-bit vectors.
define <8 x i16> @smin_v8i16(<8 x i16> %op1, <8 x i16> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: smin_v8i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smin v0.8h, v0.8h, v1.8h
; CHECK-NEXT:    ret
  %res = call <8 x i16> @llvm.smin.v8i16(<8 x i16> %op1, <8 x i16> %op2)
  ret <8 x i16> %res
}

define void @smin_v16i16(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: smin_v16i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl16
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    ld1h { z1.h }, p0/z, [x1]
; CHECK-NEXT:    smin z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <16 x i16>, ptr %a
  %op2 = load <16 x i16>, ptr %b
  %res = call <16 x i16> @llvm.smin.v16i16(<16 x i16> %op1, <16 x i16> %op2)
  store <16 x i16> %res, ptr %a
  ret void
}

define void @smin_v32i16(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: smin_v32i16:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.h, vl16
; VBITS_GE_256-NEXT:    mov x8, #16 // =0x10
; VBITS_GE_256-NEXT:    ld1h { z0.h }, p0/z, [x0, x8, lsl #1]
; VBITS_GE_256-NEXT:    ld1h { z1.h }, p0/z, [x1, x8, lsl #1]
; VBITS_GE_256-NEXT:    ld1h { z2.h }, p0/z, [x0]
; VBITS_GE_256-NEXT:    ld1h { z3.h }, p0/z, [x1]
; VBITS_GE_256-NEXT:    smin z0.h, p0/m, z0.h, z1.h
; VBITS_GE_256-NEXT:    movprfx z1, z2
; VBITS_GE_256-NEXT:    smin z1.h, p0/m, z1.h, z3.h
; VBITS_GE_256-NEXT:    st1h { z0.h }, p0, [x0, x8, lsl #1]
; VBITS_GE_256-NEXT:    st1h { z1.h }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: smin_v32i16:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.h, vl32
; VBITS_GE_512-NEXT:    ld1h { z0.h }, p0/z, [x0]
; VBITS_GE_512-NEXT:    ld1h { z1.h }, p0/z, [x1]
; VBITS_GE_512-NEXT:    smin z0.h, p0/m, z0.h, z1.h
; VBITS_GE_512-NEXT:    st1h { z0.h }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <32 x i16>, ptr %a
  %op2 = load <32 x i16>, ptr %b
  %res = call <32 x i16> @llvm.smin.v32i16(<32 x i16> %op1, <32 x i16> %op2)
  store <32 x i16> %res, ptr %a
  ret void
}

define void @smin_v64i16(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: smin_v64i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl64
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    ld1h { z1.h }, p0/z, [x1]
; CHECK-NEXT:    smin z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <64 x i16>, ptr %a
  %op2 = load <64 x i16>, ptr %b
  %res = call <64 x i16> @llvm.smin.v64i16(<64 x i16> %op1, <64 x i16> %op2)
  store <64 x i16> %res, ptr %a
  ret void
}

define void @smin_v128i16(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: smin_v128i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl128
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    ld1h { z1.h }, p0/z, [x1]
; CHECK-NEXT:    smin z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <128 x i16>, ptr %a
  %op2 = load <128 x i16>, ptr %b
  %res = call <128 x i16> @llvm.smin.v128i16(<128 x i16> %op1, <128 x i16> %op2)
  store <128 x i16> %res, ptr %a
  ret void
}

; Don't use SVE for 64-bit vectors.
define <2 x i32> @smin_v2i32(<2 x i32> %op1, <2 x i32> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: smin_v2i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smin v0.2s, v0.2s, v1.2s
; CHECK-NEXT:    ret
  %res = call <2 x i32> @llvm.smin.v2i32(<2 x i32> %op1, <2 x i32> %op2)
  ret <2 x i32> %res
}

; Don't use SVE for 128-bit vectors.
define <4 x i32> @smin_v4i32(<4 x i32> %op1, <4 x i32> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: smin_v4i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smin v0.4s, v0.4s, v1.4s
; CHECK-NEXT:    ret
  %res = call <4 x i32> @llvm.smin.v4i32(<4 x i32> %op1, <4 x i32> %op2)
  ret <4 x i32> %res
}

define void @smin_v8i32(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: smin_v8i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl8
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    ld1w { z1.s }, p0/z, [x1]
; CHECK-NEXT:    smin z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <8 x i32>, ptr %a
  %op2 = load <8 x i32>, ptr %b
  %res = call <8 x i32> @llvm.smin.v8i32(<8 x i32> %op1, <8 x i32> %op2)
  store <8 x i32> %res, ptr %a
  ret void
}

define void @smin_v16i32(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: smin_v16i32:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.s, vl8
; VBITS_GE_256-NEXT:    mov x8, #8 // =0x8
; VBITS_GE_256-NEXT:    ld1w { z0.s }, p0/z, [x0, x8, lsl #2]
; VBITS_GE_256-NEXT:    ld1w { z1.s }, p0/z, [x1, x8, lsl #2]
; VBITS_GE_256-NEXT:    ld1w { z2.s }, p0/z, [x0]
; VBITS_GE_256-NEXT:    ld1w { z3.s }, p0/z, [x1]
; VBITS_GE_256-NEXT:    smin z0.s, p0/m, z0.s, z1.s
; VBITS_GE_256-NEXT:    movprfx z1, z2
; VBITS_GE_256-NEXT:    smin z1.s, p0/m, z1.s, z3.s
; VBITS_GE_256-NEXT:    st1w { z0.s }, p0, [x0, x8, lsl #2]
; VBITS_GE_256-NEXT:    st1w { z1.s }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: smin_v16i32:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.s, vl16
; VBITS_GE_512-NEXT:    ld1w { z0.s }, p0/z, [x0]
; VBITS_GE_512-NEXT:    ld1w { z1.s }, p0/z, [x1]
; VBITS_GE_512-NEXT:    smin z0.s, p0/m, z0.s, z1.s
; VBITS_GE_512-NEXT:    st1w { z0.s }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <16 x i32>, ptr %a
  %op2 = load <16 x i32>, ptr %b
  %res = call <16 x i32> @llvm.smin.v16i32(<16 x i32> %op1, <16 x i32> %op2)
  store <16 x i32> %res, ptr %a
  ret void
}

define void @smin_v32i32(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: smin_v32i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl32
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    ld1w { z1.s }, p0/z, [x1]
; CHECK-NEXT:    smin z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <32 x i32>, ptr %a
  %op2 = load <32 x i32>, ptr %b
  %res = call <32 x i32> @llvm.smin.v32i32(<32 x i32> %op1, <32 x i32> %op2)
  store <32 x i32> %res, ptr %a
  ret void
}

define void @smin_v64i32(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: smin_v64i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl64
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    ld1w { z1.s }, p0/z, [x1]
; CHECK-NEXT:    smin z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <64 x i32>, ptr %a
  %op2 = load <64 x i32>, ptr %b
  %res = call <64 x i32> @llvm.smin.v64i32(<64 x i32> %op1, <64 x i32> %op2)
  store <64 x i32> %res, ptr %a
  ret void
}

; Vector i64 min are not legal for NEON so use SVE when available.
define <1 x i64> @smin_v1i64(<1 x i64> %op1, <1 x i64> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: smin_v1i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl1
; CHECK-NEXT:    // kill: def $d0 killed $d0 def $z0
; CHECK-NEXT:    // kill: def $d1 killed $d1 def $z1
; CHECK-NEXT:    smin z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $z0
; CHECK-NEXT:    ret
  %res = call <1 x i64> @llvm.smin.v1i64(<1 x i64> %op1, <1 x i64> %op2)
  ret <1 x i64> %res
}

; Vector i64 min are not legal for NEON so use SVE when available.
define <2 x i64> @smin_v2i64(<2 x i64> %op1, <2 x i64> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: smin_v2i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl2
; CHECK-NEXT:    // kill: def $q0 killed $q0 def $z0
; CHECK-NEXT:    // kill: def $q1 killed $q1 def $z1
; CHECK-NEXT:    smin z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    // kill: def $q0 killed $q0 killed $z0
; CHECK-NEXT:    ret
  %res = call <2 x i64> @llvm.smin.v2i64(<2 x i64> %op1, <2 x i64> %op2)
  ret <2 x i64> %res
}

define void @smin_v4i64(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: smin_v4i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    ld1d { z1.d }, p0/z, [x1]
; CHECK-NEXT:    smin z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <4 x i64>, ptr %a
  %op2 = load <4 x i64>, ptr %b
  %res = call <4 x i64> @llvm.smin.v4i64(<4 x i64> %op1, <4 x i64> %op2)
  store <4 x i64> %res, ptr %a
  ret void
}

define void @smin_v8i64(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: smin_v8i64:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.d, vl4
; VBITS_GE_256-NEXT:    mov x8, #4 // =0x4
; VBITS_GE_256-NEXT:    ld1d { z0.d }, p0/z, [x0, x8, lsl #3]
; VBITS_GE_256-NEXT:    ld1d { z1.d }, p0/z, [x1, x8, lsl #3]
; VBITS_GE_256-NEXT:    ld1d { z2.d }, p0/z, [x0]
; VBITS_GE_256-NEXT:    ld1d { z3.d }, p0/z, [x1]
; VBITS_GE_256-NEXT:    smin z0.d, p0/m, z0.d, z1.d
; VBITS_GE_256-NEXT:    movprfx z1, z2
; VBITS_GE_256-NEXT:    smin z1.d, p0/m, z1.d, z3.d
; VBITS_GE_256-NEXT:    st1d { z0.d }, p0, [x0, x8, lsl #3]
; VBITS_GE_256-NEXT:    st1d { z1.d }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: smin_v8i64:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.d, vl8
; VBITS_GE_512-NEXT:    ld1d { z0.d }, p0/z, [x0]
; VBITS_GE_512-NEXT:    ld1d { z1.d }, p0/z, [x1]
; VBITS_GE_512-NEXT:    smin z0.d, p0/m, z0.d, z1.d
; VBITS_GE_512-NEXT:    st1d { z0.d }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <8 x i64>, ptr %a
  %op2 = load <8 x i64>, ptr %b
  %res = call <8 x i64> @llvm.smin.v8i64(<8 x i64> %op1, <8 x i64> %op2)
  store <8 x i64> %res, ptr %a
  ret void
}

define void @smin_v16i64(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: smin_v16i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl16
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    ld1d { z1.d }, p0/z, [x1]
; CHECK-NEXT:    smin z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <16 x i64>, ptr %a
  %op2 = load <16 x i64>, ptr %b
  %res = call <16 x i64> @llvm.smin.v16i64(<16 x i64> %op1, <16 x i64> %op2)
  store <16 x i64> %res, ptr %a
  ret void
}

define void @smin_v32i64(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: smin_v32i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl32
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    ld1d { z1.d }, p0/z, [x1]
; CHECK-NEXT:    smin z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <32 x i64>, ptr %a
  %op2 = load <32 x i64>, ptr %b
  %res = call <32 x i64> @llvm.smin.v32i64(<32 x i64> %op1, <32 x i64> %op2)
  store <32 x i64> %res, ptr %a
  ret void
}

;
; UMAX
;

; Don't use SVE for 64-bit vectors.
define <8 x i8> @umax_v8i8(<8 x i8> %op1, <8 x i8> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: umax_v8i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umax v0.8b, v0.8b, v1.8b
; CHECK-NEXT:    ret
  %res = call <8 x i8> @llvm.umax.v8i8(<8 x i8> %op1, <8 x i8> %op2)
  ret <8 x i8> %res
}

; Don't use SVE for 128-bit vectors.
define <16 x i8> @umax_v16i8(<16 x i8> %op1, <16 x i8> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: umax_v16i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umax v0.16b, v0.16b, v1.16b
; CHECK-NEXT:    ret
  %res = call <16 x i8> @llvm.umax.v16i8(<16 x i8> %op1, <16 x i8> %op2)
  ret <16 x i8> %res
}

define void @umax_v32i8(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: umax_v32i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.b, vl32
; CHECK-NEXT:    ld1b { z0.b }, p0/z, [x0]
; CHECK-NEXT:    ld1b { z1.b }, p0/z, [x1]
; CHECK-NEXT:    umax z0.b, p0/m, z0.b, z1.b
; CHECK-NEXT:    st1b { z0.b }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <32 x i8>, ptr %a
  %op2 = load <32 x i8>, ptr %b
  %res = call <32 x i8> @llvm.umax.v32i8(<32 x i8> %op1, <32 x i8> %op2)
  store <32 x i8> %res, ptr %a
  ret void
}

define void @umax_v64i8(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: umax_v64i8:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.b, vl32
; VBITS_GE_256-NEXT:    mov w8, #32 // =0x20
; VBITS_GE_256-NEXT:    ld1b { z0.b }, p0/z, [x0, x8]
; VBITS_GE_256-NEXT:    ld1b { z1.b }, p0/z, [x1, x8]
; VBITS_GE_256-NEXT:    ld1b { z2.b }, p0/z, [x0]
; VBITS_GE_256-NEXT:    ld1b { z3.b }, p0/z, [x1]
; VBITS_GE_256-NEXT:    umax z0.b, p0/m, z0.b, z1.b
; VBITS_GE_256-NEXT:    movprfx z1, z2
; VBITS_GE_256-NEXT:    umax z1.b, p0/m, z1.b, z3.b
; VBITS_GE_256-NEXT:    st1b { z0.b }, p0, [x0, x8]
; VBITS_GE_256-NEXT:    st1b { z1.b }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: umax_v64i8:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.b, vl64
; VBITS_GE_512-NEXT:    ld1b { z0.b }, p0/z, [x0]
; VBITS_GE_512-NEXT:    ld1b { z1.b }, p0/z, [x1]
; VBITS_GE_512-NEXT:    umax z0.b, p0/m, z0.b, z1.b
; VBITS_GE_512-NEXT:    st1b { z0.b }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <64 x i8>, ptr %a
  %op2 = load <64 x i8>, ptr %b
  %res = call <64 x i8> @llvm.umax.v64i8(<64 x i8> %op1, <64 x i8> %op2)
  store <64 x i8> %res, ptr %a
  ret void
}

define void @umax_v128i8(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: umax_v128i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.b, vl128
; CHECK-NEXT:    ld1b { z0.b }, p0/z, [x0]
; CHECK-NEXT:    ld1b { z1.b }, p0/z, [x1]
; CHECK-NEXT:    umax z0.b, p0/m, z0.b, z1.b
; CHECK-NEXT:    st1b { z0.b }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <128 x i8>, ptr %a
  %op2 = load <128 x i8>, ptr %b
  %res = call <128 x i8> @llvm.umax.v128i8(<128 x i8> %op1, <128 x i8> %op2)
  store <128 x i8> %res, ptr %a
  ret void
}

define void @umax_v256i8(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: umax_v256i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.b, vl256
; CHECK-NEXT:    ld1b { z0.b }, p0/z, [x0]
; CHECK-NEXT:    ld1b { z1.b }, p0/z, [x1]
; CHECK-NEXT:    umax z0.b, p0/m, z0.b, z1.b
; CHECK-NEXT:    st1b { z0.b }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <256 x i8>, ptr %a
  %op2 = load <256 x i8>, ptr %b
  %res = call <256 x i8> @llvm.umax.v256i8(<256 x i8> %op1, <256 x i8> %op2)
  store <256 x i8> %res, ptr %a
  ret void
}

; Don't use SVE for 64-bit vectors.
define <4 x i16> @umax_v4i16(<4 x i16> %op1, <4 x i16> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: umax_v4i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umax v0.4h, v0.4h, v1.4h
; CHECK-NEXT:    ret
  %res = call <4 x i16> @llvm.umax.v4i16(<4 x i16> %op1, <4 x i16> %op2)
  ret <4 x i16> %res
}

; Don't use SVE for 128-bit vectors.
define <8 x i16> @umax_v8i16(<8 x i16> %op1, <8 x i16> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: umax_v8i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umax v0.8h, v0.8h, v1.8h
; CHECK-NEXT:    ret
  %res = call <8 x i16> @llvm.umax.v8i16(<8 x i16> %op1, <8 x i16> %op2)
  ret <8 x i16> %res
}

define void @umax_v16i16(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: umax_v16i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl16
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    ld1h { z1.h }, p0/z, [x1]
; CHECK-NEXT:    umax z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <16 x i16>, ptr %a
  %op2 = load <16 x i16>, ptr %b
  %res = call <16 x i16> @llvm.umax.v16i16(<16 x i16> %op1, <16 x i16> %op2)
  store <16 x i16> %res, ptr %a
  ret void
}

define void @umax_v32i16(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: umax_v32i16:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.h, vl16
; VBITS_GE_256-NEXT:    mov x8, #16 // =0x10
; VBITS_GE_256-NEXT:    ld1h { z0.h }, p0/z, [x0, x8, lsl #1]
; VBITS_GE_256-NEXT:    ld1h { z1.h }, p0/z, [x1, x8, lsl #1]
; VBITS_GE_256-NEXT:    ld1h { z2.h }, p0/z, [x0]
; VBITS_GE_256-NEXT:    ld1h { z3.h }, p0/z, [x1]
; VBITS_GE_256-NEXT:    umax z0.h, p0/m, z0.h, z1.h
; VBITS_GE_256-NEXT:    movprfx z1, z2
; VBITS_GE_256-NEXT:    umax z1.h, p0/m, z1.h, z3.h
; VBITS_GE_256-NEXT:    st1h { z0.h }, p0, [x0, x8, lsl #1]
; VBITS_GE_256-NEXT:    st1h { z1.h }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: umax_v32i16:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.h, vl32
; VBITS_GE_512-NEXT:    ld1h { z0.h }, p0/z, [x0]
; VBITS_GE_512-NEXT:    ld1h { z1.h }, p0/z, [x1]
; VBITS_GE_512-NEXT:    umax z0.h, p0/m, z0.h, z1.h
; VBITS_GE_512-NEXT:    st1h { z0.h }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <32 x i16>, ptr %a
  %op2 = load <32 x i16>, ptr %b
  %res = call <32 x i16> @llvm.umax.v32i16(<32 x i16> %op1, <32 x i16> %op2)
  store <32 x i16> %res, ptr %a
  ret void
}

define void @umax_v64i16(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: umax_v64i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl64
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    ld1h { z1.h }, p0/z, [x1]
; CHECK-NEXT:    umax z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <64 x i16>, ptr %a
  %op2 = load <64 x i16>, ptr %b
  %res = call <64 x i16> @llvm.umax.v64i16(<64 x i16> %op1, <64 x i16> %op2)
  store <64 x i16> %res, ptr %a
  ret void
}

define void @umax_v128i16(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: umax_v128i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl128
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    ld1h { z1.h }, p0/z, [x1]
; CHECK-NEXT:    umax z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <128 x i16>, ptr %a
  %op2 = load <128 x i16>, ptr %b
  %res = call <128 x i16> @llvm.umax.v128i16(<128 x i16> %op1, <128 x i16> %op2)
  store <128 x i16> %res, ptr %a
  ret void
}

; Don't use SVE for 64-bit vectors.
define <2 x i32> @umax_v2i32(<2 x i32> %op1, <2 x i32> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: umax_v2i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umax v0.2s, v0.2s, v1.2s
; CHECK-NEXT:    ret
  %res = call <2 x i32> @llvm.umax.v2i32(<2 x i32> %op1, <2 x i32> %op2)
  ret <2 x i32> %res
}

; Don't use SVE for 128-bit vectors.
define <4 x i32> @umax_v4i32(<4 x i32> %op1, <4 x i32> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: umax_v4i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umax v0.4s, v0.4s, v1.4s
; CHECK-NEXT:    ret
  %res = call <4 x i32> @llvm.umax.v4i32(<4 x i32> %op1, <4 x i32> %op2)
  ret <4 x i32> %res
}

define void @umax_v8i32(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: umax_v8i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl8
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    ld1w { z1.s }, p0/z, [x1]
; CHECK-NEXT:    umax z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <8 x i32>, ptr %a
  %op2 = load <8 x i32>, ptr %b
  %res = call <8 x i32> @llvm.umax.v8i32(<8 x i32> %op1, <8 x i32> %op2)
  store <8 x i32> %res, ptr %a
  ret void
}

define void @umax_v16i32(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: umax_v16i32:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.s, vl8
; VBITS_GE_256-NEXT:    mov x8, #8 // =0x8
; VBITS_GE_256-NEXT:    ld1w { z0.s }, p0/z, [x0, x8, lsl #2]
; VBITS_GE_256-NEXT:    ld1w { z1.s }, p0/z, [x1, x8, lsl #2]
; VBITS_GE_256-NEXT:    ld1w { z2.s }, p0/z, [x0]
; VBITS_GE_256-NEXT:    ld1w { z3.s }, p0/z, [x1]
; VBITS_GE_256-NEXT:    umax z0.s, p0/m, z0.s, z1.s
; VBITS_GE_256-NEXT:    movprfx z1, z2
; VBITS_GE_256-NEXT:    umax z1.s, p0/m, z1.s, z3.s
; VBITS_GE_256-NEXT:    st1w { z0.s }, p0, [x0, x8, lsl #2]
; VBITS_GE_256-NEXT:    st1w { z1.s }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: umax_v16i32:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.s, vl16
; VBITS_GE_512-NEXT:    ld1w { z0.s }, p0/z, [x0]
; VBITS_GE_512-NEXT:    ld1w { z1.s }, p0/z, [x1]
; VBITS_GE_512-NEXT:    umax z0.s, p0/m, z0.s, z1.s
; VBITS_GE_512-NEXT:    st1w { z0.s }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <16 x i32>, ptr %a
  %op2 = load <16 x i32>, ptr %b
  %res = call <16 x i32> @llvm.umax.v16i32(<16 x i32> %op1, <16 x i32> %op2)
  store <16 x i32> %res, ptr %a
  ret void
}

define void @umax_v32i32(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: umax_v32i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl32
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    ld1w { z1.s }, p0/z, [x1]
; CHECK-NEXT:    umax z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <32 x i32>, ptr %a
  %op2 = load <32 x i32>, ptr %b
  %res = call <32 x i32> @llvm.umax.v32i32(<32 x i32> %op1, <32 x i32> %op2)
  store <32 x i32> %res, ptr %a
  ret void
}

define void @umax_v64i32(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: umax_v64i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl64
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    ld1w { z1.s }, p0/z, [x1]
; CHECK-NEXT:    umax z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <64 x i32>, ptr %a
  %op2 = load <64 x i32>, ptr %b
  %res = call <64 x i32> @llvm.umax.v64i32(<64 x i32> %op1, <64 x i32> %op2)
  store <64 x i32> %res, ptr %a
  ret void
}

; Vector i64 max are not legal for NEON so use SVE when available.
define <1 x i64> @umax_v1i64(<1 x i64> %op1, <1 x i64> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: umax_v1i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl1
; CHECK-NEXT:    // kill: def $d0 killed $d0 def $z0
; CHECK-NEXT:    // kill: def $d1 killed $d1 def $z1
; CHECK-NEXT:    umax z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $z0
; CHECK-NEXT:    ret
  %res = call <1 x i64> @llvm.umax.v1i64(<1 x i64> %op1, <1 x i64> %op2)
  ret <1 x i64> %res
}

; Vector i64 max are not legal for NEON so use SVE when available.
define <2 x i64> @umax_v2i64(<2 x i64> %op1, <2 x i64> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: umax_v2i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl2
; CHECK-NEXT:    // kill: def $q0 killed $q0 def $z0
; CHECK-NEXT:    // kill: def $q1 killed $q1 def $z1
; CHECK-NEXT:    umax z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    // kill: def $q0 killed $q0 killed $z0
; CHECK-NEXT:    ret
  %res = call <2 x i64> @llvm.umax.v2i64(<2 x i64> %op1, <2 x i64> %op2)
  ret <2 x i64> %res
}

define void @umax_v4i64(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: umax_v4i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    ld1d { z1.d }, p0/z, [x1]
; CHECK-NEXT:    umax z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <4 x i64>, ptr %a
  %op2 = load <4 x i64>, ptr %b
  %res = call <4 x i64> @llvm.umax.v4i64(<4 x i64> %op1, <4 x i64> %op2)
  store <4 x i64> %res, ptr %a
  ret void
}

define void @umax_v8i64(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: umax_v8i64:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.d, vl4
; VBITS_GE_256-NEXT:    mov x8, #4 // =0x4
; VBITS_GE_256-NEXT:    ld1d { z0.d }, p0/z, [x0, x8, lsl #3]
; VBITS_GE_256-NEXT:    ld1d { z1.d }, p0/z, [x1, x8, lsl #3]
; VBITS_GE_256-NEXT:    ld1d { z2.d }, p0/z, [x0]
; VBITS_GE_256-NEXT:    ld1d { z3.d }, p0/z, [x1]
; VBITS_GE_256-NEXT:    umax z0.d, p0/m, z0.d, z1.d
; VBITS_GE_256-NEXT:    movprfx z1, z2
; VBITS_GE_256-NEXT:    umax z1.d, p0/m, z1.d, z3.d
; VBITS_GE_256-NEXT:    st1d { z0.d }, p0, [x0, x8, lsl #3]
; VBITS_GE_256-NEXT:    st1d { z1.d }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: umax_v8i64:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.d, vl8
; VBITS_GE_512-NEXT:    ld1d { z0.d }, p0/z, [x0]
; VBITS_GE_512-NEXT:    ld1d { z1.d }, p0/z, [x1]
; VBITS_GE_512-NEXT:    umax z0.d, p0/m, z0.d, z1.d
; VBITS_GE_512-NEXT:    st1d { z0.d }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <8 x i64>, ptr %a
  %op2 = load <8 x i64>, ptr %b
  %res = call <8 x i64> @llvm.umax.v8i64(<8 x i64> %op1, <8 x i64> %op2)
  store <8 x i64> %res, ptr %a
  ret void
}

define void @umax_v16i64(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: umax_v16i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl16
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    ld1d { z1.d }, p0/z, [x1]
; CHECK-NEXT:    umax z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <16 x i64>, ptr %a
  %op2 = load <16 x i64>, ptr %b
  %res = call <16 x i64> @llvm.umax.v16i64(<16 x i64> %op1, <16 x i64> %op2)
  store <16 x i64> %res, ptr %a
  ret void
}

define void @umax_v32i64(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: umax_v32i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl32
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    ld1d { z1.d }, p0/z, [x1]
; CHECK-NEXT:    umax z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <32 x i64>, ptr %a
  %op2 = load <32 x i64>, ptr %b
  %res = call <32 x i64> @llvm.umax.v32i64(<32 x i64> %op1, <32 x i64> %op2)
  store <32 x i64> %res, ptr %a
  ret void
}

;
; UMIN
;

; Don't use SVE for 64-bit vectors.
define <8 x i8> @umin_v8i8(<8 x i8> %op1, <8 x i8> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: umin_v8i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umin v0.8b, v0.8b, v1.8b
; CHECK-NEXT:    ret
  %res = call <8 x i8> @llvm.umin.v8i8(<8 x i8> %op1, <8 x i8> %op2)
  ret <8 x i8> %res
}

; Don't use SVE for 128-bit vectors.
define <16 x i8> @umin_v16i8(<16 x i8> %op1, <16 x i8> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: umin_v16i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umin v0.16b, v0.16b, v1.16b
; CHECK-NEXT:    ret
  %res = call <16 x i8> @llvm.umin.v16i8(<16 x i8> %op1, <16 x i8> %op2)
  ret <16 x i8> %res
}

define void @umin_v32i8(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: umin_v32i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.b, vl32
; CHECK-NEXT:    ld1b { z0.b }, p0/z, [x0]
; CHECK-NEXT:    ld1b { z1.b }, p0/z, [x1]
; CHECK-NEXT:    umin z0.b, p0/m, z0.b, z1.b
; CHECK-NEXT:    st1b { z0.b }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <32 x i8>, ptr %a
  %op2 = load <32 x i8>, ptr %b
  %res = call <32 x i8> @llvm.umin.v32i8(<32 x i8> %op1, <32 x i8> %op2)
  store <32 x i8> %res, ptr %a
  ret void
}

define void @umin_v64i8(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: umin_v64i8:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.b, vl32
; VBITS_GE_256-NEXT:    mov w8, #32 // =0x20
; VBITS_GE_256-NEXT:    ld1b { z0.b }, p0/z, [x0, x8]
; VBITS_GE_256-NEXT:    ld1b { z1.b }, p0/z, [x1, x8]
; VBITS_GE_256-NEXT:    ld1b { z2.b }, p0/z, [x0]
; VBITS_GE_256-NEXT:    ld1b { z3.b }, p0/z, [x1]
; VBITS_GE_256-NEXT:    umin z0.b, p0/m, z0.b, z1.b
; VBITS_GE_256-NEXT:    movprfx z1, z2
; VBITS_GE_256-NEXT:    umin z1.b, p0/m, z1.b, z3.b
; VBITS_GE_256-NEXT:    st1b { z0.b }, p0, [x0, x8]
; VBITS_GE_256-NEXT:    st1b { z1.b }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: umin_v64i8:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.b, vl64
; VBITS_GE_512-NEXT:    ld1b { z0.b }, p0/z, [x0]
; VBITS_GE_512-NEXT:    ld1b { z1.b }, p0/z, [x1]
; VBITS_GE_512-NEXT:    umin z0.b, p0/m, z0.b, z1.b
; VBITS_GE_512-NEXT:    st1b { z0.b }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <64 x i8>, ptr %a
  %op2 = load <64 x i8>, ptr %b
  %res = call <64 x i8> @llvm.umin.v64i8(<64 x i8> %op1, <64 x i8> %op2)
  store <64 x i8> %res, ptr %a
  ret void
}

define void @umin_v128i8(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: umin_v128i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.b, vl128
; CHECK-NEXT:    ld1b { z0.b }, p0/z, [x0]
; CHECK-NEXT:    ld1b { z1.b }, p0/z, [x1]
; CHECK-NEXT:    umin z0.b, p0/m, z0.b, z1.b
; CHECK-NEXT:    st1b { z0.b }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <128 x i8>, ptr %a
  %op2 = load <128 x i8>, ptr %b
  %res = call <128 x i8> @llvm.umin.v128i8(<128 x i8> %op1, <128 x i8> %op2)
  store <128 x i8> %res, ptr %a
  ret void
}

define void @umin_v256i8(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: umin_v256i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.b, vl256
; CHECK-NEXT:    ld1b { z0.b }, p0/z, [x0]
; CHECK-NEXT:    ld1b { z1.b }, p0/z, [x1]
; CHECK-NEXT:    umin z0.b, p0/m, z0.b, z1.b
; CHECK-NEXT:    st1b { z0.b }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <256 x i8>, ptr %a
  %op2 = load <256 x i8>, ptr %b
  %res = call <256 x i8> @llvm.umin.v256i8(<256 x i8> %op1, <256 x i8> %op2)
  store <256 x i8> %res, ptr %a
  ret void
}

; Don't use SVE for 64-bit vectors.
define <4 x i16> @umin_v4i16(<4 x i16> %op1, <4 x i16> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: umin_v4i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umin v0.4h, v0.4h, v1.4h
; CHECK-NEXT:    ret
  %res = call <4 x i16> @llvm.umin.v4i16(<4 x i16> %op1, <4 x i16> %op2)
  ret <4 x i16> %res
}

; Don't use SVE for 128-bit vectors.
define <8 x i16> @umin_v8i16(<8 x i16> %op1, <8 x i16> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: umin_v8i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umin v0.8h, v0.8h, v1.8h
; CHECK-NEXT:    ret
  %res = call <8 x i16> @llvm.umin.v8i16(<8 x i16> %op1, <8 x i16> %op2)
  ret <8 x i16> %res
}

define void @umin_v16i16(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: umin_v16i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl16
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    ld1h { z1.h }, p0/z, [x1]
; CHECK-NEXT:    umin z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <16 x i16>, ptr %a
  %op2 = load <16 x i16>, ptr %b
  %res = call <16 x i16> @llvm.umin.v16i16(<16 x i16> %op1, <16 x i16> %op2)
  store <16 x i16> %res, ptr %a
  ret void
}

define void @umin_v32i16(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: umin_v32i16:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.h, vl16
; VBITS_GE_256-NEXT:    mov x8, #16 // =0x10
; VBITS_GE_256-NEXT:    ld1h { z0.h }, p0/z, [x0, x8, lsl #1]
; VBITS_GE_256-NEXT:    ld1h { z1.h }, p0/z, [x1, x8, lsl #1]
; VBITS_GE_256-NEXT:    ld1h { z2.h }, p0/z, [x0]
; VBITS_GE_256-NEXT:    ld1h { z3.h }, p0/z, [x1]
; VBITS_GE_256-NEXT:    umin z0.h, p0/m, z0.h, z1.h
; VBITS_GE_256-NEXT:    movprfx z1, z2
; VBITS_GE_256-NEXT:    umin z1.h, p0/m, z1.h, z3.h
; VBITS_GE_256-NEXT:    st1h { z0.h }, p0, [x0, x8, lsl #1]
; VBITS_GE_256-NEXT:    st1h { z1.h }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: umin_v32i16:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.h, vl32
; VBITS_GE_512-NEXT:    ld1h { z0.h }, p0/z, [x0]
; VBITS_GE_512-NEXT:    ld1h { z1.h }, p0/z, [x1]
; VBITS_GE_512-NEXT:    umin z0.h, p0/m, z0.h, z1.h
; VBITS_GE_512-NEXT:    st1h { z0.h }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <32 x i16>, ptr %a
  %op2 = load <32 x i16>, ptr %b
  %res = call <32 x i16> @llvm.umin.v32i16(<32 x i16> %op1, <32 x i16> %op2)
  store <32 x i16> %res, ptr %a
  ret void
}

define void @umin_v64i16(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: umin_v64i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl64
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    ld1h { z1.h }, p0/z, [x1]
; CHECK-NEXT:    umin z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <64 x i16>, ptr %a
  %op2 = load <64 x i16>, ptr %b
  %res = call <64 x i16> @llvm.umin.v64i16(<64 x i16> %op1, <64 x i16> %op2)
  store <64 x i16> %res, ptr %a
  ret void
}

define void @umin_v128i16(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: umin_v128i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl128
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    ld1h { z1.h }, p0/z, [x1]
; CHECK-NEXT:    umin z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <128 x i16>, ptr %a
  %op2 = load <128 x i16>, ptr %b
  %res = call <128 x i16> @llvm.umin.v128i16(<128 x i16> %op1, <128 x i16> %op2)
  store <128 x i16> %res, ptr %a
  ret void
}

; Don't use SVE for 64-bit vectors.
define <2 x i32> @umin_v2i32(<2 x i32> %op1, <2 x i32> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: umin_v2i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umin v0.2s, v0.2s, v1.2s
; CHECK-NEXT:    ret
  %res = call <2 x i32> @llvm.umin.v2i32(<2 x i32> %op1, <2 x i32> %op2)
  ret <2 x i32> %res
}

; Don't use SVE for 128-bit vectors.
define <4 x i32> @umin_v4i32(<4 x i32> %op1, <4 x i32> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: umin_v4i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umin v0.4s, v0.4s, v1.4s
; CHECK-NEXT:    ret
  %res = call <4 x i32> @llvm.umin.v4i32(<4 x i32> %op1, <4 x i32> %op2)
  ret <4 x i32> %res
}

define void @umin_v8i32(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: umin_v8i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl8
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    ld1w { z1.s }, p0/z, [x1]
; CHECK-NEXT:    umin z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <8 x i32>, ptr %a
  %op2 = load <8 x i32>, ptr %b
  %res = call <8 x i32> @llvm.umin.v8i32(<8 x i32> %op1, <8 x i32> %op2)
  store <8 x i32> %res, ptr %a
  ret void
}

define void @umin_v16i32(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: umin_v16i32:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.s, vl8
; VBITS_GE_256-NEXT:    mov x8, #8 // =0x8
; VBITS_GE_256-NEXT:    ld1w { z0.s }, p0/z, [x0, x8, lsl #2]
; VBITS_GE_256-NEXT:    ld1w { z1.s }, p0/z, [x1, x8, lsl #2]
; VBITS_GE_256-NEXT:    ld1w { z2.s }, p0/z, [x0]
; VBITS_GE_256-NEXT:    ld1w { z3.s }, p0/z, [x1]
; VBITS_GE_256-NEXT:    umin z0.s, p0/m, z0.s, z1.s
; VBITS_GE_256-NEXT:    movprfx z1, z2
; VBITS_GE_256-NEXT:    umin z1.s, p0/m, z1.s, z3.s
; VBITS_GE_256-NEXT:    st1w { z0.s }, p0, [x0, x8, lsl #2]
; VBITS_GE_256-NEXT:    st1w { z1.s }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: umin_v16i32:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.s, vl16
; VBITS_GE_512-NEXT:    ld1w { z0.s }, p0/z, [x0]
; VBITS_GE_512-NEXT:    ld1w { z1.s }, p0/z, [x1]
; VBITS_GE_512-NEXT:    umin z0.s, p0/m, z0.s, z1.s
; VBITS_GE_512-NEXT:    st1w { z0.s }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <16 x i32>, ptr %a
  %op2 = load <16 x i32>, ptr %b
  %res = call <16 x i32> @llvm.umin.v16i32(<16 x i32> %op1, <16 x i32> %op2)
  store <16 x i32> %res, ptr %a
  ret void
}

define void @umin_v32i32(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: umin_v32i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl32
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    ld1w { z1.s }, p0/z, [x1]
; CHECK-NEXT:    umin z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <32 x i32>, ptr %a
  %op2 = load <32 x i32>, ptr %b
  %res = call <32 x i32> @llvm.umin.v32i32(<32 x i32> %op1, <32 x i32> %op2)
  store <32 x i32> %res, ptr %a
  ret void
}

define void @umin_v64i32(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: umin_v64i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl64
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    ld1w { z1.s }, p0/z, [x1]
; CHECK-NEXT:    umin z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <64 x i32>, ptr %a
  %op2 = load <64 x i32>, ptr %b
  %res = call <64 x i32> @llvm.umin.v64i32(<64 x i32> %op1, <64 x i32> %op2)
  store <64 x i32> %res, ptr %a
  ret void
}

; Vector i64 min are not legal for NEON so use SVE when available.
define <1 x i64> @umin_v1i64(<1 x i64> %op1, <1 x i64> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: umin_v1i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl1
; CHECK-NEXT:    // kill: def $d0 killed $d0 def $z0
; CHECK-NEXT:    // kill: def $d1 killed $d1 def $z1
; CHECK-NEXT:    umin z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $z0
; CHECK-NEXT:    ret
  %res = call <1 x i64> @llvm.umin.v1i64(<1 x i64> %op1, <1 x i64> %op2)
  ret <1 x i64> %res
}

; Vector i64 min are not legal for NEON so use SVE when available.
define <2 x i64> @umin_v2i64(<2 x i64> %op1, <2 x i64> %op2) vscale_range(2,0) #0 {
; CHECK-LABEL: umin_v2i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl2
; CHECK-NEXT:    // kill: def $q0 killed $q0 def $z0
; CHECK-NEXT:    // kill: def $q1 killed $q1 def $z1
; CHECK-NEXT:    umin z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    // kill: def $q0 killed $q0 killed $z0
; CHECK-NEXT:    ret
  %res = call <2 x i64> @llvm.umin.v2i64(<2 x i64> %op1, <2 x i64> %op2)
  ret <2 x i64> %res
}

define void @umin_v4i64(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: umin_v4i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    ld1d { z1.d }, p0/z, [x1]
; CHECK-NEXT:    umin z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <4 x i64>, ptr %a
  %op2 = load <4 x i64>, ptr %b
  %res = call <4 x i64> @llvm.umin.v4i64(<4 x i64> %op1, <4 x i64> %op2)
  store <4 x i64> %res, ptr %a
  ret void
}

define void @umin_v8i64(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: umin_v8i64:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.d, vl4
; VBITS_GE_256-NEXT:    mov x8, #4 // =0x4
; VBITS_GE_256-NEXT:    ld1d { z0.d }, p0/z, [x0, x8, lsl #3]
; VBITS_GE_256-NEXT:    ld1d { z1.d }, p0/z, [x1, x8, lsl #3]
; VBITS_GE_256-NEXT:    ld1d { z2.d }, p0/z, [x0]
; VBITS_GE_256-NEXT:    ld1d { z3.d }, p0/z, [x1]
; VBITS_GE_256-NEXT:    umin z0.d, p0/m, z0.d, z1.d
; VBITS_GE_256-NEXT:    movprfx z1, z2
; VBITS_GE_256-NEXT:    umin z1.d, p0/m, z1.d, z3.d
; VBITS_GE_256-NEXT:    st1d { z0.d }, p0, [x0, x8, lsl #3]
; VBITS_GE_256-NEXT:    st1d { z1.d }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: umin_v8i64:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.d, vl8
; VBITS_GE_512-NEXT:    ld1d { z0.d }, p0/z, [x0]
; VBITS_GE_512-NEXT:    ld1d { z1.d }, p0/z, [x1]
; VBITS_GE_512-NEXT:    umin z0.d, p0/m, z0.d, z1.d
; VBITS_GE_512-NEXT:    st1d { z0.d }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <8 x i64>, ptr %a
  %op2 = load <8 x i64>, ptr %b
  %res = call <8 x i64> @llvm.umin.v8i64(<8 x i64> %op1, <8 x i64> %op2)
  store <8 x i64> %res, ptr %a
  ret void
}

define void @umin_v16i64(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: umin_v16i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl16
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    ld1d { z1.d }, p0/z, [x1]
; CHECK-NEXT:    umin z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <16 x i64>, ptr %a
  %op2 = load <16 x i64>, ptr %b
  %res = call <16 x i64> @llvm.umin.v16i64(<16 x i64> %op1, <16 x i64> %op2)
  store <16 x i64> %res, ptr %a
  ret void
}

define void @umin_v32i64(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: umin_v32i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl32
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    ld1d { z1.d }, p0/z, [x1]
; CHECK-NEXT:    umin z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op1 = load <32 x i64>, ptr %a
  %op2 = load <32 x i64>, ptr %b
  %res = call <32 x i64> @llvm.umin.v32i64(<32 x i64> %op1, <32 x i64> %op2)
  store <32 x i64> %res, ptr %a
  ret void
}

attributes #0 = { "target-features"="+sve" }

declare <8 x i8> @llvm.smin.v8i8(<8 x i8>, <8 x i8>)
declare <16 x i8> @llvm.smin.v16i8(<16 x i8>, <16 x i8>)
declare <32 x i8> @llvm.smin.v32i8(<32 x i8>, <32 x i8>)
declare <64 x i8> @llvm.smin.v64i8(<64 x i8>, <64 x i8>)
declare <128 x i8> @llvm.smin.v128i8(<128 x i8>, <128 x i8>)
declare <256 x i8> @llvm.smin.v256i8(<256 x i8>, <256 x i8>)
declare <4 x i16> @llvm.smin.v4i16(<4 x i16>, <4 x i16>)
declare <8 x i16> @llvm.smin.v8i16(<8 x i16>, <8 x i16>)
declare <16 x i16> @llvm.smin.v16i16(<16 x i16>, <16 x i16>)
declare <32 x i16> @llvm.smin.v32i16(<32 x i16>, <32 x i16>)
declare <64 x i16> @llvm.smin.v64i16(<64 x i16>, <64 x i16>)
declare <128 x i16> @llvm.smin.v128i16(<128 x i16>, <128 x i16>)
declare <2 x i32> @llvm.smin.v2i32(<2 x i32>, <2 x i32>)
declare <4 x i32> @llvm.smin.v4i32(<4 x i32>, <4 x i32>)
declare <8 x i32> @llvm.smin.v8i32(<8 x i32>, <8 x i32>)
declare <16 x i32> @llvm.smin.v16i32(<16 x i32>, <16 x i32>)
declare <32 x i32> @llvm.smin.v32i32(<32 x i32>, <32 x i32>)
declare <64 x i32> @llvm.smin.v64i32(<64 x i32>, <64 x i32>)
declare <1 x i64> @llvm.smin.v1i64(<1 x i64>, <1 x i64>)
declare <2 x i64> @llvm.smin.v2i64(<2 x i64>, <2 x i64>)
declare <4 x i64> @llvm.smin.v4i64(<4 x i64>, <4 x i64>)
declare <8 x i64> @llvm.smin.v8i64(<8 x i64>, <8 x i64>)
declare <16 x i64> @llvm.smin.v16i64(<16 x i64>, <16 x i64>)
declare <32 x i64> @llvm.smin.v32i64(<32 x i64>, <32 x i64>)

declare <8 x i8> @llvm.smax.v8i8(<8 x i8>, <8 x i8>)
declare <16 x i8> @llvm.smax.v16i8(<16 x i8>, <16 x i8>)
declare <32 x i8> @llvm.smax.v32i8(<32 x i8>, <32 x i8>)
declare <64 x i8> @llvm.smax.v64i8(<64 x i8>, <64 x i8>)
declare <128 x i8> @llvm.smax.v128i8(<128 x i8>, <128 x i8>)
declare <256 x i8> @llvm.smax.v256i8(<256 x i8>, <256 x i8>)
declare <4 x i16> @llvm.smax.v4i16(<4 x i16>, <4 x i16>)
declare <8 x i16> @llvm.smax.v8i16(<8 x i16>, <8 x i16>)
declare <16 x i16> @llvm.smax.v16i16(<16 x i16>, <16 x i16>)
declare <32 x i16> @llvm.smax.v32i16(<32 x i16>, <32 x i16>)
declare <64 x i16> @llvm.smax.v64i16(<64 x i16>, <64 x i16>)
declare <128 x i16> @llvm.smax.v128i16(<128 x i16>, <128 x i16>)
declare <2 x i32> @llvm.smax.v2i32(<2 x i32>, <2 x i32>)
declare <4 x i32> @llvm.smax.v4i32(<4 x i32>, <4 x i32>)
declare <8 x i32> @llvm.smax.v8i32(<8 x i32>, <8 x i32>)
declare <16 x i32> @llvm.smax.v16i32(<16 x i32>, <16 x i32>)
declare <32 x i32> @llvm.smax.v32i32(<32 x i32>, <32 x i32>)
declare <64 x i32> @llvm.smax.v64i32(<64 x i32>, <64 x i32>)
declare <1 x i64> @llvm.smax.v1i64(<1 x i64>, <1 x i64>)
declare <2 x i64> @llvm.smax.v2i64(<2 x i64>, <2 x i64>)
declare <4 x i64> @llvm.smax.v4i64(<4 x i64>, <4 x i64>)
declare <8 x i64> @llvm.smax.v8i64(<8 x i64>, <8 x i64>)
declare <16 x i64> @llvm.smax.v16i64(<16 x i64>, <16 x i64>)
declare <32 x i64> @llvm.smax.v32i64(<32 x i64>, <32 x i64>)

declare <8 x i8> @llvm.umin.v8i8(<8 x i8>, <8 x i8>)
declare <16 x i8> @llvm.umin.v16i8(<16 x i8>, <16 x i8>)
declare <32 x i8> @llvm.umin.v32i8(<32 x i8>, <32 x i8>)
declare <64 x i8> @llvm.umin.v64i8(<64 x i8>, <64 x i8>)
declare <128 x i8> @llvm.umin.v128i8(<128 x i8>, <128 x i8>)
declare <256 x i8> @llvm.umin.v256i8(<256 x i8>, <256 x i8>)
declare <4 x i16> @llvm.umin.v4i16(<4 x i16>, <4 x i16>)
declare <8 x i16> @llvm.umin.v8i16(<8 x i16>, <8 x i16>)
declare <16 x i16> @llvm.umin.v16i16(<16 x i16>, <16 x i16>)
declare <32 x i16> @llvm.umin.v32i16(<32 x i16>, <32 x i16>)
declare <64 x i16> @llvm.umin.v64i16(<64 x i16>, <64 x i16>)
declare <128 x i16> @llvm.umin.v128i16(<128 x i16>, <128 x i16>)
declare <2 x i32> @llvm.umin.v2i32(<2 x i32>, <2 x i32>)
declare <4 x i32> @llvm.umin.v4i32(<4 x i32>, <4 x i32>)
declare <8 x i32> @llvm.umin.v8i32(<8 x i32>, <8 x i32>)
declare <16 x i32> @llvm.umin.v16i32(<16 x i32>, <16 x i32>)
declare <32 x i32> @llvm.umin.v32i32(<32 x i32>, <32 x i32>)
declare <64 x i32> @llvm.umin.v64i32(<64 x i32>, <64 x i32>)
declare <1 x i64> @llvm.umin.v1i64(<1 x i64>, <1 x i64>)
declare <2 x i64> @llvm.umin.v2i64(<2 x i64>, <2 x i64>)
declare <4 x i64> @llvm.umin.v4i64(<4 x i64>, <4 x i64>)
declare <8 x i64> @llvm.umin.v8i64(<8 x i64>, <8 x i64>)
declare <16 x i64> @llvm.umin.v16i64(<16 x i64>, <16 x i64>)
declare <32 x i64> @llvm.umin.v32i64(<32 x i64>, <32 x i64>)

declare <8 x i8> @llvm.umax.v8i8(<8 x i8>, <8 x i8>)
declare <16 x i8> @llvm.umax.v16i8(<16 x i8>, <16 x i8>)
declare <32 x i8> @llvm.umax.v32i8(<32 x i8>, <32 x i8>)
declare <64 x i8> @llvm.umax.v64i8(<64 x i8>, <64 x i8>)
declare <128 x i8> @llvm.umax.v128i8(<128 x i8>, <128 x i8>)
declare <256 x i8> @llvm.umax.v256i8(<256 x i8>, <256 x i8>)
declare <4 x i16> @llvm.umax.v4i16(<4 x i16>, <4 x i16>)
declare <8 x i16> @llvm.umax.v8i16(<8 x i16>, <8 x i16>)
declare <16 x i16> @llvm.umax.v16i16(<16 x i16>, <16 x i16>)
declare <32 x i16> @llvm.umax.v32i16(<32 x i16>, <32 x i16>)
declare <64 x i16> @llvm.umax.v64i16(<64 x i16>, <64 x i16>)
declare <128 x i16> @llvm.umax.v128i16(<128 x i16>, <128 x i16>)
declare <2 x i32> @llvm.umax.v2i32(<2 x i32>, <2 x i32>)
declare <4 x i32> @llvm.umax.v4i32(<4 x i32>, <4 x i32>)
declare <8 x i32> @llvm.umax.v8i32(<8 x i32>, <8 x i32>)
declare <16 x i32> @llvm.umax.v16i32(<16 x i32>, <16 x i32>)
declare <32 x i32> @llvm.umax.v32i32(<32 x i32>, <32 x i32>)
declare <64 x i32> @llvm.umax.v64i32(<64 x i32>, <64 x i32>)
declare <1 x i64> @llvm.umax.v1i64(<1 x i64>, <1 x i64>)
declare <2 x i64> @llvm.umax.v2i64(<2 x i64>, <2 x i64>)
declare <4 x i64> @llvm.umax.v4i64(<4 x i64>, <4 x i64>)
declare <8 x i64> @llvm.umax.v8i64(<8 x i64>, <8 x i64>)
declare <16 x i64> @llvm.umax.v16i64(<16 x i64>, <16 x i64>)
declare <32 x i64> @llvm.umax.v32i64(<32 x i64>, <32 x i64>)