; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sve < %s | FileCheck %s

; ADD

define <vscale x 16 x i8> @add_i8(<vscale x 16 x i8> %a) {
; CHECK-LABEL: add_i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    add z0.b, z0.b, #127 // =0x7f
; CHECK-NEXT:    ret
  %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
  %elt = insertelement <vscale x 16 x i8> undef, i8 127, i32 0
  %splat = shufflevector <vscale x 16 x i8> %elt, <vscale x 16 x i8> undef, <vscale x 16 x i32> zeroinitializer
  %out = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> %pg,
                                                               <vscale x 16 x i8> %a,
                                                               <vscale x 16 x i8> %splat)
  ret <vscale x 16 x i8> %out
}

define <vscale x 8 x i16> @add_i16(<vscale x 8 x i16> %a) {
; CHECK-LABEL: add_i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    add z0.h, z0.h, #127 // =0x7f
; CHECK-NEXT:    ret
  %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %elt = insertelement <vscale x 8 x i16> undef, i16 127, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> %pg,
                                                               <vscale x 8 x i16> %a,
                                                               <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 8 x i16> @add_i16_out_of_range(<vscale x 8 x i16> %a) {
; CHECK-LABEL: add_i16_out_of_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, #257
; CHECK-NEXT:    mov z1.h, w8
; CHECK-NEXT:    add z0.h, z0.h, z1.h
; CHECK-NEXT:    ret
  %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %elt = insertelement <vscale x 8 x i16> undef, i16 257, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> %pg,
                                                               <vscale x 8 x i16> %a,
                                                               <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 4 x i32> @add_i32(<vscale x 4 x i32> %a) {
; CHECK-LABEL: add_i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    add z0.s, z0.s, #127 // =0x7f
; CHECK-NEXT:    ret
  %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
  %elt = insertelement <vscale x 4 x i32> undef, i32 127, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> %pg,
                                                               <vscale x 4 x i32> %a,
                                                               <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 4 x i32> @add_i32_out_of_range(<vscale x 4 x i32> %a) {
; CHECK-LABEL: add_i32_out_of_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, #257
; CHECK-NEXT:    mov z1.s, w8
; CHECK-NEXT:    add z0.s, z0.s, z1.s
; CHECK-NEXT:    ret
  %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
  %elt = insertelement <vscale x 4 x i32> undef, i32 257, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> %pg,
                                                               <vscale x 4 x i32> %a,
                                                               <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 2 x i64> @add_i64(<vscale x 2 x i64> %a) {
; CHECK-LABEL: add_i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    add z0.d, z0.d, #127 // =0x7f
; CHECK-NEXT:    ret
  %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %elt = insertelement <vscale x 2 x i64> undef, i64 127, i64 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> %pg,
                                                               <vscale x 2 x i64> %a,
                                                               <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

define <vscale x 2 x i64> @add_i64_out_of_range(<vscale x 2 x i64> %a) {
; CHECK-LABEL: add_i64_out_of_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, #257
; CHECK-NEXT:    mov z1.d, x8
; CHECK-NEXT:    add z0.d, z0.d, z1.d
; CHECK-NEXT:    ret
  %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %elt = insertelement <vscale x 2 x i64> undef, i64 257, i64 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> %pg,
                                                               <vscale x 2 x i64> %a,
                                                               <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

; SUB

define <vscale x 16 x i8> @sub_i8(<vscale x 16 x i8> %a) {
; CHECK-LABEL: sub_i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    sub z0.b, z0.b, #127 // =0x7f
; CHECK-NEXT:    ret
  %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
  %elt = insertelement <vscale x 16 x i8> undef, i8 127, i32 0
  %splat = shufflevector <vscale x 16 x i8> %elt, <vscale x 16 x i8> undef, <vscale x 16 x i32> zeroinitializer
  %out = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> %pg,
                                                               <vscale x 16 x i8> %a,
                                                               <vscale x 16 x i8> %splat)
  ret <vscale x 16 x i8> %out
}

define <vscale x 8 x i16> @sub_i16(<vscale x 8 x i16> %a) {
; CHECK-LABEL: sub_i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    sub z0.h, z0.h, #127 // =0x7f
; CHECK-NEXT:    ret
  %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %elt = insertelement <vscale x 8 x i16> undef, i16 127, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> %pg,
                                                               <vscale x 8 x i16> %a,
                                                               <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 8 x i16> @sub_i16_out_of_range(<vscale x 8 x i16> %a) {
; CHECK-LABEL: sub_i16_out_of_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, #257
; CHECK-NEXT:    mov z1.h, w8
; CHECK-NEXT:    sub z0.h, z0.h, z1.h
; CHECK-NEXT:    ret
  %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %elt = insertelement <vscale x 8 x i16> undef, i16 257, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> %pg,
                                                               <vscale x 8 x i16> %a,
                                                               <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 4 x i32> @sub_i32(<vscale x 4 x i32> %a) {
; CHECK-LABEL: sub_i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    sub z0.s, z0.s, #127 // =0x7f
; CHECK-NEXT:    ret
  %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
  %elt = insertelement <vscale x 4 x i32> undef, i32 127, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> %pg,
                                                               <vscale x 4 x i32> %a,
                                                               <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 4 x i32> @sub_i32_out_of_range(<vscale x 4 x i32> %a) {
; CHECK-LABEL: sub_i32_out_of_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, #257
; CHECK-NEXT:    mov z1.s, w8
; CHECK-NEXT:    sub z0.s, z0.s, z1.s
; CHECK-NEXT:    ret
  %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
  %elt = insertelement <vscale x 4 x i32> undef, i32 257, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> %pg,
                                                               <vscale x 4 x i32> %a,
                                                               <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 2 x i64> @sub_i64(<vscale x 2 x i64> %a) {
; CHECK-LABEL: sub_i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    sub z0.d, z0.d, #127 // =0x7f
; CHECK-NEXT:    ret
  %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %elt = insertelement <vscale x 2 x i64> undef, i64 127, i64 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> %pg,
                                                               <vscale x 2 x i64> %a,
                                                               <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

define <vscale x 2 x i64> @sub_i64_out_of_range(<vscale x 2 x i64> %a) {
; CHECK-LABEL: sub_i64_out_of_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, #257
; CHECK-NEXT:    mov z1.d, x8
; CHECK-NEXT:    sub z0.d, z0.d, z1.d
; CHECK-NEXT:    ret
  %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %elt = insertelement <vscale x 2 x i64> undef, i64 257, i64 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> %pg,
                                                               <vscale x 2 x i64> %a,
                                                               <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

; As sub_i32 but where pg is i8 based and thus compatible for i32.
define <vscale x 4 x i32> @sub_i32_ptrue_all_b(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: sub_i32_ptrue_all_b:
; CHECK:       // %bb.0:
; CHECK-NEXT:    sub z0.s, z0.s, #1 // =0x1
; CHECK-NEXT:    ret
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                    <vscale x 4 x i32> %a,
                                                                    <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

; As sub_i32 but where pg is i16 based and thus compatible for i32.
define <vscale x 4 x i32> @sub_i32_ptrue_all_h(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: sub_i32_ptrue_all_h:
; CHECK:       // %bb.0:
; CHECK-NEXT:    sub z0.s, z0.s, #1 // =0x1
; CHECK-NEXT:    ret
  %pg.h = tail call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> %pg.h)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                    <vscale x 4 x i32> %a,
                                                                    <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

; As sub_i32 but where pg is i64 based, which is not compatibile for i32 and
; thus inactive lanes are important and the immediate form cannot be used.
define <vscale x 4 x i32> @sub_i32_ptrue_all_d(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: sub_i32_ptrue_all_d:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d
; CHECK-NEXT:    mov z1.s, #1 // =0x1
; CHECK-NEXT:    sub z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    ret
  %pg.d = tail call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> %pg.d)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                    <vscale x 4 x i32> %a,
                                                                    <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

; SUBR

define <vscale x 16 x i8> @subr_i8(<vscale x 16 x i8> %a) {
; CHECK-LABEL: subr_i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    subr z0.b, z0.b, #127 // =0x7f
; CHECK-NEXT:    ret
  %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
  %elt = insertelement <vscale x 16 x i8> undef, i8 127, i32 0
  %splat = shufflevector <vscale x 16 x i8> %elt, <vscale x 16 x i8> undef, <vscale x 16 x i32> zeroinitializer
  %out = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> %pg,
                                                                <vscale x 16 x i8> %a,
                                                                <vscale x 16 x i8> %splat)
  ret <vscale x 16 x i8> %out
}

define <vscale x 8 x i16> @subr_i16(<vscale x 8 x i16> %a) {
; CHECK-LABEL: subr_i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    subr z0.h, z0.h, #127 // =0x7f
; CHECK-NEXT:    ret
  %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %elt = insertelement <vscale x 8 x i16> undef, i16 127, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> %pg,
                                                                <vscale x 8 x i16> %a,
                                                                <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 8 x i16> @subr_i16_out_of_range(<vscale x 8 x i16> %a) {
; CHECK-LABEL: subr_i16_out_of_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, #257
; CHECK-NEXT:    mov z1.h, w8
; CHECK-NEXT:    sub z0.h, z1.h, z0.h
; CHECK-NEXT:    ret
  %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %elt = insertelement <vscale x 8 x i16> undef, i16 257, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> %pg,
                                                                <vscale x 8 x i16> %a,
                                                                <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 4 x i32> @subr_i32(<vscale x 4 x i32> %a) {
; CHECK-LABEL: subr_i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    subr z0.s, z0.s, #127 // =0x7f
; CHECK-NEXT:    ret
  %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
  %elt = insertelement <vscale x 4 x i32> undef, i32 127, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> %pg,
                                                                <vscale x 4 x i32> %a,
                                                                <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 4 x i32> @subr_i32_out_of_range(<vscale x 4 x i32> %a) {
; CHECK-LABEL: subr_i32_out_of_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, #257
; CHECK-NEXT:    mov z1.s, w8
; CHECK-NEXT:    sub z0.s, z1.s, z0.s
; CHECK-NEXT:    ret
  %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
  %elt = insertelement <vscale x 4 x i32> undef, i32 257, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> %pg,
                                                                <vscale x 4 x i32> %a,
                                                                <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 2 x i64> @subr_i64(<vscale x 2 x i64> %a) {
; CHECK-LABEL: subr_i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    subr z0.d, z0.d, #127 // =0x7f
; CHECK-NEXT:    ret
  %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %elt = insertelement <vscale x 2 x i64> undef, i64 127, i64 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> %pg,
                                                                <vscale x 2 x i64> %a,
                                                                <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

define <vscale x 2 x i64> @subr_i64_out_of_range(<vscale x 2 x i64> %a) {
; CHECK-LABEL: subr_i64_out_of_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, #257
; CHECK-NEXT:    mov z1.d, x8
; CHECK-NEXT:    sub z0.d, z1.d, z0.d
; CHECK-NEXT:    ret
  %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %elt = insertelement <vscale x 2 x i64> undef, i64 257, i64 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> %pg,
                                                                <vscale x 2 x i64> %a,
                                                                <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

; As subr_i32 but where pg is i8 based and thus compatible for i32.
define <vscale x 4 x i32> @subr_i32_ptrue_all_b(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: subr_i32_ptrue_all_b:
; CHECK:       // %bb.0:
; CHECK-NEXT:    subr z0.s, z0.s, #1 // =0x1
; CHECK-NEXT:    ret
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                     <vscale x 4 x i32> %a,
                                                                     <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

; As subr_i32 but where pg is i16 based and thus compatible for i32.
define <vscale x 4 x i32> @subr_i32_ptrue_all_h(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: subr_i32_ptrue_all_h:
; CHECK:       // %bb.0:
; CHECK-NEXT:    subr z0.s, z0.s, #1 // =0x1
; CHECK-NEXT:    ret
  %pg.h = tail call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> %pg.h)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                     <vscale x 4 x i32> %a,
                                                                     <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

; As subr_i32 but where pg is i64 based, which is not compatibile for i32 and
; thus inactive lanes are important and the immediate form cannot be used.
define <vscale x 4 x i32> @subr_i32_ptrue_all_d(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: subr_i32_ptrue_all_d:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d
; CHECK-NEXT:    mov z1.s, #1 // =0x1
; CHECK-NEXT:    subr z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    ret
  %pg.d = tail call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> %pg.d)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                     <vscale x 4 x i32> %a,
                                                                     <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

; SMAX

define <vscale x 16 x i8> @smax_i8(<vscale x 16 x i8> %a) {
; CHECK-LABEL: smax_i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smax z0.b, z0.b, #-128
; CHECK-NEXT:    ret
  %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
  %elt = insertelement <vscale x 16 x i8> undef, i8 -128, i32 0
  %splat = shufflevector <vscale x 16 x i8> %elt, <vscale x 16 x i8> undef, <vscale x 16 x i32> zeroinitializer
  %out = call <vscale x 16 x i8> @llvm.aarch64.sve.smax.nxv16i8(<vscale x 16 x i1> %pg,
                                                                <vscale x 16 x i8> %a,
                                                                <vscale x 16 x i8> %splat)
  ret <vscale x 16 x i8> %out
}

define <vscale x 8 x i16> @smax_i16(<vscale x 8 x i16> %a) {
; CHECK-LABEL: smax_i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smax z0.h, z0.h, #127
; CHECK-NEXT:    ret
  %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %elt = insertelement <vscale x 8 x i16> undef, i16 127, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.smax.nxv8i16(<vscale x 8 x i1> %pg,
                                                                <vscale x 8 x i16> %a,
                                                                <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 8 x i16> @smax_i16_out_of_range(<vscale x 8 x i16> %a) {
; CHECK-LABEL: smax_i16_out_of_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, #129
; CHECK-NEXT:    ptrue p0.h
; CHECK-NEXT:    mov z1.h, w8
; CHECK-NEXT:    smax z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    ret
  %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %elt = insertelement <vscale x 8 x i16> undef, i16 129, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.smax.nxv8i16(<vscale x 8 x i1> %pg,
                                                                <vscale x 8 x i16> %a,
                                                                <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 4 x i32> @smax_i32(<vscale x 4 x i32> %a) {
; CHECK-LABEL: smax_i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smax z0.s, z0.s, #-128
; CHECK-NEXT:    ret
  %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
  %elt = insertelement <vscale x 4 x i32> undef, i32 -128, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> %pg,
                                                                <vscale x 4 x i32> %a,
                                                                <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 4 x i32> @smax_i32_out_of_range(<vscale x 4 x i32> %a) {
; CHECK-LABEL: smax_i32_out_of_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, #-129
; CHECK-NEXT:    ptrue p0.s
; CHECK-NEXT:    mov z1.s, w8
; CHECK-NEXT:    smax z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    ret
  %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
  %elt = insertelement <vscale x 4 x i32> undef, i32 -129, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> %pg,
                                                                <vscale x 4 x i32> %a,
                                                                <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 2 x i64> @smax_i64(<vscale x 2 x i64> %a) {
; CHECK-LABEL: smax_i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smax z0.d, z0.d, #127
; CHECK-NEXT:    ret
  %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %elt = insertelement <vscale x 2 x i64> undef, i64 127, i64 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.smax.nxv2i64(<vscale x 2 x i1> %pg,
                                                                <vscale x 2 x i64> %a,
                                                                <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

define <vscale x 2 x i64> @smax_i64_out_of_range(<vscale x 2 x i64> %a) {
; CHECK-LABEL: smax_i64_out_of_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, #65535
; CHECK-NEXT:    ptrue p0.d
; CHECK-NEXT:    mov z1.d, x8
; CHECK-NEXT:    smax z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    ret
  %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %elt = insertelement <vscale x 2 x i64> undef, i64 65535, i64 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.smax.nxv2i64(<vscale x 2 x i1> %pg,
                                                                <vscale x 2 x i64> %a,
                                                                <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

; As smax_i32 but where pg is i8 based and thus compatible for i32.
define <vscale x 4 x i32> @smax_i32_ptrue_all_b(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: smax_i32_ptrue_all_b:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smax z0.s, z0.s, #1
; CHECK-NEXT:    ret
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                     <vscale x 4 x i32> %a,
                                                                     <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

; As smax_i32 but where pg is i16 based and thus compatible for i32.
define <vscale x 4 x i32> @smax_i32_ptrue_all_h(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: smax_i32_ptrue_all_h:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smax z0.s, z0.s, #1
; CHECK-NEXT:    ret
  %pg.h = tail call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> %pg.h)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                     <vscale x 4 x i32> %a,
                                                                     <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

; As smax_i32 but where pg is i64 based, which is not compatibile for i32 and
; thus inactive lanes are important and the immediate form cannot be used.
define <vscale x 4 x i32> @smax_i32_ptrue_all_d(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: smax_i32_ptrue_all_d:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d
; CHECK-NEXT:    mov z1.s, #1 // =0x1
; CHECK-NEXT:    smax z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    ret
  %pg.d = tail call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> %pg.d)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                     <vscale x 4 x i32> %a,
                                                                     <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

; SMIN

define <vscale x 16 x i8> @smin_i8(<vscale x 16 x i8> %a) {
; CHECK-LABEL: smin_i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smin z0.b, z0.b, #127
; CHECK-NEXT:    ret
  %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
  %elt = insertelement <vscale x 16 x i8> undef, i8 127, i32 0
  %splat = shufflevector <vscale x 16 x i8> %elt, <vscale x 16 x i8> undef, <vscale x 16 x i32> zeroinitializer
  %out = call <vscale x 16 x i8> @llvm.aarch64.sve.smin.nxv16i8(<vscale x 16 x i1> %pg,
                                                                <vscale x 16 x i8> %a,
                                                                <vscale x 16 x i8> %splat)
  ret <vscale x 16 x i8> %out
}

define <vscale x 8 x i16> @smin_i16(<vscale x 8 x i16> %a) {
; CHECK-LABEL: smin_i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smin z0.h, z0.h, #-128
; CHECK-NEXT:    ret
  %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %elt = insertelement <vscale x 8 x i16> undef, i16 -128, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.smin.nxv8i16(<vscale x 8 x i1> %pg,
                                                                <vscale x 8 x i16> %a,
                                                                <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 8 x i16> @smin_i16_out_of_range(<vscale x 8 x i16> %a) {
; CHECK-LABEL: smin_i16_out_of_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, #-129
; CHECK-NEXT:    ptrue p0.h
; CHECK-NEXT:    mov z1.h, w8
; CHECK-NEXT:    smin z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    ret
  %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %elt = insertelement <vscale x 8 x i16> undef, i16 -129, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.smin.nxv8i16(<vscale x 8 x i1> %pg,
                                                                <vscale x 8 x i16> %a,
                                                                <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 4 x i32> @smin_i32(<vscale x 4 x i32> %a) {
; CHECK-LABEL: smin_i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smin z0.s, z0.s, #127
; CHECK-NEXT:    ret
  %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
  %elt = insertelement <vscale x 4 x i32> undef, i32 127, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> %pg,
                                                                <vscale x 4 x i32> %a,
                                                                <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 4 x i32> @smin_i32_out_of_range(<vscale x 4 x i32> %a) {
; CHECK-LABEL: smin_i32_out_of_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, #257
; CHECK-NEXT:    ptrue p0.s
; CHECK-NEXT:    mov z1.s, w8
; CHECK-NEXT:    smin z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    ret
  %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
  %elt = insertelement <vscale x 4 x i32> undef, i32 257, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> %pg,
                                                                <vscale x 4 x i32> %a,
                                                                <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}


define <vscale x 2 x i64> @smin_i64(<vscale x 2 x i64> %a) {
; CHECK-LABEL: smin_i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smin z0.d, z0.d, #-128
; CHECK-NEXT:    ret
  %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %elt = insertelement <vscale x 2 x i64> undef, i64 -128, i64 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.smin.nxv2i64(<vscale x 2 x i1> %pg,
                                                                <vscale x 2 x i64> %a,
                                                                <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

define <vscale x 2 x i64> @smin_i64_out_of_range(<vscale x 2 x i64> %a) {
; CHECK-LABEL: smin_i64_out_of_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d
; CHECK-NEXT:    mov z1.d, #-256 // =0xffffffffffffff00
; CHECK-NEXT:    smin z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    ret
  %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %elt = insertelement <vscale x 2 x i64> undef, i64 -256, i64 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.smin.nxv2i64(<vscale x 2 x i1> %pg,
                                                                <vscale x 2 x i64> %a,
                                                                <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

; As smin_i32 but where pg is i8 based and thus compatible for i32.
define <vscale x 4 x i32> @smin_i32_ptrue_all_b(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: smin_i32_ptrue_all_b:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smin z0.s, z0.s, #1
; CHECK-NEXT:    ret
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                     <vscale x 4 x i32> %a,
                                                                     <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

; As smin_i32 but where pg is i16 based and thus compatible for i32.
define <vscale x 4 x i32> @smin_i32_ptrue_all_h(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: smin_i32_ptrue_all_h:
; CHECK:       // %bb.0:
; CHECK-NEXT:    smin z0.s, z0.s, #1
; CHECK-NEXT:    ret
  %pg.h = tail call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> %pg.h)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                     <vscale x 4 x i32> %a,
                                                                     <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

; As smin_i32 but where pg is i64 based, which is not compatibile for i32 and
; thus inactive lanes are important and the immediate form cannot be used.
define <vscale x 4 x i32> @smin_i32_ptrue_all_d(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: smin_i32_ptrue_all_d:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d
; CHECK-NEXT:    mov z1.s, #1 // =0x1
; CHECK-NEXT:    smin z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    ret
  %pg.d = tail call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> %pg.d)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                     <vscale x 4 x i32> %a,
                                                                     <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

; UMAX

define <vscale x 16 x i8> @umax_i8(<vscale x 16 x i8> %a) {
; CHECK-LABEL: umax_i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umax z0.b, z0.b, #0
; CHECK-NEXT:    ret
  %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
  %elt = insertelement <vscale x 16 x i8> undef, i8 0, i32 0
  %splat = shufflevector <vscale x 16 x i8> %elt, <vscale x 16 x i8> undef, <vscale x 16 x i32> zeroinitializer
  %out = call <vscale x 16 x i8> @llvm.aarch64.sve.umax.nxv16i8(<vscale x 16 x i1> %pg,
                                                                <vscale x 16 x i8> %a,
                                                                <vscale x 16 x i8> %splat)
  ret <vscale x 16 x i8> %out
}

define <vscale x 8 x i16> @umax_i16(<vscale x 8 x i16> %a) {
; CHECK-LABEL: umax_i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umax z0.h, z0.h, #255
; CHECK-NEXT:    ret
  %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %elt = insertelement <vscale x 8 x i16> undef, i16 255, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.umax.nxv8i16(<vscale x 8 x i1> %pg,
                                                                <vscale x 8 x i16> %a,
                                                                <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 8 x i16> @umax_i16_out_of_range(<vscale x 8 x i16> %a) {
; CHECK-LABEL: umax_i16_out_of_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, #257
; CHECK-NEXT:    ptrue p0.h
; CHECK-NEXT:    mov z1.h, w8
; CHECK-NEXT:    umax z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    ret
  %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %elt = insertelement <vscale x 8 x i16> undef, i16 257, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.umax.nxv8i16(<vscale x 8 x i1> %pg,
                                                                <vscale x 8 x i16> %a,
                                                                <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 4 x i32> @umax_i32(<vscale x 4 x i32> %a) {
; CHECK-LABEL: umax_i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umax z0.s, z0.s, #0
; CHECK-NEXT:    ret
  %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
  %elt = insertelement <vscale x 4 x i32> undef, i32 0, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> %pg,
                                                                <vscale x 4 x i32> %a,
                                                                <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 4 x i32> @umax_i32_out_of_range(<vscale x 4 x i32> %a) {
; CHECK-LABEL: umax_i32_out_of_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, #257
; CHECK-NEXT:    ptrue p0.s
; CHECK-NEXT:    mov z1.s, w8
; CHECK-NEXT:    umax z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    ret
  %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
  %elt = insertelement <vscale x 4 x i32> undef, i32 257, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> %pg,
                                                                <vscale x 4 x i32> %a,
                                                                <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 2 x i64> @umax_i64(<vscale x 2 x i64> %a) {
; CHECK-LABEL: umax_i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umax z0.d, z0.d, #255
; CHECK-NEXT:    ret
  %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %elt = insertelement <vscale x 2 x i64> undef, i64 255, i64 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.umax.nxv2i64(<vscale x 2 x i1> %pg,
                                                                <vscale x 2 x i64> %a,
                                                                <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

define <vscale x 2 x i64> @umax_i64_out_of_range(<vscale x 2 x i64> %a) {
; CHECK-LABEL: umax_i64_out_of_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, #65535
; CHECK-NEXT:    ptrue p0.d
; CHECK-NEXT:    mov z1.d, x8
; CHECK-NEXT:    umax z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    ret
  %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %elt = insertelement <vscale x 2 x i64> undef, i64 65535, i64 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.umax.nxv2i64(<vscale x 2 x i1> %pg,
                                                                <vscale x 2 x i64> %a,
                                                                <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

; As umax_i32 but where pg is i8 based and thus compatible for i32.
define <vscale x 4 x i32> @umax_i32_ptrue_all_b(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: umax_i32_ptrue_all_b:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umax z0.s, z0.s, #1
; CHECK-NEXT:    ret
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                     <vscale x 4 x i32> %a,
                                                                     <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

; As umax_i32 but where pg is i16 based and thus compatible for i32.
define <vscale x 4 x i32> @umax_i32_ptrue_all_h(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: umax_i32_ptrue_all_h:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umax z0.s, z0.s, #1
; CHECK-NEXT:    ret
  %pg.h = tail call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> %pg.h)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                     <vscale x 4 x i32> %a,
                                                                     <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

; As umax_i32 but where pg is i64 based, which is not compatibile for i32 and
; thus inactive lanes are important and the immediate form cannot be used.
define <vscale x 4 x i32> @umax_i32_ptrue_all_d(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: umax_i32_ptrue_all_d:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d
; CHECK-NEXT:    mov z1.s, #1 // =0x1
; CHECK-NEXT:    umax z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    ret
  %pg.d = tail call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> %pg.d)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                     <vscale x 4 x i32> %a,
                                                                     <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

; UMIN

define <vscale x 16 x i8> @umin_i8(<vscale x 16 x i8> %a) {
; CHECK-LABEL: umin_i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umin z0.b, z0.b, #255
; CHECK-NEXT:    ret
  %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
  %elt = insertelement <vscale x 16 x i8> undef, i8 255, i32 0
  %splat = shufflevector <vscale x 16 x i8> %elt, <vscale x 16 x i8> undef, <vscale x 16 x i32> zeroinitializer
  %out = call <vscale x 16 x i8> @llvm.aarch64.sve.umin.nxv16i8(<vscale x 16 x i1> %pg,
                                                                <vscale x 16 x i8> %a,
                                                                <vscale x 16 x i8> %splat)
  ret <vscale x 16 x i8> %out
}

define <vscale x 8 x i16> @umin_i16(<vscale x 8 x i16> %a) {
; CHECK-LABEL: umin_i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umin z0.h, z0.h, #0
; CHECK-NEXT:    ret
  %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %elt = insertelement <vscale x 8 x i16> undef, i16 0, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.umin.nxv8i16(<vscale x 8 x i1> %pg,
                                                                <vscale x 8 x i16> %a,
                                                                <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 8 x i16> @umin_i16_out_of_range(<vscale x 8 x i16> %a) {
; CHECK-LABEL: umin_i16_out_of_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, #257
; CHECK-NEXT:    ptrue p0.h
; CHECK-NEXT:    mov z1.h, w8
; CHECK-NEXT:    umin z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    ret
  %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %elt = insertelement <vscale x 8 x i16> undef, i16 257, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.umin.nxv8i16(<vscale x 8 x i1> %pg,
                                                                <vscale x 8 x i16> %a,
                                                                <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 4 x i32> @umin_i32(<vscale x 4 x i32> %a) {
; CHECK-LABEL: umin_i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umin z0.s, z0.s, #255
; CHECK-NEXT:    ret
  %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
  %elt = insertelement <vscale x 4 x i32> undef, i32 255, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> %pg,
                                                                <vscale x 4 x i32> %a,
                                                                <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 4 x i32> @umin_i32_out_of_range(<vscale x 4 x i32> %a) {
; CHECK-LABEL: umin_i32_out_of_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, #257
; CHECK-NEXT:    ptrue p0.s
; CHECK-NEXT:    mov z1.s, w8
; CHECK-NEXT:    umin z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    ret
  %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
  %elt = insertelement <vscale x 4 x i32> undef, i32 257, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> %pg,
                                                                <vscale x 4 x i32> %a,
                                                                <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 2 x i64> @umin_i64(<vscale x 2 x i64> %a) {
; CHECK-LABEL: umin_i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umin z0.d, z0.d, #0
; CHECK-NEXT:    ret
  %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %elt = insertelement <vscale x 2 x i64> undef, i64 0, i64 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.umin.nxv2i64(<vscale x 2 x i1> %pg,
                                                                <vscale x 2 x i64> %a,
                                                                <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

define <vscale x 2 x i64> @umin_i64_out_of_range(<vscale x 2 x i64> %a) {
; CHECK-LABEL: umin_i64_out_of_range:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov w8, #65535
; CHECK-NEXT:    ptrue p0.d
; CHECK-NEXT:    mov z1.d, x8
; CHECK-NEXT:    umin z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    ret
  %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %elt = insertelement <vscale x 2 x i64> undef, i64 65535, i64 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.umin.nxv2i64(<vscale x 2 x i1> %pg,
                                                                <vscale x 2 x i64> %a,
                                                                <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

; As umin_i32 but where pg is i8 based and thus compatible for i32.
define <vscale x 4 x i32> @umin_i32_ptrue_all_b(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: umin_i32_ptrue_all_b:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umin z0.s, z0.s, #1
; CHECK-NEXT:    ret
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                     <vscale x 4 x i32> %a,
                                                                     <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

; As umin_i32 but where pg is i16 based and thus compatible for i32.
define <vscale x 4 x i32> @umin_i32_ptrue_all_h(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: umin_i32_ptrue_all_h:
; CHECK:       // %bb.0:
; CHECK-NEXT:    umin z0.s, z0.s, #1
; CHECK-NEXT:    ret
  %pg.h = tail call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> %pg.h)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                     <vscale x 4 x i32> %a,
                                                                     <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

; As umin_i32 but where pg is i64 based, which is not compatibile for i32 and
; thus inactive lanes are important and the immediate form cannot be used.
define <vscale x 4 x i32> @umin_i32_ptrue_all_d(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: umin_i32_ptrue_all_d:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d
; CHECK-NEXT:    mov z1.s, #1 // =0x1
; CHECK-NEXT:    umin z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    ret
  %pg.d = tail call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> %pg.d)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                     <vscale x 4 x i32> %a,
                                                                     <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

; SQADD

define <vscale x 16 x i8> @sqadd_b_lowimm(<vscale x 16 x i8> %a) {
; CHECK-LABEL: sqadd_b_lowimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    sqadd z0.b, z0.b, #27 // =0x1b
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 16 x i8> undef, i8 27, i32 0
  %splat = shufflevector <vscale x 16 x i8> %elt, <vscale x 16 x i8> undef, <vscale x 16 x i32> zeroinitializer
  %out = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.x.nxv16i8(<vscale x 16 x i8> %a,
                                                                   <vscale x 16 x i8> %splat)
  ret <vscale x 16 x i8> %out
}

define <vscale x 8 x i16> @sqadd_h_lowimm(<vscale x 8 x i16> %a) {
; CHECK-LABEL: sqadd_h_lowimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    sqadd z0.h, z0.h, #43 // =0x2b
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 8 x i16> undef, i16 43, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.x.nxv8i16(<vscale x 8 x i16> %a,
                                                                   <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 8 x i16> @sqadd_h_highimm(<vscale x 8 x i16> %a) {
; CHECK-LABEL: sqadd_h_highimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    sqadd z0.h, z0.h, #2048 // =0x800
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 8 x i16> undef, i16 2048, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.x.nxv8i16(<vscale x 8 x i16> %a,
                                                                   <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 4 x i32> @sqadd_s_lowimm(<vscale x 4 x i32> %a) {
; CHECK-LABEL: sqadd_s_lowimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    sqadd z0.s, z0.s, #1 // =0x1
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 4 x i32> undef, i32 1, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.x.nxv4i32(<vscale x 4 x i32> %a,
                                                                   <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 4 x i32> @sqadd_s_highimm(<vscale x 4 x i32> %a) {
; CHECK-LABEL: sqadd_s_highimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    sqadd z0.s, z0.s, #8192 // =0x2000
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 4 x i32> undef, i32 8192, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.x.nxv4i32(<vscale x 4 x i32> %a,
                                                                   <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 2 x i64> @sqadd_d_lowimm(<vscale x 2 x i64> %a) {
; CHECK-LABEL: sqadd_d_lowimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    sqadd z0.d, z0.d, #255 // =0xff
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 2 x i64> undef, i64 255, i32 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.x.nxv2i64(<vscale x 2 x i64> %a,
                                                                   <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

define <vscale x 2 x i64> @sqadd_d_highimm(<vscale x 2 x i64> %a) {
; CHECK-LABEL: sqadd_d_highimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    sqadd z0.d, z0.d, #65280 // =0xff00
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 2 x i64> undef, i64 65280, i32 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.x.nxv2i64(<vscale x 2 x i64> %a,
                                                                   <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

; SQSUB

define <vscale x 16 x i8> @sqsub_b_lowimm(<vscale x 16 x i8> %a) {
; CHECK-LABEL: sqsub_b_lowimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    sqsub z0.b, z0.b, #27 // =0x1b
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 16 x i8> undef, i8 27, i32 0
  %splat = shufflevector <vscale x 16 x i8> %elt, <vscale x 16 x i8> undef, <vscale x 16 x i32> zeroinitializer
  %out = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsub.x.nxv16i8(<vscale x 16 x i8> %a,
                                                                   <vscale x 16 x i8> %splat)
  ret <vscale x 16 x i8> %out
}

define <vscale x 8 x i16> @sqsub_h_lowimm(<vscale x 8 x i16> %a) {
; CHECK-LABEL: sqsub_h_lowimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    sqsub z0.h, z0.h, #43 // =0x2b
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 8 x i16> undef, i16 43, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsub.x.nxv8i16(<vscale x 8 x i16> %a,
                                                                   <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 8 x i16> @sqsub_h_highimm(<vscale x 8 x i16> %a) {
; CHECK-LABEL: sqsub_h_highimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    sqsub z0.h, z0.h, #2048 // =0x800
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 8 x i16> undef, i16 2048, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsub.x.nxv8i16(<vscale x 8 x i16> %a,
                                                                   <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 4 x i32> @sqsub_s_lowimm(<vscale x 4 x i32> %a) {
; CHECK-LABEL: sqsub_s_lowimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    sqsub z0.s, z0.s, #1 // =0x1
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 4 x i32> undef, i32 1, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsub.x.nxv4i32(<vscale x 4 x i32> %a,
                                                                   <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 4 x i32> @sqsub_s_highimm(<vscale x 4 x i32> %a) {
; CHECK-LABEL: sqsub_s_highimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    sqsub z0.s, z0.s, #8192 // =0x2000
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 4 x i32> undef, i32 8192, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsub.x.nxv4i32(<vscale x 4 x i32> %a,
                                                                   <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 2 x i64> @sqsub_d_lowimm(<vscale x 2 x i64> %a) {
; CHECK-LABEL: sqsub_d_lowimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    sqsub z0.d, z0.d, #255 // =0xff
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 2 x i64> undef, i64 255, i32 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsub.x.nxv2i64(<vscale x 2 x i64> %a,
                                                                   <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

define <vscale x 2 x i64> @sqsub_d_highimm(<vscale x 2 x i64> %a) {
; CHECK-LABEL: sqsub_d_highimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    sqsub z0.d, z0.d, #65280 // =0xff00
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 2 x i64> undef, i64 65280, i32 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsub.x.nxv2i64(<vscale x 2 x i64> %a,
                                                                   <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

; UQADD

define <vscale x 16 x i8> @uqadd_b_lowimm(<vscale x 16 x i8> %a) {
; CHECK-LABEL: uqadd_b_lowimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    uqadd z0.b, z0.b, #27 // =0x1b
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 16 x i8> undef, i8 27, i32 0
  %splat = shufflevector <vscale x 16 x i8> %elt, <vscale x 16 x i8> undef, <vscale x 16 x i32> zeroinitializer
  %out = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.x.nxv16i8(<vscale x 16 x i8> %a,
                                                                   <vscale x 16 x i8> %splat)
  ret <vscale x 16 x i8> %out
}

define <vscale x 8 x i16> @uqadd_h_lowimm(<vscale x 8 x i16> %a) {
; CHECK-LABEL: uqadd_h_lowimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    uqadd z0.h, z0.h, #43 // =0x2b
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 8 x i16> undef, i16 43, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.x.nxv8i16(<vscale x 8 x i16> %a,
                                                                   <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 8 x i16> @uqadd_h_highimm(<vscale x 8 x i16> %a) {
; CHECK-LABEL: uqadd_h_highimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    uqadd z0.h, z0.h, #2048 // =0x800
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 8 x i16> undef, i16 2048, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.x.nxv8i16(<vscale x 8 x i16> %a,
                                                                   <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 4 x i32> @uqadd_s_lowimm(<vscale x 4 x i32> %a) {
; CHECK-LABEL: uqadd_s_lowimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    uqadd z0.s, z0.s, #1 // =0x1
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 4 x i32> undef, i32 1, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.x.nxv4i32(<vscale x 4 x i32> %a,
                                                                   <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 4 x i32> @uqadd_s_highimm(<vscale x 4 x i32> %a) {
; CHECK-LABEL: uqadd_s_highimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    uqadd z0.s, z0.s, #8192 // =0x2000
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 4 x i32> undef, i32 8192, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.x.nxv4i32(<vscale x 4 x i32> %a,
                                                                   <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 2 x i64> @uqadd_d_lowimm(<vscale x 2 x i64> %a) {
; CHECK-LABEL: uqadd_d_lowimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    uqadd z0.d, z0.d, #255 // =0xff
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 2 x i64> undef, i64 255, i32 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.x.nxv2i64(<vscale x 2 x i64> %a,
                                                                   <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

define <vscale x 2 x i64> @uqadd_d_highimm(<vscale x 2 x i64> %a) {
; CHECK-LABEL: uqadd_d_highimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    uqadd z0.d, z0.d, #65280 // =0xff00
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 2 x i64> undef, i64 65280, i32 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.x.nxv2i64(<vscale x 2 x i64> %a,
                                                                   <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

; UQSUB

define <vscale x 16 x i8> @uqsub_b_lowimm(<vscale x 16 x i8> %a) {
; CHECK-LABEL: uqsub_b_lowimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    uqsub z0.b, z0.b, #27 // =0x1b
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 16 x i8> undef, i8 27, i32 0
  %splat = shufflevector <vscale x 16 x i8> %elt, <vscale x 16 x i8> undef, <vscale x 16 x i32> zeroinitializer
  %out = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsub.x.nxv16i8(<vscale x 16 x i8> %a,
                                                                   <vscale x 16 x i8> %splat)
  ret <vscale x 16 x i8> %out
}

define <vscale x 8 x i16> @uqsub_h_lowimm(<vscale x 8 x i16> %a) {
; CHECK-LABEL: uqsub_h_lowimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    uqsub z0.h, z0.h, #43 // =0x2b
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 8 x i16> undef, i16 43, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsub.x.nxv8i16(<vscale x 8 x i16> %a,
                                                                   <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 8 x i16> @uqsub_h_highimm(<vscale x 8 x i16> %a) {
; CHECK-LABEL: uqsub_h_highimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    uqsub z0.h, z0.h, #2048 // =0x800
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 8 x i16> undef, i16 2048, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsub.x.nxv8i16(<vscale x 8 x i16> %a,
                                                                   <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 4 x i32> @uqsub_s_lowimm(<vscale x 4 x i32> %a) {
; CHECK-LABEL: uqsub_s_lowimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    uqsub z0.s, z0.s, #1 // =0x1
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 4 x i32> undef, i32 1, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsub.x.nxv4i32(<vscale x 4 x i32> %a,
                                                                   <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 4 x i32> @uqsub_s_highimm(<vscale x 4 x i32> %a) {
; CHECK-LABEL: uqsub_s_highimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    uqsub z0.s, z0.s, #8192 // =0x2000
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 4 x i32> undef, i32 8192, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsub.x.nxv4i32(<vscale x 4 x i32> %a,
                                                                   <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 2 x i64> @uqsub_d_lowimm(<vscale x 2 x i64> %a) {
; CHECK-LABEL: uqsub_d_lowimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    uqsub z0.d, z0.d, #255 // =0xff
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 2 x i64> undef, i64 255, i32 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsub.x.nxv2i64(<vscale x 2 x i64> %a,
                                                                   <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

define <vscale x 2 x i64> @uqsub_d_highimm(<vscale x 2 x i64> %a) {
; CHECK-LABEL: uqsub_d_highimm:
; CHECK:       // %bb.0:
; CHECK-NEXT:    uqsub z0.d, z0.d, #65280 // =0xff00
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 2 x i64> undef, i64 65280, i32 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsub.x.nxv2i64(<vscale x 2 x i64> %a,
                                                                   <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

; ASR

define <vscale x 16 x i8> @asr_i8(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %a) {
; CHECK-LABEL: asr_i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    asr z0.b, p0/m, z0.b, #8
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 16 x i8> undef, i8 9, i32 0
  %splat = shufflevector <vscale x 16 x i8> %elt, <vscale x 16 x i8> undef, <vscale x 16 x i32> zeroinitializer
  %out = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.nxv16i8(<vscale x 16 x i1> %pg,
                                                               <vscale x 16 x i8> %a,
                                                               <vscale x 16 x i8> %splat)
  ret <vscale x 16 x i8> %out
}

define <vscale x 16 x i8> @asr_i8_all_active(<vscale x 16 x i8> %a) {
; CHECK-LABEL: asr_i8_all_active:
; CHECK:       // %bb.0:
; CHECK-NEXT:    asr z0.b, z0.b, #8
; CHECK-NEXT:    ret
  %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
  %elt = insertelement <vscale x 16 x i8> undef, i8 8, i32 0
  %splat = shufflevector <vscale x 16 x i8> %elt, <vscale x 16 x i8> undef, <vscale x 16 x i32> zeroinitializer
  %out = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.nxv16i8(<vscale x 16 x i1> %pg,
                                                               <vscale x 16 x i8> %a,
                                                               <vscale x 16 x i8> %splat)
  ret <vscale x 16 x i8> %out
}

; Ensure we don't match a right shift by zero to the immediate form.
define <vscale x 16 x i8> @asr_i8_too_small(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %a) {
; CHECK-LABEL: asr_i8_too_small:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z1.b, #0 // =0x0
; CHECK-NEXT:    asr z0.b, p0/m, z0.b, z1.b
; CHECK-NEXT:    ret
  %out = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.nxv16i8(<vscale x 16 x i1> %pg,
                                                               <vscale x 16 x i8> %a,
                                                               <vscale x 16 x i8> zeroinitializer)
  ret <vscale x 16 x i8> %out
}

define <vscale x 8 x i16> @asr_i16(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a) {
; CHECK-LABEL: asr_i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    asr z0.h, p0/m, z0.h, #16
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 8 x i16> undef, i16 17, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.nxv8i16(<vscale x 8 x i1> %pg,
                                                               <vscale x 8 x i16> %a,
                                                               <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 8 x i16> @asr_i16_all_active(<vscale x 8 x i16> %a) {
; CHECK-LABEL: asr_i16_all_active:
; CHECK:       // %bb.0:
; CHECK-NEXT:    asr z0.h, z0.h, #16
; CHECK-NEXT:    ret
  %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %elt = insertelement <vscale x 8 x i16> undef, i16 16, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.nxv8i16(<vscale x 8 x i1> %pg,
                                                               <vscale x 8 x i16> %a,
                                                               <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

; Ensure we don't match a right shift by zero to the immediate form.
define <vscale x 8 x i16> @asr_i16_too_small(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a) {
; CHECK-LABEL: asr_i16_too_small:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z1.h, #0 // =0x0
; CHECK-NEXT:    asr z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    ret
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.nxv8i16(<vscale x 8 x i1> %pg,
                                                               <vscale x 8 x i16> %a,
                                                               <vscale x 8 x i16> zeroinitializer)
  ret <vscale x 8 x i16> %out
}

define <vscale x 4 x i32> @asr_i32(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %a) {
; CHECK-LABEL: asr_i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    asr z0.s, p0/m, z0.s, #32
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 4 x i32> undef, i32 33, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.nxv4i32(<vscale x 4 x i1> %pg,
                                                               <vscale x 4 x i32> %a,
                                                               <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 4 x i32> @asr_i32_all_active(<vscale x 4 x i32> %a) {
; CHECK-LABEL: asr_i32_all_active:
; CHECK:       // %bb.0:
; CHECK-NEXT:    asr z0.s, z0.s, #32
; CHECK-NEXT:    ret
  %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
  %elt = insertelement <vscale x 4 x i32> undef, i32 32, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.nxv4i32(<vscale x 4 x i1> %pg,
                                                               <vscale x 4 x i32> %a,
                                                               <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

; Ensure we don't match a right shift by zero to the immediate form.
define <vscale x 4 x i32> @asr_i32_too_small(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %a) {
; CHECK-LABEL: asr_i32_too_small:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z1.s, #0 // =0x0
; CHECK-NEXT:    asr z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    ret
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.nxv4i32(<vscale x 4 x i1> %pg,
                                                               <vscale x 4 x i32> %a,
                                                               <vscale x 4 x i32> zeroinitializer)
  ret <vscale x 4 x i32> %out
}

define <vscale x 2 x i64> @asr_i64(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a) {
; CHECK-LABEL: asr_i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    asr z0.d, p0/m, z0.d, #64
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 2 x i64> undef, i64 65, i64 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.asr.nxv2i64(<vscale x 2 x i1> %pg,
                                                               <vscale x 2 x i64> %a,
                                                               <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

define <vscale x 2 x i64> @asr_i64_all_active(<vscale x 2 x i64> %a) {
; CHECK-LABEL: asr_i64_all_active:
; CHECK:       // %bb.0:
; CHECK-NEXT:    asr z0.d, z0.d, #64
; CHECK-NEXT:    ret
  %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %elt = insertelement <vscale x 2 x i64> undef, i64 64, i64 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.asr.nxv2i64(<vscale x 2 x i1> %pg,
                                                               <vscale x 2 x i64> %a,
                                                               <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

; Ensure we don't match a right shift by zero to the immediate form.
define <vscale x 2 x i64> @asr_i64_too_small(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a) {
; CHECK-LABEL: asr_i64_too_small:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z1.d, #0 // =0x0
; CHECK-NEXT:    asr z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    ret
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.asr.nxv2i64(<vscale x 2 x i1> %pg,
                                                               <vscale x 2 x i64> %a,
                                                               <vscale x 2 x i64> zeroinitializer)
  ret <vscale x 2 x i64> %out
}

; LSL

define <vscale x 16 x i8> @lsl_i8(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %a) {
; CHECK-LABEL: lsl_i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    lsl z0.b, p0/m, z0.b, #7
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 16 x i8> undef, i8 7, i32 0
  %splat = shufflevector <vscale x 16 x i8> %elt, <vscale x 16 x i8> undef, <vscale x 16 x i32> zeroinitializer
  %out = call <vscale x 16 x i8> @llvm.aarch64.sve.lsl.nxv16i8(<vscale x 16 x i1> %pg,
                                                               <vscale x 16 x i8> %a,
                                                               <vscale x 16 x i8> %splat)
  ret <vscale x 16 x i8> %out
}

define <vscale x 16 x i8> @lsl_i8_all_active(<vscale x 16 x i8> %a) {
; CHECK-LABEL: lsl_i8_all_active:
; CHECK:       // %bb.0:
; CHECK-NEXT:    lsl z0.b, z0.b, #7
; CHECK-NEXT:    ret
  %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
  %elt = insertelement <vscale x 16 x i8> undef, i8 7, i32 0
  %splat = shufflevector <vscale x 16 x i8> %elt, <vscale x 16 x i8> undef, <vscale x 16 x i32> zeroinitializer
  %out = call <vscale x 16 x i8> @llvm.aarch64.sve.lsl.nxv16i8(<vscale x 16 x i1> %pg,
                                                               <vscale x 16 x i8> %a,
                                                               <vscale x 16 x i8> %splat)
  ret <vscale x 16 x i8> %out
}

; Ensure we don't match a left shift bigger than its bitwidth to the immediate form.
define <vscale x 16 x i8> @lsl_i8_too_big(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %a) {
; CHECK-LABEL: lsl_i8_too_big:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z1.b, #8 // =0x8
; CHECK-NEXT:    lsl z0.b, p0/m, z0.b, z1.b
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 16 x i8> undef, i8 8, i32 0
  %splat = shufflevector <vscale x 16 x i8> %elt, <vscale x 16 x i8> undef, <vscale x 16 x i32> zeroinitializer
  %out = call <vscale x 16 x i8> @llvm.aarch64.sve.lsl.nxv16i8(<vscale x 16 x i1> %pg,
                                                               <vscale x 16 x i8> %a,
                                                               <vscale x 16 x i8> %splat)
  ret <vscale x 16 x i8> %out
}

define <vscale x 16 x i8> @lsl_i8_zero(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %a) {
; CHECK-LABEL: lsl_i8_zero:
; CHECK:       // %bb.0:
; CHECK-NEXT:    lsl z0.b, p0/m, z0.b, #0
; CHECK-NEXT:    ret
  %out = call <vscale x 16 x i8> @llvm.aarch64.sve.lsl.nxv16i8(<vscale x 16 x i1> %pg,
                                                               <vscale x 16 x i8> %a,
                                                               <vscale x 16 x i8> zeroinitializer)
  ret <vscale x 16 x i8> %out
}

define <vscale x 8 x i16> @lsl_i16(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a) {
; CHECK-LABEL: lsl_i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    lsl z0.h, p0/m, z0.h, #15
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 8 x i16> undef, i16 15, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.lsl.nxv8i16(<vscale x 8 x i1> %pg,
                                                               <vscale x 8 x i16> %a,
                                                               <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 8 x i16> @lsl_i16_all_active(<vscale x 8 x i16> %a) {
; CHECK-LABEL: lsl_i16_all_active:
; CHECK:       // %bb.0:
; CHECK-NEXT:    lsl z0.h, z0.h, #15
; CHECK-NEXT:    ret
  %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %elt = insertelement <vscale x 8 x i16> undef, i16 15, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.lsl.nxv8i16(<vscale x 8 x i1> %pg,
                                                               <vscale x 8 x i16> %a,
                                                               <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

; Ensure we don't match a left shift bigger than its bitwidth to the immediate form.
define <vscale x 8 x i16> @lsl_i16_too_big(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a) {
; CHECK-LABEL: lsl_i16_too_big:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z1.h, #16 // =0x10
; CHECK-NEXT:    lsl z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 8 x i16> undef, i16 16, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.lsl.nxv8i16(<vscale x 8 x i1> %pg,
                                                               <vscale x 8 x i16> %a,
                                                               <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 8 x i16> @lsl_i16_zero(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a) {
; CHECK-LABEL: lsl_i16_zero:
; CHECK:       // %bb.0:
; CHECK-NEXT:    lsl z0.h, p0/m, z0.h, #0
; CHECK-NEXT:    ret
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.lsl.nxv8i16(<vscale x 8 x i1> %pg,
                                                               <vscale x 8 x i16> %a,
                                                               <vscale x 8 x i16> zeroinitializer)
  ret <vscale x 8 x i16> %out
}

define <vscale x 4 x i32> @lsl_i32(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %a) {
; CHECK-LABEL: lsl_i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    lsl z0.s, p0/m, z0.s, #31
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 4 x i32> undef, i32 31, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.lsl.nxv4i32(<vscale x 4 x i1> %pg,
                                                               <vscale x 4 x i32> %a,
                                                               <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 4 x i32> @lsl_i32_all_active(<vscale x 4 x i32> %a) {
; CHECK-LABEL: lsl_i32_all_active:
; CHECK:       // %bb.0:
; CHECK-NEXT:    lsl z0.s, z0.s, #31
; CHECK-NEXT:    ret
  %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
  %elt = insertelement <vscale x 4 x i32> undef, i32 31, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.lsl.nxv4i32(<vscale x 4 x i1> %pg,
                                                               <vscale x 4 x i32> %a,
                                                               <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

; Ensure we don't match a left shift bigger than its bitwidth to the immediate form.
define <vscale x 4 x i32> @lsl_i32_too_big(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %a) {
; CHECK-LABEL: lsl_i32_too_big:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z1.s, #32 // =0x20
; CHECK-NEXT:    lsl z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 4 x i32> undef, i32 32, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.lsl.nxv4i32(<vscale x 4 x i1> %pg,
                                                               <vscale x 4 x i32> %a,
                                                               <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 4 x i32> @lsl_i32_zero(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %a) {
; CHECK-LABEL: lsl_i32_zero:
; CHECK:       // %bb.0:
; CHECK-NEXT:    lsl z0.s, p0/m, z0.s, #0
; CHECK-NEXT:    ret
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.lsl.nxv4i32(<vscale x 4 x i1> %pg,
                                                               <vscale x 4 x i32> %a,
                                                               <vscale x 4 x i32> zeroinitializer)
  ret <vscale x 4 x i32> %out
}

define <vscale x 2 x i64> @lsl_i64(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a) {
; CHECK-LABEL: lsl_i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    lsl z0.d, p0/m, z0.d, #63
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 2 x i64> undef, i64 63, i64 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.lsl.nxv2i64(<vscale x 2 x i1> %pg,
                                                               <vscale x 2 x i64> %a,
                                                               <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

define <vscale x 2 x i64> @lsl_i64_all_active(<vscale x 2 x i64> %a) {
; CHECK-LABEL: lsl_i64_all_active:
; CHECK:       // %bb.0:
; CHECK-NEXT:    lsl z0.d, z0.d, #63
; CHECK-NEXT:    ret
  %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %elt = insertelement <vscale x 2 x i64> undef, i64 63, i64 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.lsl.nxv2i64(<vscale x 2 x i1> %pg,
                                                               <vscale x 2 x i64> %a,
                                                               <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

; Ensure we don't match a left shift bigger than its bitwidth to the immediate form.
define <vscale x 2 x i64> @lsl_i64_too_big(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a) {
; CHECK-LABEL: lsl_i64_too_big:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z1.d, #64 // =0x40
; CHECK-NEXT:    lsl z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 2 x i64> undef, i64 64, i64 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.lsl.nxv2i64(<vscale x 2 x i1> %pg,
                                                               <vscale x 2 x i64> %a,
                                                               <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

define <vscale x 2 x i64> @lsl_i64_zero(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a) {
; CHECK-LABEL: lsl_i64_zero:
; CHECK:       // %bb.0:
; CHECK-NEXT:    lsl z0.d, p0/m, z0.d, #0
; CHECK-NEXT:    ret
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.lsl.nxv2i64(<vscale x 2 x i1> %pg,
                                                               <vscale x 2 x i64> %a,
                                                               <vscale x 2 x i64> zeroinitializer)
  ret <vscale x 2 x i64> %out
}

; LSR

define <vscale x 16 x i8> @lsr_i8(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %a) {
; CHECK-LABEL: lsr_i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    lsr z0.b, p0/m, z0.b, #8
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 16 x i8> undef, i8 9, i32 0
  %splat = shufflevector <vscale x 16 x i8> %elt, <vscale x 16 x i8> undef, <vscale x 16 x i32> zeroinitializer
  %out = call <vscale x 16 x i8> @llvm.aarch64.sve.lsr.nxv16i8(<vscale x 16 x i1> %pg,
                                                               <vscale x 16 x i8> %a,
                                                               <vscale x 16 x i8> %splat)
  ret <vscale x 16 x i8> %out
}

define <vscale x 16 x i8> @lsr_i8_all_active(<vscale x 16 x i8> %a) {
; CHECK-LABEL: lsr_i8_all_active:
; CHECK:       // %bb.0:
; CHECK-NEXT:    lsr z0.b, z0.b, #8
; CHECK-NEXT:    ret
  %pg = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
  %elt = insertelement <vscale x 16 x i8> undef, i8 8, i32 0
  %splat = shufflevector <vscale x 16 x i8> %elt, <vscale x 16 x i8> undef, <vscale x 16 x i32> zeroinitializer
  %out = call <vscale x 16 x i8> @llvm.aarch64.sve.lsr.nxv16i8(<vscale x 16 x i1> %pg,
                                                               <vscale x 16 x i8> %a,
                                                               <vscale x 16 x i8> %splat)
  ret <vscale x 16 x i8> %out
}

; Ensure we don't match a right shift by zero to the immediate form.
define <vscale x 16 x i8> @lsr_i8_too_small(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %a) {
; CHECK-LABEL: lsr_i8_too_small:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z1.b, #0 // =0x0
; CHECK-NEXT:    lsr z0.b, p0/m, z0.b, z1.b
; CHECK-NEXT:    ret
  %out = call <vscale x 16 x i8> @llvm.aarch64.sve.lsr.nxv16i8(<vscale x 16 x i1> %pg,
                                                               <vscale x 16 x i8> %a,
                                                               <vscale x 16 x i8> zeroinitializer)
  ret <vscale x 16 x i8> %out
}

define <vscale x 8 x i16> @lsr_i16(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a) {
; CHECK-LABEL: lsr_i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    lsr z0.h, p0/m, z0.h, #16
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 8 x i16> undef, i16 17, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.lsr.nxv8i16(<vscale x 8 x i1> %pg,
                                                               <vscale x 8 x i16> %a,
                                                               <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

define <vscale x 8 x i16> @lsr_i16_all_active(<vscale x 8 x i16> %a) {
; CHECK-LABEL: lsr_i16_all_active:
; CHECK:       // %bb.0:
; CHECK-NEXT:    lsr z0.h, z0.h, #16
; CHECK-NEXT:    ret
  %pg = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %elt = insertelement <vscale x 8 x i16> undef, i16 16, i32 0
  %splat = shufflevector <vscale x 8 x i16> %elt, <vscale x 8 x i16> undef, <vscale x 8 x i32> zeroinitializer
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.lsr.nxv8i16(<vscale x 8 x i1> %pg,
                                                               <vscale x 8 x i16> %a,
                                                               <vscale x 8 x i16> %splat)
  ret <vscale x 8 x i16> %out
}

; Ensure we don't match a right shift by zero to the immediate form.
define <vscale x 8 x i16> @lsr_i16_too_small(<vscale x 8 x i1> %pg, <vscale x 8 x i16> %a) {
; CHECK-LABEL: lsr_i16_too_small:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z1.h, #0 // =0x0
; CHECK-NEXT:    lsr z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    ret
  %out = call <vscale x 8 x i16> @llvm.aarch64.sve.lsr.nxv8i16(<vscale x 8 x i1> %pg,
                                                               <vscale x 8 x i16> %a,
                                                               <vscale x 8 x i16> zeroinitializer)
  ret <vscale x 8 x i16> %out
}

define <vscale x 4 x i32> @lsr_i32(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %a) {
; CHECK-LABEL: lsr_i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    lsr z0.s, p0/m, z0.s, #32
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 4 x i32> undef, i32 33, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.nxv4i32(<vscale x 4 x i1> %pg,
                                                               <vscale x 4 x i32> %a,
                                                               <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

define <vscale x 4 x i32> @lsr_i32_all_active(<vscale x 4 x i32> %a) {
; CHECK-LABEL: lsr_i32_all_active:
; CHECK:       // %bb.0:
; CHECK-NEXT:    lsr z0.s, z0.s, #32
; CHECK-NEXT:    ret
  %pg = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
  %elt = insertelement <vscale x 4 x i32> undef, i32 32, i32 0
  %splat = shufflevector <vscale x 4 x i32> %elt, <vscale x 4 x i32> undef, <vscale x 4 x i32> zeroinitializer
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.nxv4i32(<vscale x 4 x i1> %pg,
                                                               <vscale x 4 x i32> %a,
                                                               <vscale x 4 x i32> %splat)
  ret <vscale x 4 x i32> %out
}

; Ensure we don't match a right shift by zero to the immediate form.
define <vscale x 4 x i32> @lsr_i32_too_small(<vscale x 4 x i1> %pg, <vscale x 4 x i32> %a) {
; CHECK-LABEL: lsr_i32_too_small:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z1.s, #0 // =0x0
; CHECK-NEXT:    lsr z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    ret
  %out = call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.nxv4i32(<vscale x 4 x i1> %pg,
                                                               <vscale x 4 x i32> %a,
                                                               <vscale x 4 x i32> zeroinitializer)
  ret <vscale x 4 x i32> %out
}

define <vscale x 2 x i64> @lsr_i64(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a) {
; CHECK-LABEL: lsr_i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    lsr z0.d, p0/m, z0.d, #64
; CHECK-NEXT:    ret
  %elt = insertelement <vscale x 2 x i64> undef, i64 65, i64 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.lsr.nxv2i64(<vscale x 2 x i1> %pg,
                                                               <vscale x 2 x i64> %a,
                                                               <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

define <vscale x 2 x i64> @lsr_i64_all_active(<vscale x 2 x i64> %a) {
; CHECK-LABEL: lsr_i64_all_active:
; CHECK:       // %bb.0:
; CHECK-NEXT:    lsr z0.d, z0.d, #64
; CHECK-NEXT:    ret
  %pg = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %elt = insertelement <vscale x 2 x i64> undef, i64 64, i64 0
  %splat = shufflevector <vscale x 2 x i64> %elt, <vscale x 2 x i64> undef, <vscale x 2 x i32> zeroinitializer
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.lsr.nxv2i64(<vscale x 2 x i1> %pg,
                                                               <vscale x 2 x i64> %a,
                                                               <vscale x 2 x i64> %splat)
  ret <vscale x 2 x i64> %out
}

; Ensure we don't match a right shift by zero to the immediate form.
define <vscale x 2 x i64> @lsr_i64_too_small(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a) {
; CHECK-LABEL: lsr_i64_too_small:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z1.d, #0 // =0x0
; CHECK-NEXT:    lsr z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    ret
  %out = call <vscale x 2 x i64> @llvm.aarch64.sve.lsr.nxv2i64(<vscale x 2 x i1> %pg,
                                                               <vscale x 2 x i64> %a,
                                                               <vscale x 2 x i64> zeroinitializer)
  ret <vscale x 2 x i64> %out
}

; As lsr_i32 but where pg is i8 based and thus compatible for i32.
define <vscale x 4 x i32> @lsr_i32_ptrue_all_b(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: lsr_i32_ptrue_all_b:
; CHECK:       // %bb.0:
; CHECK-NEXT:    lsr z0.s, z0.s, #1
; CHECK-NEXT:    ret
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                    <vscale x 4 x i32> %a,
                                                                    <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

; As lsr_i32 but where pg is i16 based and thus compatible for i32.
define <vscale x 4 x i32> @lsr_i32_ptrue_all_h(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: lsr_i32_ptrue_all_h:
; CHECK:       // %bb.0:
; CHECK-NEXT:    lsr z0.s, z0.s, #1
; CHECK-NEXT:    ret
  %pg.h = tail call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> %pg.h)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                    <vscale x 4 x i32> %a,
                                                                    <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

; As lsr_i32 but where pg is i64 based, which is not compatibile for i32 and
; thus inactive lanes are important and the immediate form cannot be used.
define <vscale x 4 x i32> @lsr_i32_ptrue_all_d(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: lsr_i32_ptrue_all_d:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d
; CHECK-NEXT:    lsr z0.s, p0/m, z0.s, #1
; CHECK-NEXT:    ret
  %pg.d = tail call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> %pg.d)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                    <vscale x 4 x i32> %a,
                                                                    <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

;
; MUL
;

; As mul_i32 but where pg is i8 based and thus compatible for i32.
define <vscale x 4 x i32> @mul_i32_ptrue_all_b(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: mul_i32_ptrue_all_b:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mul z0.s, z0.s, #1
; CHECK-NEXT:    ret
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                     <vscale x 4 x i32> %a,
                                                                     <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

; As mul_i32 but where pg is i16 based and thus compatible for i32.
define <vscale x 4 x i32> @mul_i32_ptrue_all_h(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: mul_i32_ptrue_all_h:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mul z0.s, z0.s, #1
; CHECK-NEXT:    ret
  %pg.h = tail call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> %pg.h)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                     <vscale x 4 x i32> %a,
                                                                     <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

; As mul_i32 but where pg is i64 based, which is not compatibile for i32 and
; thus inactive lanes are important and the immediate form cannot be used.
define <vscale x 4 x i32> @mul_i32_ptrue_all_d(<vscale x 4 x i32> %a) #0 {
; CHECK-LABEL: mul_i32_ptrue_all_d:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d
; CHECK-NEXT:    mov z1.s, #1 // =0x1
; CHECK-NEXT:    mul z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    ret
  %pg.d = tail call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
  %pg.b = tail call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> %pg.d)
  %pg.s = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> %pg.b)
  %b = tail call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 1)
  %out = tail call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> %pg.s,
                                                                     <vscale x 4 x i32> %a,
                                                                     <vscale x 4 x i32> %b)
  ret <vscale x 4 x i32> %out
}

declare <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1>, <vscale x 16 x i8>, <vscale x 16 x i8>)
declare <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1>, <vscale x 8 x i16>, <vscale x 8 x i16>)
declare <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1>, <vscale x 4 x i32>, <vscale x 4 x i32>)
declare <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1>, <vscale x 2 x i64>, <vscale x 2 x i64>)

declare <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1>, <vscale x 16 x i8>, <vscale x 16 x i8>)
declare <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1>, <vscale x 8 x i16>, <vscale x 8 x i16>)
declare <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1>, <vscale x 4 x i32>, <vscale x 4 x i32>)
declare <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1>, <vscale x 2 x i64>, <vscale x 2 x i64>)

declare <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1>, <vscale x 16 x i8>, <vscale x 16 x i8>)
declare <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1>, <vscale x 8 x i16>, <vscale x 8 x i16>)
declare <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1>, <vscale x 4 x i32>, <vscale x 4 x i32>)
declare <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1>, <vscale x 2 x i64>, <vscale x 2 x i64>)

declare <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.x.nxv16i8(<vscale x 16 x i8>, <vscale x 16 x i8>)
declare <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.x.nxv8i16(<vscale x 8 x i16>, <vscale x 8 x i16>)
declare <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.x.nxv4i32(<vscale x 4 x i32>, <vscale x 4 x i32>)
declare <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.x.nxv2i64(<vscale x 2 x i64>, <vscale x 2 x i64>)

declare <vscale x 16 x i8> @llvm.aarch64.sve.sqsub.x.nxv16i8(<vscale x 16 x i8>, <vscale x 16 x i8>)
declare <vscale x 8 x i16> @llvm.aarch64.sve.sqsub.x.nxv8i16(<vscale x 8 x i16>, <vscale x 8 x i16>)
declare <vscale x 4 x i32> @llvm.aarch64.sve.sqsub.x.nxv4i32(<vscale x 4 x i32>, <vscale x 4 x i32>)
declare <vscale x 2 x i64> @llvm.aarch64.sve.sqsub.x.nxv2i64(<vscale x 2 x i64>, <vscale x 2 x i64>)

declare <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.x.nxv16i8(<vscale x 16 x i8>, <vscale x 16 x i8>)
declare <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.x.nxv8i16(<vscale x 8 x i16>, <vscale x 8 x i16>)
declare <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.x.nxv4i32(<vscale x 4 x i32>, <vscale x 4 x i32>)
declare <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.x.nxv2i64(<vscale x 2 x i64>, <vscale x 2 x i64>)

declare <vscale x 16 x i8> @llvm.aarch64.sve.uqsub.x.nxv16i8(<vscale x 16 x i8>, <vscale x 16 x i8>)
declare <vscale x 8 x i16> @llvm.aarch64.sve.uqsub.x.nxv8i16(<vscale x 8 x i16>, <vscale x 8 x i16>)
declare <vscale x 4 x i32> @llvm.aarch64.sve.uqsub.x.nxv4i32(<vscale x 4 x i32>, <vscale x 4 x i32>)
declare <vscale x 2 x i64> @llvm.aarch64.sve.uqsub.x.nxv2i64(<vscale x 2 x i64>, <vscale x 2 x i64>)

declare <vscale x 16 x i8> @llvm.aarch64.sve.smax.nxv16i8(<vscale x 16 x i1>, <vscale x 16 x i8>, <vscale x 16 x i8>)
declare <vscale x 8 x i16> @llvm.aarch64.sve.smax.nxv8i16(<vscale x 8 x i1>, <vscale x 8 x i16>, <vscale x 8 x i16>)
declare <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1>, <vscale x 4 x i32>, <vscale x 4 x i32>)
declare <vscale x 2 x i64> @llvm.aarch64.sve.smax.nxv2i64(<vscale x 2 x i1>, <vscale x 2 x i64>, <vscale x 2 x i64>)

declare <vscale x 16 x i8> @llvm.aarch64.sve.smin.nxv16i8(<vscale x 16 x i1>, <vscale x 16 x i8>, <vscale x 16 x i8>)
declare <vscale x 8 x i16> @llvm.aarch64.sve.smin.nxv8i16(<vscale x 8 x i1>, <vscale x 8 x i16>, <vscale x 8 x i16>)
declare <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1>, <vscale x 4 x i32>, <vscale x 4 x i32>)
declare <vscale x 2 x i64> @llvm.aarch64.sve.smin.nxv2i64(<vscale x 2 x i1>, <vscale x 2 x i64>, <vscale x 2 x i64>)

declare <vscale x 16 x i8> @llvm.aarch64.sve.umax.nxv16i8(<vscale x 16 x i1>, <vscale x 16 x i8>, <vscale x 16 x i8>)
declare <vscale x 8 x i16> @llvm.aarch64.sve.umax.nxv8i16(<vscale x 8 x i1>, <vscale x 8 x i16>, <vscale x 8 x i16>)
declare <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1>, <vscale x 4 x i32>, <vscale x 4 x i32>)
declare <vscale x 2 x i64> @llvm.aarch64.sve.umax.nxv2i64(<vscale x 2 x i1>, <vscale x 2 x i64>, <vscale x 2 x i64>)

declare <vscale x 16 x i8> @llvm.aarch64.sve.umin.nxv16i8(<vscale x 16 x i1>, <vscale x 16 x i8>, <vscale x 16 x i8>)
declare <vscale x 8 x i16> @llvm.aarch64.sve.umin.nxv8i16(<vscale x 8 x i1>, <vscale x 8 x i16>, <vscale x 8 x i16>)
declare <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1>, <vscale x 4 x i32>, <vscale x 4 x i32>)
declare <vscale x 2 x i64> @llvm.aarch64.sve.umin.nxv2i64(<vscale x 2 x i1>, <vscale x 2 x i64>, <vscale x 2 x i64>)

declare <vscale x 16 x i8> @llvm.aarch64.sve.asr.nxv16i8(<vscale x 16 x i1>, <vscale x 16 x i8>, <vscale x 16 x i8>)
declare <vscale x 8 x i16> @llvm.aarch64.sve.asr.nxv8i16(<vscale x 8 x i1>, <vscale x 8 x i16>, <vscale x 8 x i16>)
declare <vscale x 4 x i32> @llvm.aarch64.sve.asr.nxv4i32(<vscale x 4 x i1>, <vscale x 4 x i32>, <vscale x 4 x i32>)
declare <vscale x 2 x i64> @llvm.aarch64.sve.asr.nxv2i64(<vscale x 2 x i1>, <vscale x 2 x i64>, <vscale x 2 x i64>)

declare <vscale x 16 x i8> @llvm.aarch64.sve.lsl.nxv16i8(<vscale x 16 x i1>, <vscale x 16 x i8>, <vscale x 16 x i8>)
declare <vscale x 8 x i16> @llvm.aarch64.sve.lsl.nxv8i16(<vscale x 8 x i1>, <vscale x 8 x i16>, <vscale x 8 x i16>)
declare <vscale x 4 x i32> @llvm.aarch64.sve.lsl.nxv4i32(<vscale x 4 x i1>, <vscale x 4 x i32>, <vscale x 4 x i32>)
declare <vscale x 2 x i64> @llvm.aarch64.sve.lsl.nxv2i64(<vscale x 2 x i1>, <vscale x 2 x i64>, <vscale x 2 x i64>)

declare <vscale x 16 x i8> @llvm.aarch64.sve.lsr.nxv16i8(<vscale x 16 x i1>, <vscale x 16 x i8>, <vscale x 16 x i8>)
declare <vscale x 8 x i16> @llvm.aarch64.sve.lsr.nxv8i16(<vscale x 8 x i1>, <vscale x 8 x i16>, <vscale x 8 x i16>)
declare <vscale x 4 x i32> @llvm.aarch64.sve.lsr.nxv4i32(<vscale x 4 x i1>, <vscale x 4 x i32>, <vscale x 4 x i32>)
declare <vscale x 2 x i64> @llvm.aarch64.sve.lsr.nxv2i64(<vscale x 2 x i1>, <vscale x 2 x i64>, <vscale x 2 x i64>)

declare <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1>, <vscale x 16 x i8>, <vscale x 16 x i8>)
declare <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1>, <vscale x 8 x i16>, <vscale x 8 x i16>)
declare <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1>, <vscale x 4 x i32>, <vscale x 4 x i32>)
declare <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1>, <vscale x 2 x i64>, <vscale x 2 x i64>)

declare <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1>)
declare <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1>)
declare <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1>)

declare <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 16 x i1>)
declare <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1>)
declare <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1>)

declare <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32)

declare <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 %pattern)
declare <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 %pattern)
declare <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 %pattern)
declare <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 %pattern)