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

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

;
; FCVTZU H -> H
;

; Don't use SVE for 64-bit vectors.
define <4 x i16> @fcvtzu_v4f16_v4i16(<4 x half> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v4f16_v4i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtzu v0.4h, v0.4h
; CHECK-NEXT:    ret
  %res = fptoui <4 x half> %op1 to <4 x i16>
  ret <4 x i16> %res
}

; Don't use SVE for 128-bit vectors.
define void @fcvtzu_v8f16_v8i16(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v8f16_v8i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr q0, [x0]
; CHECK-NEXT:    fcvtzu v0.8h, v0.8h
; CHECK-NEXT:    str q0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <8 x half>, ptr %a
  %res = fptoui <8 x half> %op1 to <8 x i16>
  store <8 x i16> %res, ptr %b
  ret void
}

define void @fcvtzu_v16f16_v16i16(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v16f16_v16i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl16
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <16 x half>, ptr %a
  %res = fptoui <16 x half> %op1 to <16 x i16>
  store <16 x i16> %res, ptr %b
  ret void
}

define void @fcvtzu_v32f16_v32i16(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: fcvtzu_v32f16_v32i16:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.h, vl16
; VBITS_GE_256-NEXT:    mov x8, #16 // =0x10
; VBITS_GE_256-NEXT:    ld1h { z0.h }, p0/z, [x0, x8, lsl #1]
; VBITS_GE_256-NEXT:    ld1h { z1.h }, p0/z, [x0]
; VBITS_GE_256-NEXT:    fcvtzu z0.h, p0/m, z0.h
; VBITS_GE_256-NEXT:    fcvtzu z1.h, p0/m, z1.h
; VBITS_GE_256-NEXT:    st1h { z0.h }, p0, [x1, x8, lsl #1]
; VBITS_GE_256-NEXT:    st1h { z1.h }, p0, [x1]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: fcvtzu_v32f16_v32i16:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.h, vl32
; VBITS_GE_512-NEXT:    ld1h { z0.h }, p0/z, [x0]
; VBITS_GE_512-NEXT:    fcvtzu z0.h, p0/m, z0.h
; VBITS_GE_512-NEXT:    st1h { z0.h }, p0, [x1]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <32 x half>, ptr %a
  %res = fptoui <32 x half> %op1 to <32 x i16>
  store <32 x i16> %res, ptr %b
  ret void
}

define void @fcvtzu_v64f16_v64i16(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: fcvtzu_v64f16_v64i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl64
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <64 x half>, ptr %a
  %res = fptoui <64 x half> %op1 to <64 x i16>
  store <64 x i16> %res, ptr %b
  ret void
}

define void @fcvtzu_v128f16_v128i16(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: fcvtzu_v128f16_v128i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl128
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <128 x half>, ptr %a
  %res = fptoui <128 x half> %op1 to <128 x i16>
  store <128 x i16> %res, ptr %b
  ret void
}

;
; FCVTZU H -> S
;

; Don't use SVE for 64-bit vectors.
define <2 x i32> @fcvtzu_v2f16_v2i32(<2 x half> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v2f16_v2i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtl v0.4s, v0.4h
; CHECK-NEXT:    fcvtzu v0.4s, v0.4s
; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $q0
; CHECK-NEXT:    ret
  %res = fptoui <2 x half> %op1 to <2 x i32>
  ret <2 x i32> %res
}

; Don't use SVE for 128-bit vectors.
define <4 x i32> @fcvtzu_v4f16_v4i32(<4 x half> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v4f16_v4i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtl v0.4s, v0.4h
; CHECK-NEXT:    fcvtzu v0.4s, v0.4s
; CHECK-NEXT:    ret
  %res = fptoui <4 x half> %op1 to <4 x i32>
  ret <4 x i32> %res
}

define void @fcvtzu_v8f16_v8i32(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v8f16_v8i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr q0, [x0]
; CHECK-NEXT:    ptrue p0.s, vl8
; CHECK-NEXT:    uunpklo z0.s, z0.h
; CHECK-NEXT:    fcvtzu z0.s, p0/m, z0.h
; CHECK-NEXT:    st1w { z0.s }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <8 x half>, ptr %a
  %res = fptoui <8 x half> %op1 to <8 x i32>
  store <8 x i32> %res, ptr %b
  ret void
}

define void @fcvtzu_v16f16_v16i32(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: fcvtzu_v16f16_v16i32:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.h, vl16
; VBITS_GE_256-NEXT:    mov x8, #8 // =0x8
; VBITS_GE_256-NEXT:    ld1h { z0.h }, p0/z, [x0]
; VBITS_GE_256-NEXT:    ptrue p0.s, vl8
; VBITS_GE_256-NEXT:    uunpklo z1.s, z0.h
; VBITS_GE_256-NEXT:    ext z0.b, z0.b, z0.b, #16
; VBITS_GE_256-NEXT:    uunpklo z0.s, z0.h
; VBITS_GE_256-NEXT:    fcvtzu z1.s, p0/m, z1.h
; VBITS_GE_256-NEXT:    fcvtzu z0.s, p0/m, z0.h
; VBITS_GE_256-NEXT:    st1w { z1.s }, p0, [x1]
; VBITS_GE_256-NEXT:    st1w { z0.s }, p0, [x1, x8, lsl #2]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: fcvtzu_v16f16_v16i32:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.s, vl16
; VBITS_GE_512-NEXT:    ld1h { z0.s }, p0/z, [x0]
; VBITS_GE_512-NEXT:    fcvtzu z0.s, p0/m, z0.h
; VBITS_GE_512-NEXT:    st1w { z0.s }, p0, [x1]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <16 x half>, ptr %a
  %res = fptoui <16 x half> %op1 to <16 x i32>
  store <16 x i32> %res, ptr %b
  ret void
}

define void @fcvtzu_v32f16_v32i32(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: fcvtzu_v32f16_v32i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl32
; CHECK-NEXT:    ld1h { z0.s }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.s, p0/m, z0.h
; CHECK-NEXT:    st1w { z0.s }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <32 x half>, ptr %a
  %res = fptoui <32 x half> %op1 to <32 x i32>
  store <32 x i32> %res, ptr %b
  ret void
}

define void @fcvtzu_v64f16_v64i32(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: fcvtzu_v64f16_v64i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl64
; CHECK-NEXT:    ld1h { z0.s }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.s, p0/m, z0.h
; CHECK-NEXT:    st1w { z0.s }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <64 x half>, ptr %a
  %res = fptoui <64 x half> %op1 to <64 x i32>
  store <64 x i32> %res, ptr %b
  ret void
}

;
; FCVTZU H -> D
;

; Don't use SVE for 64-bit vectors.
define <1 x i64> @fcvtzu_v1f16_v1i64(<1 x half> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v1f16_v1i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtzu x8, h0
; CHECK-NEXT:    fmov d0, x8
; CHECK-NEXT:    ret
  %res = fptoui <1 x half> %op1 to <1 x i64>
  ret <1 x i64> %res
}

; v2f16 is not legal for NEON, so use SVE
define <2 x i64> @fcvtzu_v2f16_v2i64(<2 x half> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v2f16_v2i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    // kill: def $d0 killed $d0 def $z0
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    uunpklo z0.s, z0.h
; CHECK-NEXT:    uunpklo z0.d, z0.s
; CHECK-NEXT:    fcvtzu z0.d, p0/m, z0.h
; CHECK-NEXT:    // kill: def $q0 killed $q0 killed $z0
; CHECK-NEXT:    ret
  %res = fptoui <2 x half> %op1 to <2 x i64>
  ret <2 x i64> %res
}

define void @fcvtzu_v4f16_v4i64(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v4f16_v4i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr d0, [x0]
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    uunpklo z0.s, z0.h
; CHECK-NEXT:    uunpklo z0.d, z0.s
; CHECK-NEXT:    fcvtzu z0.d, p0/m, z0.h
; CHECK-NEXT:    st1d { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <4 x half>, ptr %a
  %res = fptoui <4 x half> %op1 to <4 x i64>
  store <4 x i64> %res, ptr %b
  ret void
}

define void @fcvtzu_v8f16_v8i64(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: fcvtzu_v8f16_v8i64:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ldr q0, [x0]
; VBITS_GE_256-NEXT:    ptrue p0.d, vl4
; VBITS_GE_256-NEXT:    mov x8, #4 // =0x4
; VBITS_GE_256-NEXT:    ext v1.16b, v0.16b, v0.16b, #8
; VBITS_GE_256-NEXT:    uunpklo z0.s, z0.h
; VBITS_GE_256-NEXT:    uunpklo z1.s, z1.h
; VBITS_GE_256-NEXT:    uunpklo z0.d, z0.s
; VBITS_GE_256-NEXT:    uunpklo z1.d, z1.s
; VBITS_GE_256-NEXT:    fcvtzu z0.d, p0/m, z0.h
; VBITS_GE_256-NEXT:    fcvtzu z1.d, p0/m, z1.h
; VBITS_GE_256-NEXT:    st1d { z0.d }, p0, [x1]
; VBITS_GE_256-NEXT:    st1d { z1.d }, p0, [x1, x8, lsl #3]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: fcvtzu_v8f16_v8i64:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ldr q0, [x0]
; VBITS_GE_512-NEXT:    ptrue p0.d, vl8
; VBITS_GE_512-NEXT:    uunpklo z0.s, z0.h
; VBITS_GE_512-NEXT:    uunpklo z0.d, z0.s
; VBITS_GE_512-NEXT:    fcvtzu z0.d, p0/m, z0.h
; VBITS_GE_512-NEXT:    st1d { z0.d }, p0, [x1]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <8 x half>, ptr %a
  %res = fptoui <8 x half> %op1 to <8 x i64>
  store <8 x i64> %res, ptr %b
  ret void
}

define void @fcvtzu_v16f16_v16i64(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: fcvtzu_v16f16_v16i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl16
; CHECK-NEXT:    ld1h { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.d, p0/m, z0.h
; CHECK-NEXT:    st1d { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <16 x half>, ptr %a
  %res = fptoui <16 x half> %op1 to <16 x i64>
  store <16 x i64> %res, ptr %b
  ret void
}

define void @fcvtzu_v32f16_v32i64(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: fcvtzu_v32f16_v32i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl32
; CHECK-NEXT:    ld1h { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.d, p0/m, z0.h
; CHECK-NEXT:    st1d { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <32 x half>, ptr %a
  %res = fptoui <32 x half> %op1 to <32 x i64>
  store <32 x i64> %res, ptr %b
  ret void
}

;
; FCVTZU S -> H
;

; Don't use SVE for 64-bit vectors.
define <2 x i16> @fcvtzu_v2f32_v2i16(<2 x float> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v2f32_v2i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtzs v0.2s, v0.2s
; CHECK-NEXT:    ret
  %res = fptoui <2 x float> %op1 to <2 x i16>
  ret <2 x i16> %res
}

; Don't use SVE for 128-bit vectors.
define <4 x i16> @fcvtzu_v4f32_v4i16(<4 x float> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v4f32_v4i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtzu v1.4s, v0.4s
; CHECK-NEXT:    mov w8, v1.s[1]
; CHECK-NEXT:    mov v0.16b, v1.16b
; CHECK-NEXT:    mov w9, v1.s[2]
; CHECK-NEXT:    mov v0.h[1], w8
; CHECK-NEXT:    mov w8, v1.s[3]
; CHECK-NEXT:    mov v0.h[2], w9
; CHECK-NEXT:    mov v0.h[3], w8
; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $q0
; CHECK-NEXT:    ret
  %res = fptoui <4 x float> %op1 to <4 x i16>
  ret <4 x i16> %res
}

define <8 x i16> @fcvtzu_v8f32_v8i16(ptr %a) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v8f32_v8i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl8
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.s, p0/m, z0.s
; CHECK-NEXT:    uzp1 z0.h, z0.h, z0.h
; CHECK-NEXT:    // kill: def $q0 killed $q0 killed $z0
; CHECK-NEXT:    ret
  %op1 = load <8 x float>, ptr %a
  %res = fptoui <8 x float> %op1 to <8 x i16>
  ret <8 x i16> %res
}

define void @fcvtzu_v16f32_v16i16(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: fcvtzu_v16f32_v16i16:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.s, vl8
; VBITS_GE_256-NEXT:    mov x8, #8 // =0x8
; VBITS_GE_256-NEXT:    ld1w { z0.s }, p0/z, [x0, x8, lsl #2]
; VBITS_GE_256-NEXT:    ld1w { z1.s }, p0/z, [x0]
; VBITS_GE_256-NEXT:    fcvtzu z0.s, p0/m, z0.s
; VBITS_GE_256-NEXT:    fcvtzu z1.s, p0/m, z1.s
; VBITS_GE_256-NEXT:    ptrue p0.h, vl8
; VBITS_GE_256-NEXT:    uzp1 z0.h, z0.h, z0.h
; VBITS_GE_256-NEXT:    uzp1 z1.h, z1.h, z1.h
; VBITS_GE_256-NEXT:    splice z1.h, p0, z1.h, z0.h
; VBITS_GE_256-NEXT:    ptrue p0.h, vl16
; VBITS_GE_256-NEXT:    st1h { z1.h }, p0, [x1]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: fcvtzu_v16f32_v16i16:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.s, vl16
; VBITS_GE_512-NEXT:    ld1w { z0.s }, p0/z, [x0]
; VBITS_GE_512-NEXT:    fcvtzu z0.s, p0/m, z0.s
; VBITS_GE_512-NEXT:    st1h { z0.s }, p0, [x1]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <16 x float>, ptr %a
  %res = fptoui <16 x float> %op1 to <16 x i16>
  store <16 x i16> %res, ptr %b
  ret void
}

define void @fcvtzu_v32f32_v32i16(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: fcvtzu_v32f32_v32i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl32
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.s, p0/m, z0.s
; CHECK-NEXT:    st1h { z0.s }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <32 x float>, ptr %a
  %res = fptoui <32 x float> %op1 to <32 x i16>
  store <32 x i16> %res, ptr %b
  ret void
}

define void @fcvtzu_v64f32_v64i16(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: fcvtzu_v64f32_v64i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl64
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.s, p0/m, z0.s
; CHECK-NEXT:    st1h { z0.s }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <64 x float>, ptr %a
  %res = fptoui <64 x float> %op1 to <64 x i16>
  store <64 x i16> %res, ptr %b
  ret void
}

;
; FCVTZU S -> S
;

; Don't use SVE for 64-bit vectors.
define <2 x i32> @fcvtzu_v2f32_v2i32(<2 x float> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v2f32_v2i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtzu v0.2s, v0.2s
; CHECK-NEXT:    ret
  %res = fptoui <2 x float> %op1 to <2 x i32>
  ret <2 x i32> %res
}

; Don't use SVE for 128-bit vectors.
define <4 x i32> @fcvtzu_v4f32_v4i32(<4 x float> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v4f32_v4i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtzu v0.4s, v0.4s
; CHECK-NEXT:    ret
  %res = fptoui <4 x float> %op1 to <4 x i32>
  ret <4 x i32> %res
}

define void @fcvtzu_v8f32_v8i32(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v8f32_v8i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl8
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <8 x float>, ptr %a
  %res = fptoui <8 x float> %op1 to <8 x i32>
  store <8 x i32> %res, ptr %b
  ret void
}

define void @fcvtzu_v16f32_v16i32(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: fcvtzu_v16f32_v16i32:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.s, vl8
; VBITS_GE_256-NEXT:    mov x8, #8 // =0x8
; VBITS_GE_256-NEXT:    ld1w { z0.s }, p0/z, [x0, x8, lsl #2]
; VBITS_GE_256-NEXT:    ld1w { z1.s }, p0/z, [x0]
; VBITS_GE_256-NEXT:    fcvtzu z0.s, p0/m, z0.s
; VBITS_GE_256-NEXT:    fcvtzu z1.s, p0/m, z1.s
; VBITS_GE_256-NEXT:    st1w { z0.s }, p0, [x1, x8, lsl #2]
; VBITS_GE_256-NEXT:    st1w { z1.s }, p0, [x1]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: fcvtzu_v16f32_v16i32:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.s, vl16
; VBITS_GE_512-NEXT:    ld1w { z0.s }, p0/z, [x0]
; VBITS_GE_512-NEXT:    fcvtzu z0.s, p0/m, z0.s
; VBITS_GE_512-NEXT:    st1w { z0.s }, p0, [x1]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <16 x float>, ptr %a
  %res = fptoui <16 x float> %op1 to <16 x i32>
  store <16 x i32> %res, ptr %b
  ret void
}

define void @fcvtzu_v32f32_v32i32(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: fcvtzu_v32f32_v32i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl32
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <32 x float>, ptr %a
  %res = fptoui <32 x float> %op1 to <32 x i32>
  store <32 x i32> %res, ptr %b
  ret void
}

define void @fcvtzu_v64f32_v64i32(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: fcvtzu_v64f32_v64i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl64
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <64 x float>, ptr %a
  %res = fptoui <64 x float> %op1 to <64 x i32>
  store <64 x i32> %res, ptr %b
  ret void
}

;
; FCVTZU S -> D
;

; Don't use SVE for 64-bit vectors.
define <1 x i64> @fcvtzu_v1f32_v1i64(<1 x float> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v1f32_v1i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtl v0.2d, v0.2s
; CHECK-NEXT:    fcvtzu v0.2d, v0.2d
; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $q0
; CHECK-NEXT:    ret
  %res = fptoui <1 x float> %op1 to <1 x i64>
  ret <1 x i64> %res
}

; Don't use SVE for 128-bit vectors.
define <2 x i64> @fcvtzu_v2f32_v2i64(<2 x float> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v2f32_v2i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtl v0.2d, v0.2s
; CHECK-NEXT:    fcvtzu v0.2d, v0.2d
; CHECK-NEXT:    ret
  %res = fptoui <2 x float> %op1 to <2 x i64>
  ret <2 x i64> %res
}

define void @fcvtzu_v4f32_v4i64(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v4f32_v4i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr q0, [x0]
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    uunpklo z0.d, z0.s
; CHECK-NEXT:    fcvtzu z0.d, p0/m, z0.s
; CHECK-NEXT:    st1d { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <4 x float>, ptr %a
  %res = fptoui <4 x float> %op1 to <4 x i64>
  store <4 x i64> %res, ptr %b
  ret void
}

define void @fcvtzu_v8f32_v8i64(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: fcvtzu_v8f32_v8i64:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.s, vl8
; VBITS_GE_256-NEXT:    mov x8, #4 // =0x4
; VBITS_GE_256-NEXT:    ld1w { z0.s }, p0/z, [x0]
; VBITS_GE_256-NEXT:    ptrue p0.d, vl4
; VBITS_GE_256-NEXT:    uunpklo z1.d, z0.s
; VBITS_GE_256-NEXT:    ext z0.b, z0.b, z0.b, #16
; VBITS_GE_256-NEXT:    uunpklo z0.d, z0.s
; VBITS_GE_256-NEXT:    fcvtzu z1.d, p0/m, z1.s
; VBITS_GE_256-NEXT:    fcvtzu z0.d, p0/m, z0.s
; VBITS_GE_256-NEXT:    st1d { z1.d }, p0, [x1]
; VBITS_GE_256-NEXT:    st1d { z0.d }, p0, [x1, x8, lsl #3]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: fcvtzu_v8f32_v8i64:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.d, vl8
; VBITS_GE_512-NEXT:    ld1w { z0.d }, p0/z, [x0]
; VBITS_GE_512-NEXT:    fcvtzu z0.d, p0/m, z0.s
; VBITS_GE_512-NEXT:    st1d { z0.d }, p0, [x1]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <8 x float>, ptr %a
  %res = fptoui <8 x float> %op1 to <8 x i64>
  store <8 x i64> %res, ptr %b
  ret void
}

define void @fcvtzu_v16f32_v16i64(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: fcvtzu_v16f32_v16i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl16
; CHECK-NEXT:    ld1w { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.d, p0/m, z0.s
; CHECK-NEXT:    st1d { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <16 x float>, ptr %a
  %res = fptoui <16 x float> %op1 to <16 x i64>
  store <16 x i64> %res, ptr %b
  ret void
}

define void @fcvtzu_v32f32_v32i64(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: fcvtzu_v32f32_v32i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl32
; CHECK-NEXT:    ld1w { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.d, p0/m, z0.s
; CHECK-NEXT:    st1d { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <32 x float>, ptr %a
  %res = fptoui <32 x float> %op1 to <32 x i64>
  store <32 x i64> %res, ptr %b
  ret void
}


;
; FCVTZU D -> H
;

; v1f64 is perfered to be widened to v4f64, so use SVE
define <1 x i16> @fcvtzu_v1f64_v1i16(<1 x double> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v1f64_v1i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    // kill: def $d0 killed $d0 def $z0
; CHECK-NEXT:    fcvtzu z0.d, p0/m, z0.d
; CHECK-NEXT:    uzp1 z0.s, z0.s, z0.s
; CHECK-NEXT:    uzp1 z0.h, z0.h, z0.h
; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $z0
; CHECK-NEXT:    ret
  %res = fptoui <1 x double> %op1 to <1 x i16>
  ret <1 x i16> %res
}

; Don't use SVE for 128-bit vectors.
define <2 x i16> @fcvtzu_v2f64_v2i16(<2 x double> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v2f64_v2i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtzs v0.2d, v0.2d
; CHECK-NEXT:    xtn v0.2s, v0.2d
; CHECK-NEXT:    ret
  %res = fptoui <2 x double> %op1 to <2 x i16>
  ret <2 x i16> %res
}

define <4 x i16> @fcvtzu_v4f64_v4i16(ptr %a) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v4f64_v4i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.d, p0/m, z0.d
; CHECK-NEXT:    uzp1 z0.s, z0.s, z0.s
; CHECK-NEXT:    uzp1 z0.h, z0.h, z0.h
; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $z0
; CHECK-NEXT:    ret
  %op1 = load <4 x double>, ptr %a
  %res = fptoui <4 x double> %op1 to <4 x i16>
  ret <4 x i16> %res
}

define <8 x i16> @fcvtzu_v8f64_v8i16(ptr %a) #0 {
; VBITS_GE_256-LABEL: fcvtzu_v8f64_v8i16:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.d, vl4
; VBITS_GE_256-NEXT:    mov x8, #4 // =0x4
; VBITS_GE_256-NEXT:    ld1d { z0.d }, p0/z, [x0, x8, lsl #3]
; VBITS_GE_256-NEXT:    ld1d { z1.d }, p0/z, [x0]
; VBITS_GE_256-NEXT:    fcvtzu z0.d, p0/m, z0.d
; VBITS_GE_256-NEXT:    fcvtzu z1.d, p0/m, z1.d
; VBITS_GE_256-NEXT:    uzp1 z0.s, z0.s, z0.s
; VBITS_GE_256-NEXT:    uzp1 z1.s, z1.s, z1.s
; VBITS_GE_256-NEXT:    uzp1 z2.h, z0.h, z0.h
; VBITS_GE_256-NEXT:    uzp1 z0.h, z1.h, z1.h
; VBITS_GE_256-NEXT:    mov v0.d[1], v2.d[0]
; VBITS_GE_256-NEXT:    // kill: def $q0 killed $q0 killed $z0
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: fcvtzu_v8f64_v8i16:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.d, vl8
; VBITS_GE_512-NEXT:    ld1d { z0.d }, p0/z, [x0]
; VBITS_GE_512-NEXT:    fcvtzu z0.d, p0/m, z0.d
; VBITS_GE_512-NEXT:    uzp1 z0.s, z0.s, z0.s
; VBITS_GE_512-NEXT:    uzp1 z0.h, z0.h, z0.h
; VBITS_GE_512-NEXT:    // kill: def $q0 killed $q0 killed $z0
; VBITS_GE_512-NEXT:    ret
  %op1 = load <8 x double>, ptr %a
  %res = fptoui <8 x double> %op1 to <8 x i16>
  ret <8 x i16> %res
}

define void @fcvtzu_v16f64_v16i16(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: fcvtzu_v16f64_v16i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl16
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.d, p0/m, z0.d
; CHECK-NEXT:    st1h { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <16 x double>, ptr %a
  %res = fptoui <16 x double> %op1 to <16 x i16>
  store <16 x i16> %res, ptr %b
  ret void
}

define void @fcvtzu_v32f64_v32i16(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: fcvtzu_v32f64_v32i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl32
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.d, p0/m, z0.d
; CHECK-NEXT:    st1h { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <32 x double>, ptr %a
  %res = fptoui <32 x double> %op1 to <32 x i16>
  store <32 x i16> %res, ptr %b
  ret void
}

;
; FCVTZU D -> S
;

; Don't use SVE for 64-bit vectors.
define <1 x i32> @fcvtzu_v1f64_v1i32(<1 x double> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v1f64_v1i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    // kill: def $d0 killed $d0 def $q0
; CHECK-NEXT:    fcvtzu v0.2d, v0.2d
; CHECK-NEXT:    xtn v0.2s, v0.2d
; CHECK-NEXT:    ret
  %res = fptoui <1 x double> %op1 to <1 x i32>
  ret <1 x i32> %res
}

; Don't use SVE for 128-bit vectors.
define <2 x i32> @fcvtzu_v2f64_v2i32(<2 x double> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v2f64_v2i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtzu v0.2d, v0.2d
; CHECK-NEXT:    xtn v0.2s, v0.2d
; CHECK-NEXT:    ret
  %res = fptoui <2 x double> %op1 to <2 x i32>
  ret <2 x i32> %res
}

define <4 x i32> @fcvtzu_v4f64_v4i32(ptr %a) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v4f64_v4i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.d, p0/m, z0.d
; CHECK-NEXT:    uzp1 z0.s, z0.s, z0.s
; CHECK-NEXT:    // kill: def $q0 killed $q0 killed $z0
; CHECK-NEXT:    ret
  %op1 = load <4 x double>, ptr %a
  %res = fptoui <4 x double> %op1 to <4 x i32>
  ret <4 x i32> %res
}

define void @fcvtzu_v8f64_v8i32(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: fcvtzu_v8f64_v8i32:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.d, vl4
; VBITS_GE_256-NEXT:    mov x8, #4 // =0x4
; VBITS_GE_256-NEXT:    ld1d { z0.d }, p0/z, [x0, x8, lsl #3]
; VBITS_GE_256-NEXT:    ld1d { z1.d }, p0/z, [x0]
; VBITS_GE_256-NEXT:    fcvtzu z0.d, p0/m, z0.d
; VBITS_GE_256-NEXT:    fcvtzu z1.d, p0/m, z1.d
; VBITS_GE_256-NEXT:    ptrue p0.s, vl4
; VBITS_GE_256-NEXT:    uzp1 z0.s, z0.s, z0.s
; VBITS_GE_256-NEXT:    uzp1 z1.s, z1.s, z1.s
; VBITS_GE_256-NEXT:    splice z1.s, p0, z1.s, z0.s
; VBITS_GE_256-NEXT:    ptrue p0.s, vl8
; VBITS_GE_256-NEXT:    st1w { z1.s }, p0, [x1]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: fcvtzu_v8f64_v8i32:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.d, vl8
; VBITS_GE_512-NEXT:    ld1d { z0.d }, p0/z, [x0]
; VBITS_GE_512-NEXT:    fcvtzu z0.d, p0/m, z0.d
; VBITS_GE_512-NEXT:    st1w { z0.d }, p0, [x1]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <8 x double>, ptr %a
  %res = fptoui <8 x double> %op1 to <8 x i32>
  store <8 x i32> %res, ptr %b
  ret void
}

define void @fcvtzu_v16f64_v16i32(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: fcvtzu_v16f64_v16i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl16
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.d, p0/m, z0.d
; CHECK-NEXT:    st1w { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <16 x double>, ptr %a
  %res = fptoui <16 x double> %op1 to <16 x i32>
  store <16 x i32> %res, ptr %b
  ret void
}

define void @fcvtzu_v32f64_v32i32(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: fcvtzu_v32f64_v32i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl32
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.d, p0/m, z0.d
; CHECK-NEXT:    st1w { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <32 x double>, ptr %a
  %res = fptoui <32 x double> %op1 to <32 x i32>
  store <32 x i32> %res, ptr %b
  ret void
}

;
; FCVTZU D -> D
;

; Don't use SVE for 64-bit vectors.
define <1 x i64> @fcvtzu_v1f64_v1i64(<1 x double> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v1f64_v1i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtzu x8, d0
; CHECK-NEXT:    fmov d0, x8
; CHECK-NEXT:    ret
  %res = fptoui <1 x double> %op1 to <1 x i64>
  ret <1 x i64> %res
}

; Don't use SVE for 128-bit vectors.
define <2 x i64> @fcvtzu_v2f64_v2i64(<2 x double> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v2f64_v2i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtzu v0.2d, v0.2d
; CHECK-NEXT:    ret
  %res = fptoui <2 x double> %op1 to <2 x i64>
  ret <2 x i64> %res
}

define void @fcvtzu_v4f64_v4i64(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzu_v4f64_v4i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <4 x double>, ptr %a
  %res = fptoui <4 x double> %op1 to <4 x i64>
  store <4 x i64> %res, ptr %b
  ret void
}

define void @fcvtzu_v8f64_v8i64(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: fcvtzu_v8f64_v8i64:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.d, vl4
; VBITS_GE_256-NEXT:    mov x8, #4 // =0x4
; VBITS_GE_256-NEXT:    ld1d { z0.d }, p0/z, [x0, x8, lsl #3]
; VBITS_GE_256-NEXT:    ld1d { z1.d }, p0/z, [x0]
; VBITS_GE_256-NEXT:    fcvtzu z0.d, p0/m, z0.d
; VBITS_GE_256-NEXT:    fcvtzu z1.d, p0/m, z1.d
; VBITS_GE_256-NEXT:    st1d { z0.d }, p0, [x1, x8, lsl #3]
; VBITS_GE_256-NEXT:    st1d { z1.d }, p0, [x1]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: fcvtzu_v8f64_v8i64:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.d, vl8
; VBITS_GE_512-NEXT:    ld1d { z0.d }, p0/z, [x0]
; VBITS_GE_512-NEXT:    fcvtzu z0.d, p0/m, z0.d
; VBITS_GE_512-NEXT:    st1d { z0.d }, p0, [x1]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <8 x double>, ptr %a
  %res = fptoui <8 x double> %op1 to <8 x i64>
  store <8 x i64> %res, ptr %b
  ret void
}

define void @fcvtzu_v16f64_v16i64(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: fcvtzu_v16f64_v16i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl16
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <16 x double>, ptr %a
  %res = fptoui <16 x double> %op1 to <16 x i64>
  store <16 x i64> %res, ptr %b
  ret void
}

define void @fcvtzu_v32f64_v32i64(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: fcvtzu_v32f64_v32i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl32
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzu z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <32 x double>, ptr %a
  %res = fptoui <32 x double> %op1 to <32 x i64>
  store <32 x i64> %res, ptr %b
  ret void
}

;
; FCVTZS H -> H
;

; Don't use SVE for 64-bit vectors.
define <4 x i16> @fcvtzs_v4f16_v4i16(<4 x half> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v4f16_v4i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtzs v0.4h, v0.4h
; CHECK-NEXT:    ret
  %res = fptosi <4 x half> %op1 to <4 x i16>
  ret <4 x i16> %res
}

; Don't use SVE for 128-bit vectors.
define void @fcvtzs_v8f16_v8i16(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v8f16_v8i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr q0, [x0]
; CHECK-NEXT:    fcvtzs v0.8h, v0.8h
; CHECK-NEXT:    str q0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <8 x half>, ptr %a
  %res = fptosi <8 x half> %op1 to <8 x i16>
  store <8 x i16> %res, ptr %b
  ret void
}

define void @fcvtzs_v16f16_v16i16(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v16f16_v16i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl16
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <16 x half>, ptr %a
  %res = fptosi <16 x half> %op1 to <16 x i16>
  store <16 x i16> %res, ptr %b
  ret void
}

define void @fcvtzs_v32f16_v32i16(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: fcvtzs_v32f16_v32i16:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.h, vl16
; VBITS_GE_256-NEXT:    mov x8, #16 // =0x10
; VBITS_GE_256-NEXT:    ld1h { z0.h }, p0/z, [x0, x8, lsl #1]
; VBITS_GE_256-NEXT:    ld1h { z1.h }, p0/z, [x0]
; VBITS_GE_256-NEXT:    fcvtzs z0.h, p0/m, z0.h
; VBITS_GE_256-NEXT:    fcvtzs z1.h, p0/m, z1.h
; VBITS_GE_256-NEXT:    st1h { z0.h }, p0, [x1, x8, lsl #1]
; VBITS_GE_256-NEXT:    st1h { z1.h }, p0, [x1]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: fcvtzs_v32f16_v32i16:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.h, vl32
; VBITS_GE_512-NEXT:    ld1h { z0.h }, p0/z, [x0]
; VBITS_GE_512-NEXT:    fcvtzs z0.h, p0/m, z0.h
; VBITS_GE_512-NEXT:    st1h { z0.h }, p0, [x1]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <32 x half>, ptr %a
  %res = fptosi <32 x half> %op1 to <32 x i16>
  store <32 x i16> %res, ptr %b
  ret void
}

define void @fcvtzs_v64f16_v64i16(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: fcvtzs_v64f16_v64i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl64
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <64 x half>, ptr %a
  %res = fptosi <64 x half> %op1 to <64 x i16>
  store <64 x i16> %res, ptr %b
  ret void
}

define void @fcvtzs_v128f16_v128i16(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: fcvtzs_v128f16_v128i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl128
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <128 x half>, ptr %a
  %res = fptosi <128 x half> %op1 to <128 x i16>
  store <128 x i16> %res, ptr %b
  ret void
}

;
; FCVTZS H -> S
;

; Don't use SVE for 64-bit vectors.
define <2 x i32> @fcvtzs_v2f16_v2i32(<2 x half> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v2f16_v2i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtl v0.4s, v0.4h
; CHECK-NEXT:    fcvtzs v0.4s, v0.4s
; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $q0
; CHECK-NEXT:    ret
  %res = fptosi <2 x half> %op1 to <2 x i32>
  ret <2 x i32> %res
}

; Don't use SVE for 128-bit vectors.
define <4 x i32> @fcvtzs_v4f16_v4i32(<4 x half> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v4f16_v4i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtl v0.4s, v0.4h
; CHECK-NEXT:    fcvtzs v0.4s, v0.4s
; CHECK-NEXT:    ret
  %res = fptosi <4 x half> %op1 to <4 x i32>
  ret <4 x i32> %res
}

define void @fcvtzs_v8f16_v8i32(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v8f16_v8i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr q0, [x0]
; CHECK-NEXT:    ptrue p0.s, vl8
; CHECK-NEXT:    uunpklo z0.s, z0.h
; CHECK-NEXT:    fcvtzs z0.s, p0/m, z0.h
; CHECK-NEXT:    st1w { z0.s }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <8 x half>, ptr %a
  %res = fptosi <8 x half> %op1 to <8 x i32>
  store <8 x i32> %res, ptr %b
  ret void
}

define void @fcvtzs_v16f16_v16i32(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: fcvtzs_v16f16_v16i32:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.h, vl16
; VBITS_GE_256-NEXT:    mov x8, #8 // =0x8
; VBITS_GE_256-NEXT:    ld1h { z0.h }, p0/z, [x0]
; VBITS_GE_256-NEXT:    ptrue p0.s, vl8
; VBITS_GE_256-NEXT:    uunpklo z1.s, z0.h
; VBITS_GE_256-NEXT:    ext z0.b, z0.b, z0.b, #16
; VBITS_GE_256-NEXT:    uunpklo z0.s, z0.h
; VBITS_GE_256-NEXT:    fcvtzs z1.s, p0/m, z1.h
; VBITS_GE_256-NEXT:    fcvtzs z0.s, p0/m, z0.h
; VBITS_GE_256-NEXT:    st1w { z1.s }, p0, [x1]
; VBITS_GE_256-NEXT:    st1w { z0.s }, p0, [x1, x8, lsl #2]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: fcvtzs_v16f16_v16i32:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.s, vl16
; VBITS_GE_512-NEXT:    ld1h { z0.s }, p0/z, [x0]
; VBITS_GE_512-NEXT:    fcvtzs z0.s, p0/m, z0.h
; VBITS_GE_512-NEXT:    st1w { z0.s }, p0, [x1]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <16 x half>, ptr %a
  %res = fptosi <16 x half> %op1 to <16 x i32>
  store <16 x i32> %res, ptr %b
  ret void
}

define void @fcvtzs_v32f16_v32i32(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: fcvtzs_v32f16_v32i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl32
; CHECK-NEXT:    ld1h { z0.s }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.s, p0/m, z0.h
; CHECK-NEXT:    st1w { z0.s }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <32 x half>, ptr %a
  %res = fptosi <32 x half> %op1 to <32 x i32>
  store <32 x i32> %res, ptr %b
  ret void
}

define void @fcvtzs_v64f16_v64i32(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: fcvtzs_v64f16_v64i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl64
; CHECK-NEXT:    ld1h { z0.s }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.s, p0/m, z0.h
; CHECK-NEXT:    st1w { z0.s }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <64 x half>, ptr %a
  %res = fptosi <64 x half> %op1 to <64 x i32>
  store <64 x i32> %res, ptr %b
  ret void
}

;
; FCVTZS H -> D
;

; Don't use SVE for 64-bit vectors.
define <1 x i64> @fcvtzs_v1f16_v1i64(<1 x half> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v1f16_v1i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtzs x8, h0
; CHECK-NEXT:    fmov d0, x8
; CHECK-NEXT:    ret
  %res = fptosi <1 x half> %op1 to <1 x i64>
  ret <1 x i64> %res
}

; v2f16 is not legal for NEON, so use SVE
define <2 x i64> @fcvtzs_v2f16_v2i64(<2 x half> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v2f16_v2i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    // kill: def $d0 killed $d0 def $z0
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    uunpklo z0.s, z0.h
; CHECK-NEXT:    uunpklo z0.d, z0.s
; CHECK-NEXT:    fcvtzs z0.d, p0/m, z0.h
; CHECK-NEXT:    // kill: def $q0 killed $q0 killed $z0
; CHECK-NEXT:    ret
  %res = fptosi <2 x half> %op1 to <2 x i64>
  ret <2 x i64> %res
}

define void @fcvtzs_v4f16_v4i64(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v4f16_v4i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr d0, [x0]
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    uunpklo z0.s, z0.h
; CHECK-NEXT:    uunpklo z0.d, z0.s
; CHECK-NEXT:    fcvtzs z0.d, p0/m, z0.h
; CHECK-NEXT:    st1d { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <4 x half>, ptr %a
  %res = fptosi <4 x half> %op1 to <4 x i64>
  store <4 x i64> %res, ptr %b
  ret void
}

define void @fcvtzs_v8f16_v8i64(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: fcvtzs_v8f16_v8i64:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ldr q0, [x0]
; VBITS_GE_256-NEXT:    ptrue p0.d, vl4
; VBITS_GE_256-NEXT:    mov x8, #4 // =0x4
; VBITS_GE_256-NEXT:    ext v1.16b, v0.16b, v0.16b, #8
; VBITS_GE_256-NEXT:    uunpklo z0.s, z0.h
; VBITS_GE_256-NEXT:    uunpklo z1.s, z1.h
; VBITS_GE_256-NEXT:    uunpklo z0.d, z0.s
; VBITS_GE_256-NEXT:    uunpklo z1.d, z1.s
; VBITS_GE_256-NEXT:    fcvtzs z0.d, p0/m, z0.h
; VBITS_GE_256-NEXT:    fcvtzs z1.d, p0/m, z1.h
; VBITS_GE_256-NEXT:    st1d { z0.d }, p0, [x1]
; VBITS_GE_256-NEXT:    st1d { z1.d }, p0, [x1, x8, lsl #3]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: fcvtzs_v8f16_v8i64:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ldr q0, [x0]
; VBITS_GE_512-NEXT:    ptrue p0.d, vl8
; VBITS_GE_512-NEXT:    uunpklo z0.s, z0.h
; VBITS_GE_512-NEXT:    uunpklo z0.d, z0.s
; VBITS_GE_512-NEXT:    fcvtzs z0.d, p0/m, z0.h
; VBITS_GE_512-NEXT:    st1d { z0.d }, p0, [x1]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <8 x half>, ptr %a
  %res = fptosi <8 x half> %op1 to <8 x i64>
  store <8 x i64> %res, ptr %b
  ret void
}

define void @fcvtzs_v16f16_v16i64(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: fcvtzs_v16f16_v16i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl16
; CHECK-NEXT:    ld1h { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.d, p0/m, z0.h
; CHECK-NEXT:    st1d { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <16 x half>, ptr %a
  %res = fptosi <16 x half> %op1 to <16 x i64>
  store <16 x i64> %res, ptr %b
  ret void
}

define void @fcvtzs_v32f16_v32i64(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: fcvtzs_v32f16_v32i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl32
; CHECK-NEXT:    ld1h { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.d, p0/m, z0.h
; CHECK-NEXT:    st1d { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <32 x half>, ptr %a
  %res = fptosi <32 x half> %op1 to <32 x i64>
  store <32 x i64> %res, ptr %b
  ret void
}

;
; FCVTZS S -> H
;

; Don't use SVE for 64-bit vectors.
define <2 x i16> @fcvtzs_v2f32_v2i16(<2 x float> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v2f32_v2i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtzs v0.2s, v0.2s
; CHECK-NEXT:    ret
  %res = fptosi <2 x float> %op1 to <2 x i16>
  ret <2 x i16> %res
}

; Don't use SVE for 128-bit vectors.
define <4 x i16> @fcvtzs_v4f32_v4i16(<4 x float> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v4f32_v4i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtzs v1.4s, v0.4s
; CHECK-NEXT:    mov w8, v1.s[1]
; CHECK-NEXT:    mov v0.16b, v1.16b
; CHECK-NEXT:    mov w9, v1.s[2]
; CHECK-NEXT:    mov v0.h[1], w8
; CHECK-NEXT:    mov w8, v1.s[3]
; CHECK-NEXT:    mov v0.h[2], w9
; CHECK-NEXT:    mov v0.h[3], w8
; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $q0
; CHECK-NEXT:    ret
  %res = fptosi <4 x float> %op1 to <4 x i16>
  ret <4 x i16> %res
}

define <8 x i16> @fcvtzs_v8f32_v8i16(ptr %a) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v8f32_v8i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl8
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.s, p0/m, z0.s
; CHECK-NEXT:    uzp1 z0.h, z0.h, z0.h
; CHECK-NEXT:    // kill: def $q0 killed $q0 killed $z0
; CHECK-NEXT:    ret
  %op1 = load <8 x float>, ptr %a
  %res = fptosi <8 x float> %op1 to <8 x i16>
  ret <8 x i16> %res
}

define void @fcvtzs_v16f32_v16i16(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: fcvtzs_v16f32_v16i16:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.s, vl8
; VBITS_GE_256-NEXT:    mov x8, #8 // =0x8
; VBITS_GE_256-NEXT:    ld1w { z0.s }, p0/z, [x0, x8, lsl #2]
; VBITS_GE_256-NEXT:    ld1w { z1.s }, p0/z, [x0]
; VBITS_GE_256-NEXT:    fcvtzs z0.s, p0/m, z0.s
; VBITS_GE_256-NEXT:    fcvtzs z1.s, p0/m, z1.s
; VBITS_GE_256-NEXT:    ptrue p0.h, vl8
; VBITS_GE_256-NEXT:    uzp1 z0.h, z0.h, z0.h
; VBITS_GE_256-NEXT:    uzp1 z1.h, z1.h, z1.h
; VBITS_GE_256-NEXT:    splice z1.h, p0, z1.h, z0.h
; VBITS_GE_256-NEXT:    ptrue p0.h, vl16
; VBITS_GE_256-NEXT:    st1h { z1.h }, p0, [x1]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: fcvtzs_v16f32_v16i16:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.s, vl16
; VBITS_GE_512-NEXT:    ld1w { z0.s }, p0/z, [x0]
; VBITS_GE_512-NEXT:    fcvtzs z0.s, p0/m, z0.s
; VBITS_GE_512-NEXT:    st1h { z0.s }, p0, [x1]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <16 x float>, ptr %a
  %res = fptosi <16 x float> %op1 to <16 x i16>
  store <16 x i16> %res, ptr %b
  ret void
}

define void @fcvtzs_v32f32_v32i16(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: fcvtzs_v32f32_v32i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl32
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.s, p0/m, z0.s
; CHECK-NEXT:    st1h { z0.s }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <32 x float>, ptr %a
  %res = fptosi <32 x float> %op1 to <32 x i16>
  store <32 x i16> %res, ptr %b
  ret void
}

define void @fcvtzs_v64f32_v64i16(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: fcvtzs_v64f32_v64i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl64
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.s, p0/m, z0.s
; CHECK-NEXT:    st1h { z0.s }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <64 x float>, ptr %a
  %res = fptosi <64 x float> %op1 to <64 x i16>
  store <64 x i16> %res, ptr %b
  ret void
}

;
; FCVTZS S -> S
;

; Don't use SVE for 64-bit vectors.
define <2 x i32> @fcvtzs_v2f32_v2i32(<2 x float> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v2f32_v2i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtzs v0.2s, v0.2s
; CHECK-NEXT:    ret
  %res = fptosi <2 x float> %op1 to <2 x i32>
  ret <2 x i32> %res
}

; Don't use SVE for 128-bit vectors.
define <4 x i32> @fcvtzs_v4f32_v4i32(<4 x float> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v4f32_v4i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtzs v0.4s, v0.4s
; CHECK-NEXT:    ret
  %res = fptosi <4 x float> %op1 to <4 x i32>
  ret <4 x i32> %res
}

define void @fcvtzs_v8f32_v8i32(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v8f32_v8i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl8
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <8 x float>, ptr %a
  %res = fptosi <8 x float> %op1 to <8 x i32>
  store <8 x i32> %res, ptr %b
  ret void
}

define void @fcvtzs_v16f32_v16i32(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: fcvtzs_v16f32_v16i32:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.s, vl8
; VBITS_GE_256-NEXT:    mov x8, #8 // =0x8
; VBITS_GE_256-NEXT:    ld1w { z0.s }, p0/z, [x0, x8, lsl #2]
; VBITS_GE_256-NEXT:    ld1w { z1.s }, p0/z, [x0]
; VBITS_GE_256-NEXT:    fcvtzs z0.s, p0/m, z0.s
; VBITS_GE_256-NEXT:    fcvtzs z1.s, p0/m, z1.s
; VBITS_GE_256-NEXT:    st1w { z0.s }, p0, [x1, x8, lsl #2]
; VBITS_GE_256-NEXT:    st1w { z1.s }, p0, [x1]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: fcvtzs_v16f32_v16i32:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.s, vl16
; VBITS_GE_512-NEXT:    ld1w { z0.s }, p0/z, [x0]
; VBITS_GE_512-NEXT:    fcvtzs z0.s, p0/m, z0.s
; VBITS_GE_512-NEXT:    st1w { z0.s }, p0, [x1]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <16 x float>, ptr %a
  %res = fptosi <16 x float> %op1 to <16 x i32>
  store <16 x i32> %res, ptr %b
  ret void
}

define void @fcvtzs_v32f32_v32i32(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: fcvtzs_v32f32_v32i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl32
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <32 x float>, ptr %a
  %res = fptosi <32 x float> %op1 to <32 x i32>
  store <32 x i32> %res, ptr %b
  ret void
}

define void @fcvtzs_v64f32_v64i32(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: fcvtzs_v64f32_v64i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl64
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <64 x float>, ptr %a
  %res = fptosi <64 x float> %op1 to <64 x i32>
  store <64 x i32> %res, ptr %b
  ret void
}

;
; FCVTZS S -> D
;

; Don't use SVE for 64-bit vectors.
define <1 x i64> @fcvtzs_v1f32_v1i64(<1 x float> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v1f32_v1i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtl v0.2d, v0.2s
; CHECK-NEXT:    fcvtzs v0.2d, v0.2d
; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $q0
; CHECK-NEXT:    ret
  %res = fptosi <1 x float> %op1 to <1 x i64>
  ret <1 x i64> %res
}

; Don't use SVE for 128-bit vectors.
define <2 x i64> @fcvtzs_v2f32_v2i64(<2 x float> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v2f32_v2i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtl v0.2d, v0.2s
; CHECK-NEXT:    fcvtzs v0.2d, v0.2d
; CHECK-NEXT:    ret
  %res = fptosi <2 x float> %op1 to <2 x i64>
  ret <2 x i64> %res
}

define void @fcvtzs_v4f32_v4i64(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v4f32_v4i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr q0, [x0]
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    uunpklo z0.d, z0.s
; CHECK-NEXT:    fcvtzs z0.d, p0/m, z0.s
; CHECK-NEXT:    st1d { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <4 x float>, ptr %a
  %res = fptosi <4 x float> %op1 to <4 x i64>
  store <4 x i64> %res, ptr %b
  ret void
}

define void @fcvtzs_v8f32_v8i64(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: fcvtzs_v8f32_v8i64:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.s, vl8
; VBITS_GE_256-NEXT:    mov x8, #4 // =0x4
; VBITS_GE_256-NEXT:    ld1w { z0.s }, p0/z, [x0]
; VBITS_GE_256-NEXT:    ptrue p0.d, vl4
; VBITS_GE_256-NEXT:    uunpklo z1.d, z0.s
; VBITS_GE_256-NEXT:    ext z0.b, z0.b, z0.b, #16
; VBITS_GE_256-NEXT:    uunpklo z0.d, z0.s
; VBITS_GE_256-NEXT:    fcvtzs z1.d, p0/m, z1.s
; VBITS_GE_256-NEXT:    fcvtzs z0.d, p0/m, z0.s
; VBITS_GE_256-NEXT:    st1d { z1.d }, p0, [x1]
; VBITS_GE_256-NEXT:    st1d { z0.d }, p0, [x1, x8, lsl #3]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: fcvtzs_v8f32_v8i64:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.d, vl8
; VBITS_GE_512-NEXT:    ld1w { z0.d }, p0/z, [x0]
; VBITS_GE_512-NEXT:    fcvtzs z0.d, p0/m, z0.s
; VBITS_GE_512-NEXT:    st1d { z0.d }, p0, [x1]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <8 x float>, ptr %a
  %res = fptosi <8 x float> %op1 to <8 x i64>
  store <8 x i64> %res, ptr %b
  ret void
}

define void @fcvtzs_v16f32_v16i64(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: fcvtzs_v16f32_v16i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl16
; CHECK-NEXT:    ld1w { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.d, p0/m, z0.s
; CHECK-NEXT:    st1d { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <16 x float>, ptr %a
  %res = fptosi <16 x float> %op1 to <16 x i64>
  store <16 x i64> %res, ptr %b
  ret void
}

define void @fcvtzs_v32f32_v32i64(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: fcvtzs_v32f32_v32i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl32
; CHECK-NEXT:    ld1w { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.d, p0/m, z0.s
; CHECK-NEXT:    st1d { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <32 x float>, ptr %a
  %res = fptosi <32 x float> %op1 to <32 x i64>
  store <32 x i64> %res, ptr %b
  ret void
}


;
; FCVTZS D -> H
;

; v1f64 is perfered to be widened to v4f64, so use SVE
define <1 x i16> @fcvtzs_v1f64_v1i16(<1 x double> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v1f64_v1i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    // kill: def $d0 killed $d0 def $z0
; CHECK-NEXT:    fcvtzs z0.d, p0/m, z0.d
; CHECK-NEXT:    uzp1 z0.s, z0.s, z0.s
; CHECK-NEXT:    uzp1 z0.h, z0.h, z0.h
; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $z0
; CHECK-NEXT:    ret
  %res = fptosi <1 x double> %op1 to <1 x i16>
  ret <1 x i16> %res
}

; Don't use SVE for 128-bit vectors.
define <2 x i16> @fcvtzs_v2f64_v2i16(<2 x double> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v2f64_v2i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtzs v0.2d, v0.2d
; CHECK-NEXT:    xtn v0.2s, v0.2d
; CHECK-NEXT:    ret
  %res = fptosi <2 x double> %op1 to <2 x i16>
  ret <2 x i16> %res
}

define <4 x i16> @fcvtzs_v4f64_v4i16(ptr %a) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v4f64_v4i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.d, p0/m, z0.d
; CHECK-NEXT:    uzp1 z0.s, z0.s, z0.s
; CHECK-NEXT:    uzp1 z0.h, z0.h, z0.h
; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $z0
; CHECK-NEXT:    ret
  %op1 = load <4 x double>, ptr %a
  %res = fptosi <4 x double> %op1 to <4 x i16>
  ret <4 x i16> %res
}

define <8 x i16> @fcvtzs_v8f64_v8i16(ptr %a) #0 {
; VBITS_GE_256-LABEL: fcvtzs_v8f64_v8i16:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.d, vl4
; VBITS_GE_256-NEXT:    mov x8, #4 // =0x4
; VBITS_GE_256-NEXT:    ld1d { z0.d }, p0/z, [x0, x8, lsl #3]
; VBITS_GE_256-NEXT:    ld1d { z1.d }, p0/z, [x0]
; VBITS_GE_256-NEXT:    fcvtzs z0.d, p0/m, z0.d
; VBITS_GE_256-NEXT:    fcvtzs z1.d, p0/m, z1.d
; VBITS_GE_256-NEXT:    uzp1 z0.s, z0.s, z0.s
; VBITS_GE_256-NEXT:    uzp1 z1.s, z1.s, z1.s
; VBITS_GE_256-NEXT:    uzp1 z2.h, z0.h, z0.h
; VBITS_GE_256-NEXT:    uzp1 z0.h, z1.h, z1.h
; VBITS_GE_256-NEXT:    mov v0.d[1], v2.d[0]
; VBITS_GE_256-NEXT:    // kill: def $q0 killed $q0 killed $z0
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: fcvtzs_v8f64_v8i16:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.d, vl8
; VBITS_GE_512-NEXT:    ld1d { z0.d }, p0/z, [x0]
; VBITS_GE_512-NEXT:    fcvtzs z0.d, p0/m, z0.d
; VBITS_GE_512-NEXT:    uzp1 z0.s, z0.s, z0.s
; VBITS_GE_512-NEXT:    uzp1 z0.h, z0.h, z0.h
; VBITS_GE_512-NEXT:    // kill: def $q0 killed $q0 killed $z0
; VBITS_GE_512-NEXT:    ret
  %op1 = load <8 x double>, ptr %a
  %res = fptosi <8 x double> %op1 to <8 x i16>
  ret <8 x i16> %res
}

define void @fcvtzs_v16f64_v16i16(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: fcvtzs_v16f64_v16i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl16
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.d, p0/m, z0.d
; CHECK-NEXT:    st1h { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <16 x double>, ptr %a
  %res = fptosi <16 x double> %op1 to <16 x i16>
  store <16 x i16> %res, ptr %b
  ret void
}

define void @fcvtzs_v32f64_v32i16(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: fcvtzs_v32f64_v32i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl32
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.d, p0/m, z0.d
; CHECK-NEXT:    st1h { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <32 x double>, ptr %a
  %res = fptosi <32 x double> %op1 to <32 x i16>
  store <32 x i16> %res, ptr %b
  ret void
}

;
; FCVTZS D -> S
;

; Don't use SVE for 64-bit vectors.
define <1 x i32> @fcvtzs_v1f64_v1i32(<1 x double> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v1f64_v1i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    // kill: def $d0 killed $d0 def $q0
; CHECK-NEXT:    fcvtzs v0.2d, v0.2d
; CHECK-NEXT:    xtn v0.2s, v0.2d
; CHECK-NEXT:    ret
  %res = fptosi <1 x double> %op1 to <1 x i32>
  ret <1 x i32> %res
}

; Don't use SVE for 128-bit vectors.
define <2 x i32> @fcvtzs_v2f64_v2i32(<2 x double> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v2f64_v2i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtzs v0.2d, v0.2d
; CHECK-NEXT:    xtn v0.2s, v0.2d
; CHECK-NEXT:    ret
  %res = fptosi <2 x double> %op1 to <2 x i32>
  ret <2 x i32> %res
}

define <4 x i32> @fcvtzs_v4f64_v4i32(ptr %a) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v4f64_v4i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.d, p0/m, z0.d
; CHECK-NEXT:    uzp1 z0.s, z0.s, z0.s
; CHECK-NEXT:    // kill: def $q0 killed $q0 killed $z0
; CHECK-NEXT:    ret
  %op1 = load <4 x double>, ptr %a
  %res = fptosi <4 x double> %op1 to <4 x i32>
  ret <4 x i32> %res
}

define void @fcvtzs_v8f64_v8i32(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: fcvtzs_v8f64_v8i32:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.d, vl4
; VBITS_GE_256-NEXT:    mov x8, #4 // =0x4
; VBITS_GE_256-NEXT:    ld1d { z0.d }, p0/z, [x0, x8, lsl #3]
; VBITS_GE_256-NEXT:    ld1d { z1.d }, p0/z, [x0]
; VBITS_GE_256-NEXT:    fcvtzs z0.d, p0/m, z0.d
; VBITS_GE_256-NEXT:    fcvtzs z1.d, p0/m, z1.d
; VBITS_GE_256-NEXT:    ptrue p0.s, vl4
; VBITS_GE_256-NEXT:    uzp1 z0.s, z0.s, z0.s
; VBITS_GE_256-NEXT:    uzp1 z1.s, z1.s, z1.s
; VBITS_GE_256-NEXT:    splice z1.s, p0, z1.s, z0.s
; VBITS_GE_256-NEXT:    ptrue p0.s, vl8
; VBITS_GE_256-NEXT:    st1w { z1.s }, p0, [x1]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: fcvtzs_v8f64_v8i32:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.d, vl8
; VBITS_GE_512-NEXT:    ld1d { z0.d }, p0/z, [x0]
; VBITS_GE_512-NEXT:    fcvtzs z0.d, p0/m, z0.d
; VBITS_GE_512-NEXT:    st1w { z0.d }, p0, [x1]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <8 x double>, ptr %a
  %res = fptosi <8 x double> %op1 to <8 x i32>
  store <8 x i32> %res, ptr %b
  ret void
}

define void @fcvtzs_v16f64_v16i32(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: fcvtzs_v16f64_v16i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl16
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.d, p0/m, z0.d
; CHECK-NEXT:    st1w { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <16 x double>, ptr %a
  %res = fptosi <16 x double> %op1 to <16 x i32>
  store <16 x i32> %res, ptr %b
  ret void
}

define void @fcvtzs_v32f64_v32i32(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: fcvtzs_v32f64_v32i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl32
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.d, p0/m, z0.d
; CHECK-NEXT:    st1w { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <32 x double>, ptr %a
  %res = fptosi <32 x double> %op1 to <32 x i32>
  store <32 x i32> %res, ptr %b
  ret void
}

;
; FCVTZS D -> D
;

; Don't use SVE for 64-bit vectors.
define <1 x i64> @fcvtzs_v1f64_v1i64(<1 x double> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v1f64_v1i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtzs x8, d0
; CHECK-NEXT:    fmov d0, x8
; CHECK-NEXT:    ret
  %res = fptosi <1 x double> %op1 to <1 x i64>
  ret <1 x i64> %res
}

; Don't use SVE for 128-bit vectors.
define <2 x i64> @fcvtzs_v2f64_v2i64(<2 x double> %op1) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v2f64_v2i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    fcvtzs v0.2d, v0.2d
; CHECK-NEXT:    ret
  %res = fptosi <2 x double> %op1 to <2 x i64>
  ret <2 x i64> %res
}

define void @fcvtzs_v4f64_v4i64(ptr %a, ptr %b) vscale_range(2,0) #0 {
; CHECK-LABEL: fcvtzs_v4f64_v4i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <4 x double>, ptr %a
  %res = fptosi <4 x double> %op1 to <4 x i64>
  store <4 x i64> %res, ptr %b
  ret void
}

define void @fcvtzs_v8f64_v8i64(ptr %a, ptr %b) #0 {
; VBITS_GE_256-LABEL: fcvtzs_v8f64_v8i64:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    ptrue p0.d, vl4
; VBITS_GE_256-NEXT:    mov x8, #4 // =0x4
; VBITS_GE_256-NEXT:    ld1d { z0.d }, p0/z, [x0, x8, lsl #3]
; VBITS_GE_256-NEXT:    ld1d { z1.d }, p0/z, [x0]
; VBITS_GE_256-NEXT:    fcvtzs z0.d, p0/m, z0.d
; VBITS_GE_256-NEXT:    fcvtzs z1.d, p0/m, z1.d
; VBITS_GE_256-NEXT:    st1d { z0.d }, p0, [x1, x8, lsl #3]
; VBITS_GE_256-NEXT:    st1d { z1.d }, p0, [x1]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: fcvtzs_v8f64_v8i64:
; VBITS_GE_512:       // %bb.0:
; VBITS_GE_512-NEXT:    ptrue p0.d, vl8
; VBITS_GE_512-NEXT:    ld1d { z0.d }, p0/z, [x0]
; VBITS_GE_512-NEXT:    fcvtzs z0.d, p0/m, z0.d
; VBITS_GE_512-NEXT:    st1d { z0.d }, p0, [x1]
; VBITS_GE_512-NEXT:    ret
  %op1 = load <8 x double>, ptr %a
  %res = fptosi <8 x double> %op1 to <8 x i64>
  store <8 x i64> %res, ptr %b
  ret void
}

define void @fcvtzs_v16f64_v16i64(ptr %a, ptr %b) vscale_range(8,0) #0 {
; CHECK-LABEL: fcvtzs_v16f64_v16i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl16
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <16 x double>, ptr %a
  %res = fptosi <16 x double> %op1 to <16 x i64>
  store <16 x i64> %res, ptr %b
  ret void
}

define void @fcvtzs_v32f64_v32i64(ptr %a, ptr %b) vscale_range(16,0) #0 {
; CHECK-LABEL: fcvtzs_v32f64_v32i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl32
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    fcvtzs z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x1]
; CHECK-NEXT:    ret
  %op1 = load <32 x double>, ptr %a
  %res = fptosi <32 x double> %op1 to <32 x i64>
  store <32 x i64> %res, ptr %b
  ret void
}

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