test/CodeGen/AArch64/sve-fixed-length-fp-rounding.ll

; 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"

;
; CEIL -> FRINTP
;

; Don't use SVE for 64-bit vectors.
define <4 x half> @frintp_v4f16(<4 x half> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintp_v4f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintp v0.4h, v0.4h
; CHECK-NEXT:    ret
  %res = call <4 x half> @llvm.ceil.v4f16(<4 x half> %op)
  ret <4 x half> %res
}

; Don't use SVE for 128-bit vectors.
define <8 x half> @frintp_v8f16(<8 x half> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintp_v8f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintp v0.8h, v0.8h
; CHECK-NEXT:    ret
  %res = call <8 x half> @llvm.ceil.v8f16(<8 x half> %op)
  ret <8 x half> %res
}

define void @frintp_v16f16(<16 x half>* %a) vscale_range(2,0) #0 {
; CHECK-LABEL: frintp_v16f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl16
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    frintp z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <16 x half>, <16 x half>* %a
  %res = call <16 x half> @llvm.ceil.v16f16(<16 x half> %op)
  store <16 x half> %res, <16 x half>* %a
  ret void
}

define void @frintp_v32f16(<32 x half>* %a) #0 {
; VBITS_GE_256-LABEL: frintp_v32f16:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    mov x8, #16
; VBITS_GE_256-NEXT:    ptrue p0.h, vl16
; 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:    frintp z0.h, p0/m, z0.h
; VBITS_GE_256-NEXT:    frintp z1.h, p0/m, z1.h
; VBITS_GE_256-NEXT:    st1h { z0.h }, p0, [x0, x8, lsl #1]
; VBITS_GE_256-NEXT:    st1h { z1.h }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: frintp_v32f16:
; 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:    frintp z0.h, p0/m, z0.h
; VBITS_GE_512-NEXT:    st1h { z0.h }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op = load <32 x half>, <32 x half>* %a
  %res = call <32 x half> @llvm.ceil.v32f16(<32 x half> %op)
  store <32 x half> %res, <32 x half>* %a
  ret void
}

define void @frintp_v64f16(<64 x half>* %a) vscale_range(8,0) #0 {
; CHECK-LABEL: frintp_v64f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl64
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    frintp z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <64 x half>, <64 x half>* %a
  %res = call <64 x half> @llvm.ceil.v64f16(<64 x half> %op)
  store <64 x half> %res, <64 x half>* %a
  ret void
}

define void @frintp_v128f16(<128 x half>* %a) vscale_range(16,0) #0 {
; CHECK-LABEL: frintp_v128f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl128
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    frintp z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <128 x half>, <128 x half>* %a
  %res = call <128 x half> @llvm.ceil.v128f16(<128 x half> %op)
  store <128 x half> %res, <128 x half>* %a
  ret void
}

; Don't use SVE for 64-bit vectors.
define <2 x float> @frintp_v2f32(<2 x float> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintp_v2f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintp v0.2s, v0.2s
; CHECK-NEXT:    ret
  %res = call <2 x float> @llvm.ceil.v2f32(<2 x float> %op)
  ret <2 x float> %res
}

; Don't use SVE for 128-bit vectors.
define <4 x float> @frintp_v4f32(<4 x float> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintp_v4f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintp v0.4s, v0.4s
; CHECK-NEXT:    ret
  %res = call <4 x float> @llvm.ceil.v4f32(<4 x float> %op)
  ret <4 x float> %res
}

define void @frintp_v8f32(<8 x float>* %a) vscale_range(2,0) #0 {
; CHECK-LABEL: frintp_v8f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl8
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    frintp z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <8 x float>, <8 x float>* %a
  %res = call <8 x float> @llvm.ceil.v8f32(<8 x float> %op)
  store <8 x float> %res, <8 x float>* %a
  ret void
}

define void @frintp_v16f32(<16 x float>* %a) #0 {
; VBITS_GE_256-LABEL: frintp_v16f32:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    mov x8, #8
; VBITS_GE_256-NEXT:    ptrue p0.s, vl8
; 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:    frintp z0.s, p0/m, z0.s
; VBITS_GE_256-NEXT:    frintp z1.s, p0/m, z1.s
; VBITS_GE_256-NEXT:    st1w { z0.s }, p0, [x0, x8, lsl #2]
; VBITS_GE_256-NEXT:    st1w { z1.s }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: frintp_v16f32:
; 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:    frintp z0.s, p0/m, z0.s
; VBITS_GE_512-NEXT:    st1w { z0.s }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op = load <16 x float>, <16 x float>* %a
  %res = call <16 x float> @llvm.ceil.v16f32(<16 x float> %op)
  store <16 x float> %res, <16 x float>* %a
  ret void
}

define void @frintp_v32f32(<32 x float>* %a) vscale_range(8,0) #0 {
; CHECK-LABEL: frintp_v32f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl32
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    frintp z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <32 x float>, <32 x float>* %a
  %res = call <32 x float> @llvm.ceil.v32f32(<32 x float> %op)
  store <32 x float> %res, <32 x float>* %a
  ret void
}

define void @frintp_v64f32(<64 x float>* %a) vscale_range(16,0) #0 {
; CHECK-LABEL: frintp_v64f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl64
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    frintp z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <64 x float>, <64 x float>* %a
  %res = call <64 x float> @llvm.ceil.v64f32(<64 x float> %op)
  store <64 x float> %res, <64 x float>* %a
  ret void
}

; Don't use SVE for 64-bit vectors.
define <1 x double> @frintp_v1f64(<1 x double> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintp_v1f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintp d0, d0
; CHECK-NEXT:    ret
  %res = call <1 x double> @llvm.ceil.v1f64(<1 x double> %op)
  ret <1 x double> %res
}

; Don't use SVE for 128-bit vectors.
define <2 x double> @frintp_v2f64(<2 x double> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintp_v2f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintp v0.2d, v0.2d
; CHECK-NEXT:    ret
  %res = call <2 x double> @llvm.ceil.v2f64(<2 x double> %op)
  ret <2 x double> %res
}

define void @frintp_v4f64(<4 x double>* %a) vscale_range(2,0) #0 {
; CHECK-LABEL: frintp_v4f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    frintp z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <4 x double>, <4 x double>* %a
  %res = call <4 x double> @llvm.ceil.v4f64(<4 x double> %op)
  store <4 x double> %res, <4 x double>* %a
  ret void
}

define void @frintp_v8f64(<8 x double>* %a) #0 {
; VBITS_GE_256-LABEL: frintp_v8f64:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    mov x8, #4
; VBITS_GE_256-NEXT:    ptrue p0.d, vl4
; 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:    frintp z0.d, p0/m, z0.d
; VBITS_GE_256-NEXT:    frintp z1.d, p0/m, z1.d
; VBITS_GE_256-NEXT:    st1d { z0.d }, p0, [x0, x8, lsl #3]
; VBITS_GE_256-NEXT:    st1d { z1.d }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: frintp_v8f64:
; 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:    frintp z0.d, p0/m, z0.d
; VBITS_GE_512-NEXT:    st1d { z0.d }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op = load <8 x double>, <8 x double>* %a
  %res = call <8 x double> @llvm.ceil.v8f64(<8 x double> %op)
  store <8 x double> %res, <8 x double>* %a
  ret void
}

define void @frintp_v16f64(<16 x double>* %a) vscale_range(8,0) #0 {
; CHECK-LABEL: frintp_v16f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl16
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    frintp z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <16 x double>, <16 x double>* %a
  %res = call <16 x double> @llvm.ceil.v16f64(<16 x double> %op)
  store <16 x double> %res, <16 x double>* %a
  ret void
}

define void @frintp_v32f64(<32 x double>* %a) vscale_range(16,0) #0 {
; CHECK-LABEL: frintp_v32f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl32
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    frintp z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <32 x double>, <32 x double>* %a
  %res = call <32 x double> @llvm.ceil.v32f64(<32 x double> %op)
  store <32 x double> %res, <32 x double>* %a
  ret void
}

;
; FLOOR -> FRINTM
;

; Don't use SVE for 64-bit vectors.
define <4 x half> @frintm_v4f16(<4 x half> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintm_v4f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintm v0.4h, v0.4h
; CHECK-NEXT:    ret
  %res = call <4 x half> @llvm.floor.v4f16(<4 x half> %op)
  ret <4 x half> %res
}

; Don't use SVE for 128-bit vectors.
define <8 x half> @frintm_v8f16(<8 x half> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintm_v8f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintm v0.8h, v0.8h
; CHECK-NEXT:    ret
  %res = call <8 x half> @llvm.floor.v8f16(<8 x half> %op)
  ret <8 x half> %res
}

define void @frintm_v16f16(<16 x half>* %a) vscale_range(2,0) #0 {
; CHECK-LABEL: frintm_v16f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl16
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    frintm z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <16 x half>, <16 x half>* %a
  %res = call <16 x half> @llvm.floor.v16f16(<16 x half> %op)
  store <16 x half> %res, <16 x half>* %a
  ret void
}

define void @frintm_v32f16(<32 x half>* %a) #0 {
; VBITS_GE_256-LABEL: frintm_v32f16:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    mov x8, #16
; VBITS_GE_256-NEXT:    ptrue p0.h, vl16
; 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:    frintm z0.h, p0/m, z0.h
; VBITS_GE_256-NEXT:    frintm z1.h, p0/m, z1.h
; VBITS_GE_256-NEXT:    st1h { z0.h }, p0, [x0, x8, lsl #1]
; VBITS_GE_256-NEXT:    st1h { z1.h }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: frintm_v32f16:
; 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:    frintm z0.h, p0/m, z0.h
; VBITS_GE_512-NEXT:    st1h { z0.h }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op = load <32 x half>, <32 x half>* %a
  %res = call <32 x half> @llvm.floor.v32f16(<32 x half> %op)
  store <32 x half> %res, <32 x half>* %a
  ret void
}

define void @frintm_v64f16(<64 x half>* %a) vscale_range(8,0) #0 {
; CHECK-LABEL: frintm_v64f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl64
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    frintm z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <64 x half>, <64 x half>* %a
  %res = call <64 x half> @llvm.floor.v64f16(<64 x half> %op)
  store <64 x half> %res, <64 x half>* %a
  ret void
}

define void @frintm_v128f16(<128 x half>* %a) vscale_range(16,0) #0 {
; CHECK-LABEL: frintm_v128f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl128
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    frintm z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <128 x half>, <128 x half>* %a
  %res = call <128 x half> @llvm.floor.v128f16(<128 x half> %op)
  store <128 x half> %res, <128 x half>* %a
  ret void
}

; Don't use SVE for 64-bit vectors.
define <2 x float> @frintm_v2f32(<2 x float> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintm_v2f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintm v0.2s, v0.2s
; CHECK-NEXT:    ret
  %res = call <2 x float> @llvm.floor.v2f32(<2 x float> %op)
  ret <2 x float> %res
}

; Don't use SVE for 128-bit vectors.
define <4 x float> @frintm_v4f32(<4 x float> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintm_v4f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintm v0.4s, v0.4s
; CHECK-NEXT:    ret
  %res = call <4 x float> @llvm.floor.v4f32(<4 x float> %op)
  ret <4 x float> %res
}

define void @frintm_v8f32(<8 x float>* %a) vscale_range(2,0) #0 {
; CHECK-LABEL: frintm_v8f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl8
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    frintm z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <8 x float>, <8 x float>* %a
  %res = call <8 x float> @llvm.floor.v8f32(<8 x float> %op)
  store <8 x float> %res, <8 x float>* %a
  ret void
}

define void @frintm_v16f32(<16 x float>* %a) #0 {
; VBITS_GE_256-LABEL: frintm_v16f32:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    mov x8, #8
; VBITS_GE_256-NEXT:    ptrue p0.s, vl8
; 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:    frintm z0.s, p0/m, z0.s
; VBITS_GE_256-NEXT:    frintm z1.s, p0/m, z1.s
; VBITS_GE_256-NEXT:    st1w { z0.s }, p0, [x0, x8, lsl #2]
; VBITS_GE_256-NEXT:    st1w { z1.s }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: frintm_v16f32:
; 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:    frintm z0.s, p0/m, z0.s
; VBITS_GE_512-NEXT:    st1w { z0.s }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op = load <16 x float>, <16 x float>* %a
  %res = call <16 x float> @llvm.floor.v16f32(<16 x float> %op)
  store <16 x float> %res, <16 x float>* %a
  ret void
}

define void @frintm_v32f32(<32 x float>* %a) vscale_range(8,0) #0 {
; CHECK-LABEL: frintm_v32f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl32
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    frintm z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <32 x float>, <32 x float>* %a
  %res = call <32 x float> @llvm.floor.v32f32(<32 x float> %op)
  store <32 x float> %res, <32 x float>* %a
  ret void
}

define void @frintm_v64f32(<64 x float>* %a) vscale_range(16,0) #0 {
; CHECK-LABEL: frintm_v64f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl64
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    frintm z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <64 x float>, <64 x float>* %a
  %res = call <64 x float> @llvm.floor.v64f32(<64 x float> %op)
  store <64 x float> %res, <64 x float>* %a
  ret void
}

; Don't use SVE for 64-bit vectors.
define <1 x double> @frintm_v1f64(<1 x double> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintm_v1f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintm d0, d0
; CHECK-NEXT:    ret
  %res = call <1 x double> @llvm.floor.v1f64(<1 x double> %op)
  ret <1 x double> %res
}

; Don't use SVE for 128-bit vectors.
define <2 x double> @frintm_v2f64(<2 x double> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintm_v2f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintm v0.2d, v0.2d
; CHECK-NEXT:    ret
  %res = call <2 x double> @llvm.floor.v2f64(<2 x double> %op)
  ret <2 x double> %res
}

define void @frintm_v4f64(<4 x double>* %a) vscale_range(2,0) #0 {
; CHECK-LABEL: frintm_v4f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    frintm z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <4 x double>, <4 x double>* %a
  %res = call <4 x double> @llvm.floor.v4f64(<4 x double> %op)
  store <4 x double> %res, <4 x double>* %a
  ret void
}

define void @frintm_v8f64(<8 x double>* %a) #0 {
; VBITS_GE_256-LABEL: frintm_v8f64:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    mov x8, #4
; VBITS_GE_256-NEXT:    ptrue p0.d, vl4
; 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:    frintm z0.d, p0/m, z0.d
; VBITS_GE_256-NEXT:    frintm z1.d, p0/m, z1.d
; VBITS_GE_256-NEXT:    st1d { z0.d }, p0, [x0, x8, lsl #3]
; VBITS_GE_256-NEXT:    st1d { z1.d }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: frintm_v8f64:
; 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:    frintm z0.d, p0/m, z0.d
; VBITS_GE_512-NEXT:    st1d { z0.d }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op = load <8 x double>, <8 x double>* %a
  %res = call <8 x double> @llvm.floor.v8f64(<8 x double> %op)
  store <8 x double> %res, <8 x double>* %a
  ret void
}

define void @frintm_v16f64(<16 x double>* %a) vscale_range(8,0) #0 {
; CHECK-LABEL: frintm_v16f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl16
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    frintm z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <16 x double>, <16 x double>* %a
  %res = call <16 x double> @llvm.floor.v16f64(<16 x double> %op)
  store <16 x double> %res, <16 x double>* %a
  ret void
}

define void @frintm_v32f64(<32 x double>* %a) vscale_range(16,0) #0 {
; CHECK-LABEL: frintm_v32f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl32
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    frintm z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <32 x double>, <32 x double>* %a
  %res = call <32 x double> @llvm.floor.v32f64(<32 x double> %op)
  store <32 x double> %res, <32 x double>* %a
  ret void
}

;
; FNEARBYINT -> FRINTI
;

; Don't use SVE for 64-bit vectors.
define <4 x half> @frinti_v4f16(<4 x half> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frinti_v4f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frinti v0.4h, v0.4h
; CHECK-NEXT:    ret
  %res = call <4 x half> @llvm.nearbyint.v4f16(<4 x half> %op)
  ret <4 x half> %res
}

; Don't use SVE for 128-bit vectors.
define <8 x half> @frinti_v8f16(<8 x half> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frinti_v8f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frinti v0.8h, v0.8h
; CHECK-NEXT:    ret
  %res = call <8 x half> @llvm.nearbyint.v8f16(<8 x half> %op)
  ret <8 x half> %res
}

define void @frinti_v16f16(<16 x half>* %a) vscale_range(2,0) #0 {
; CHECK-LABEL: frinti_v16f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl16
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    frinti z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <16 x half>, <16 x half>* %a
  %res = call <16 x half> @llvm.nearbyint.v16f16(<16 x half> %op)
  store <16 x half> %res, <16 x half>* %a
  ret void
}

define void @frinti_v32f16(<32 x half>* %a) #0 {
; VBITS_GE_256-LABEL: frinti_v32f16:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    mov x8, #16
; VBITS_GE_256-NEXT:    ptrue p0.h, vl16
; 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:    frinti z0.h, p0/m, z0.h
; VBITS_GE_256-NEXT:    frinti z1.h, p0/m, z1.h
; VBITS_GE_256-NEXT:    st1h { z0.h }, p0, [x0, x8, lsl #1]
; VBITS_GE_256-NEXT:    st1h { z1.h }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: frinti_v32f16:
; 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:    frinti z0.h, p0/m, z0.h
; VBITS_GE_512-NEXT:    st1h { z0.h }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op = load <32 x half>, <32 x half>* %a
  %res = call <32 x half> @llvm.nearbyint.v32f16(<32 x half> %op)
  store <32 x half> %res, <32 x half>* %a
  ret void
}

define void @frinti_v64f16(<64 x half>* %a) vscale_range(8,0) #0 {
; CHECK-LABEL: frinti_v64f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl64
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    frinti z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <64 x half>, <64 x half>* %a
  %res = call <64 x half> @llvm.nearbyint.v64f16(<64 x half> %op)
  store <64 x half> %res, <64 x half>* %a
  ret void
}

define void @frinti_v128f16(<128 x half>* %a) vscale_range(16,0) #0 {
; CHECK-LABEL: frinti_v128f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl128
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    frinti z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <128 x half>, <128 x half>* %a
  %res = call <128 x half> @llvm.nearbyint.v128f16(<128 x half> %op)
  store <128 x half> %res, <128 x half>* %a
  ret void
}

; Don't use SVE for 64-bit vectors.
define <2 x float> @frinti_v2f32(<2 x float> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frinti_v2f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frinti v0.2s, v0.2s
; CHECK-NEXT:    ret
  %res = call <2 x float> @llvm.nearbyint.v2f32(<2 x float> %op)
  ret <2 x float> %res
}

; Don't use SVE for 128-bit vectors.
define <4 x float> @frinti_v4f32(<4 x float> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frinti_v4f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frinti v0.4s, v0.4s
; CHECK-NEXT:    ret
  %res = call <4 x float> @llvm.nearbyint.v4f32(<4 x float> %op)
  ret <4 x float> %res
}

define void @frinti_v8f32(<8 x float>* %a) vscale_range(2,0) #0 {
; CHECK-LABEL: frinti_v8f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl8
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    frinti z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <8 x float>, <8 x float>* %a
  %res = call <8 x float> @llvm.nearbyint.v8f32(<8 x float> %op)
  store <8 x float> %res, <8 x float>* %a
  ret void
}

define void @frinti_v16f32(<16 x float>* %a) #0 {
; VBITS_GE_256-LABEL: frinti_v16f32:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    mov x8, #8
; VBITS_GE_256-NEXT:    ptrue p0.s, vl8
; 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:    frinti z0.s, p0/m, z0.s
; VBITS_GE_256-NEXT:    frinti z1.s, p0/m, z1.s
; VBITS_GE_256-NEXT:    st1w { z0.s }, p0, [x0, x8, lsl #2]
; VBITS_GE_256-NEXT:    st1w { z1.s }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: frinti_v16f32:
; 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:    frinti z0.s, p0/m, z0.s
; VBITS_GE_512-NEXT:    st1w { z0.s }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op = load <16 x float>, <16 x float>* %a
  %res = call <16 x float> @llvm.nearbyint.v16f32(<16 x float> %op)
  store <16 x float> %res, <16 x float>* %a
  ret void
}

define void @frinti_v32f32(<32 x float>* %a) vscale_range(8,0) #0 {
; CHECK-LABEL: frinti_v32f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl32
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    frinti z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <32 x float>, <32 x float>* %a
  %res = call <32 x float> @llvm.nearbyint.v32f32(<32 x float> %op)
  store <32 x float> %res, <32 x float>* %a
  ret void
}

define void @frinti_v64f32(<64 x float>* %a) vscale_range(16,0) #0 {
; CHECK-LABEL: frinti_v64f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl64
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    frinti z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <64 x float>, <64 x float>* %a
  %res = call <64 x float> @llvm.nearbyint.v64f32(<64 x float> %op)
  store <64 x float> %res, <64 x float>* %a
  ret void
}

; Don't use SVE for 64-bit vectors.
define <1 x double> @frinti_v1f64(<1 x double> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frinti_v1f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frinti d0, d0
; CHECK-NEXT:    ret
  %res = call <1 x double> @llvm.nearbyint.v1f64(<1 x double> %op)
  ret <1 x double> %res
}

; Don't use SVE for 128-bit vectors.
define <2 x double> @frinti_v2f64(<2 x double> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frinti_v2f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frinti v0.2d, v0.2d
; CHECK-NEXT:    ret
  %res = call <2 x double> @llvm.nearbyint.v2f64(<2 x double> %op)
  ret <2 x double> %res
}

define void @frinti_v4f64(<4 x double>* %a) vscale_range(2,0) #0 {
; CHECK-LABEL: frinti_v4f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    frinti z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <4 x double>, <4 x double>* %a
  %res = call <4 x double> @llvm.nearbyint.v4f64(<4 x double> %op)
  store <4 x double> %res, <4 x double>* %a
  ret void
}

define void @frinti_v8f64(<8 x double>* %a) #0 {
; VBITS_GE_256-LABEL: frinti_v8f64:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    mov x8, #4
; VBITS_GE_256-NEXT:    ptrue p0.d, vl4
; 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:    frinti z0.d, p0/m, z0.d
; VBITS_GE_256-NEXT:    frinti z1.d, p0/m, z1.d
; VBITS_GE_256-NEXT:    st1d { z0.d }, p0, [x0, x8, lsl #3]
; VBITS_GE_256-NEXT:    st1d { z1.d }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: frinti_v8f64:
; 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:    frinti z0.d, p0/m, z0.d
; VBITS_GE_512-NEXT:    st1d { z0.d }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op = load <8 x double>, <8 x double>* %a
  %res = call <8 x double> @llvm.nearbyint.v8f64(<8 x double> %op)
  store <8 x double> %res, <8 x double>* %a
  ret void
}

define void @frinti_v16f64(<16 x double>* %a) vscale_range(8,0) #0 {
; CHECK-LABEL: frinti_v16f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl16
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    frinti z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <16 x double>, <16 x double>* %a
  %res = call <16 x double> @llvm.nearbyint.v16f64(<16 x double> %op)
  store <16 x double> %res, <16 x double>* %a
  ret void
}

define void @frinti_v32f64(<32 x double>* %a) vscale_range(16,0) #0 {
; CHECK-LABEL: frinti_v32f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl32
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    frinti z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <32 x double>, <32 x double>* %a
  %res = call <32 x double> @llvm.nearbyint.v32f64(<32 x double> %op)
  store <32 x double> %res, <32 x double>* %a
  ret void
}

;
; RINT -> FRINTX
;

; Don't use SVE for 64-bit vectors.
define <4 x half> @frintx_v4f16(<4 x half> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintx_v4f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintx v0.4h, v0.4h
; CHECK-NEXT:    ret
  %res = call <4 x half> @llvm.rint.v4f16(<4 x half> %op)
  ret <4 x half> %res
}

; Don't use SVE for 128-bit vectors.
define <8 x half> @frintx_v8f16(<8 x half> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintx_v8f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintx v0.8h, v0.8h
; CHECK-NEXT:    ret
  %res = call <8 x half> @llvm.rint.v8f16(<8 x half> %op)
  ret <8 x half> %res
}

define void @frintx_v16f16(<16 x half>* %a) vscale_range(2,0) #0 {
; CHECK-LABEL: frintx_v16f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl16
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    frintx z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <16 x half>, <16 x half>* %a
  %res = call <16 x half> @llvm.rint.v16f16(<16 x half> %op)
  store <16 x half> %res, <16 x half>* %a
  ret void
}

define void @frintx_v32f16(<32 x half>* %a) #0 {
; VBITS_GE_256-LABEL: frintx_v32f16:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    mov x8, #16
; VBITS_GE_256-NEXT:    ptrue p0.h, vl16
; 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:    frintx z0.h, p0/m, z0.h
; VBITS_GE_256-NEXT:    frintx z1.h, p0/m, z1.h
; VBITS_GE_256-NEXT:    st1h { z0.h }, p0, [x0, x8, lsl #1]
; VBITS_GE_256-NEXT:    st1h { z1.h }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: frintx_v32f16:
; 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:    frintx z0.h, p0/m, z0.h
; VBITS_GE_512-NEXT:    st1h { z0.h }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op = load <32 x half>, <32 x half>* %a
  %res = call <32 x half> @llvm.rint.v32f16(<32 x half> %op)
  store <32 x half> %res, <32 x half>* %a
  ret void
}

define void @frintx_v64f16(<64 x half>* %a) vscale_range(8,0) #0 {
; CHECK-LABEL: frintx_v64f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl64
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    frintx z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <64 x half>, <64 x half>* %a
  %res = call <64 x half> @llvm.rint.v64f16(<64 x half> %op)
  store <64 x half> %res, <64 x half>* %a
  ret void
}

define void @frintx_v128f16(<128 x half>* %a) vscale_range(16,0) #0 {
; CHECK-LABEL: frintx_v128f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl128
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    frintx z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <128 x half>, <128 x half>* %a
  %res = call <128 x half> @llvm.rint.v128f16(<128 x half> %op)
  store <128 x half> %res, <128 x half>* %a
  ret void
}

; Don't use SVE for 64-bit vectors.
define <2 x float> @frintx_v2f32(<2 x float> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintx_v2f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintx v0.2s, v0.2s
; CHECK-NEXT:    ret
  %res = call <2 x float> @llvm.rint.v2f32(<2 x float> %op)
  ret <2 x float> %res
}

; Don't use SVE for 128-bit vectors.
define <4 x float> @frintx_v4f32(<4 x float> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintx_v4f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintx v0.4s, v0.4s
; CHECK-NEXT:    ret
  %res = call <4 x float> @llvm.rint.v4f32(<4 x float> %op)
  ret <4 x float> %res
}

define void @frintx_v8f32(<8 x float>* %a) vscale_range(2,0) #0 {
; CHECK-LABEL: frintx_v8f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl8
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    frintx z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <8 x float>, <8 x float>* %a
  %res = call <8 x float> @llvm.rint.v8f32(<8 x float> %op)
  store <8 x float> %res, <8 x float>* %a
  ret void
}

define void @frintx_v16f32(<16 x float>* %a) #0 {
; VBITS_GE_256-LABEL: frintx_v16f32:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    mov x8, #8
; VBITS_GE_256-NEXT:    ptrue p0.s, vl8
; 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:    frintx z0.s, p0/m, z0.s
; VBITS_GE_256-NEXT:    frintx z1.s, p0/m, z1.s
; VBITS_GE_256-NEXT:    st1w { z0.s }, p0, [x0, x8, lsl #2]
; VBITS_GE_256-NEXT:    st1w { z1.s }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: frintx_v16f32:
; 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:    frintx z0.s, p0/m, z0.s
; VBITS_GE_512-NEXT:    st1w { z0.s }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op = load <16 x float>, <16 x float>* %a
  %res = call <16 x float> @llvm.rint.v16f32(<16 x float> %op)
  store <16 x float> %res, <16 x float>* %a
  ret void
}

define void @frintx_v32f32(<32 x float>* %a) vscale_range(8,0) #0 {
; CHECK-LABEL: frintx_v32f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl32
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    frintx z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <32 x float>, <32 x float>* %a
  %res = call <32 x float> @llvm.rint.v32f32(<32 x float> %op)
  store <32 x float> %res, <32 x float>* %a
  ret void
}

define void @frintx_v64f32(<64 x float>* %a) vscale_range(16,0) #0 {
; CHECK-LABEL: frintx_v64f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl64
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    frintx z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <64 x float>, <64 x float>* %a
  %res = call <64 x float> @llvm.rint.v64f32(<64 x float> %op)
  store <64 x float> %res, <64 x float>* %a
  ret void
}

; Don't use SVE for 64-bit vectors.
define <1 x double> @frintx_v1f64(<1 x double> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintx_v1f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintx d0, d0
; CHECK-NEXT:    ret
  %res = call <1 x double> @llvm.rint.v1f64(<1 x double> %op)
  ret <1 x double> %res
}

; Don't use SVE for 128-bit vectors.
define <2 x double> @frintx_v2f64(<2 x double> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintx_v2f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintx v0.2d, v0.2d
; CHECK-NEXT:    ret
  %res = call <2 x double> @llvm.rint.v2f64(<2 x double> %op)
  ret <2 x double> %res
}

define void @frintx_v4f64(<4 x double>* %a) vscale_range(2,0) #0 {
; CHECK-LABEL: frintx_v4f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    frintx z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <4 x double>, <4 x double>* %a
  %res = call <4 x double> @llvm.rint.v4f64(<4 x double> %op)
  store <4 x double> %res, <4 x double>* %a
  ret void
}

define void @frintx_v8f64(<8 x double>* %a) #0 {
; VBITS_GE_256-LABEL: frintx_v8f64:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    mov x8, #4
; VBITS_GE_256-NEXT:    ptrue p0.d, vl4
; 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:    frintx z0.d, p0/m, z0.d
; VBITS_GE_256-NEXT:    frintx z1.d, p0/m, z1.d
; VBITS_GE_256-NEXT:    st1d { z0.d }, p0, [x0, x8, lsl #3]
; VBITS_GE_256-NEXT:    st1d { z1.d }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: frintx_v8f64:
; 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:    frintx z0.d, p0/m, z0.d
; VBITS_GE_512-NEXT:    st1d { z0.d }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op = load <8 x double>, <8 x double>* %a
  %res = call <8 x double> @llvm.rint.v8f64(<8 x double> %op)
  store <8 x double> %res, <8 x double>* %a
  ret void
}

define void @frintx_v16f64(<16 x double>* %a) vscale_range(8,0) #0 {
; CHECK-LABEL: frintx_v16f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl16
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    frintx z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <16 x double>, <16 x double>* %a
  %res = call <16 x double> @llvm.rint.v16f64(<16 x double> %op)
  store <16 x double> %res, <16 x double>* %a
  ret void
}

define void @frintx_v32f64(<32 x double>* %a) vscale_range(16,0) #0 {
; CHECK-LABEL: frintx_v32f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl32
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    frintx z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <32 x double>, <32 x double>* %a
  %res = call <32 x double> @llvm.rint.v32f64(<32 x double> %op)
  store <32 x double> %res, <32 x double>* %a
  ret void
}

;
; ROUND -> FRINTA
;

; Don't use SVE for 64-bit vectors.
define <4 x half> @frinta_v4f16(<4 x half> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frinta_v4f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frinta v0.4h, v0.4h
; CHECK-NEXT:    ret
  %res = call <4 x half> @llvm.round.v4f16(<4 x half> %op)
  ret <4 x half> %res
}

; Don't use SVE for 128-bit vectors.
define <8 x half> @frinta_v8f16(<8 x half> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frinta_v8f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frinta v0.8h, v0.8h
; CHECK-NEXT:    ret
  %res = call <8 x half> @llvm.round.v8f16(<8 x half> %op)
  ret <8 x half> %res
}

define void @frinta_v16f16(<16 x half>* %a) vscale_range(2,0) #0 {
; CHECK-LABEL: frinta_v16f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl16
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    frinta z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <16 x half>, <16 x half>* %a
  %res = call <16 x half> @llvm.round.v16f16(<16 x half> %op)
  store <16 x half> %res, <16 x half>* %a
  ret void
}

define void @frinta_v32f16(<32 x half>* %a) #0 {
; VBITS_GE_256-LABEL: frinta_v32f16:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    mov x8, #16
; VBITS_GE_256-NEXT:    ptrue p0.h, vl16
; 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:    frinta z0.h, p0/m, z0.h
; VBITS_GE_256-NEXT:    frinta z1.h, p0/m, z1.h
; VBITS_GE_256-NEXT:    st1h { z0.h }, p0, [x0, x8, lsl #1]
; VBITS_GE_256-NEXT:    st1h { z1.h }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: frinta_v32f16:
; 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:    frinta z0.h, p0/m, z0.h
; VBITS_GE_512-NEXT:    st1h { z0.h }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op = load <32 x half>, <32 x half>* %a
  %res = call <32 x half> @llvm.round.v32f16(<32 x half> %op)
  store <32 x half> %res, <32 x half>* %a
  ret void
}

define void @frinta_v64f16(<64 x half>* %a) vscale_range(8,0) #0 {
; CHECK-LABEL: frinta_v64f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl64
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    frinta z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <64 x half>, <64 x half>* %a
  %res = call <64 x half> @llvm.round.v64f16(<64 x half> %op)
  store <64 x half> %res, <64 x half>* %a
  ret void
}

define void @frinta_v128f16(<128 x half>* %a) vscale_range(16,0) #0 {
; CHECK-LABEL: frinta_v128f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl128
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    frinta z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <128 x half>, <128 x half>* %a
  %res = call <128 x half> @llvm.round.v128f16(<128 x half> %op)
  store <128 x half> %res, <128 x half>* %a
  ret void
}

; Don't use SVE for 64-bit vectors.
define <2 x float> @frinta_v2f32(<2 x float> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frinta_v2f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frinta v0.2s, v0.2s
; CHECK-NEXT:    ret
  %res = call <2 x float> @llvm.round.v2f32(<2 x float> %op)
  ret <2 x float> %res
}

; Don't use SVE for 128-bit vectors.
define <4 x float> @frinta_v4f32(<4 x float> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frinta_v4f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frinta v0.4s, v0.4s
; CHECK-NEXT:    ret
  %res = call <4 x float> @llvm.round.v4f32(<4 x float> %op)
  ret <4 x float> %res
}

define void @frinta_v8f32(<8 x float>* %a) vscale_range(2,0) #0 {
; CHECK-LABEL: frinta_v8f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl8
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    frinta z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <8 x float>, <8 x float>* %a
  %res = call <8 x float> @llvm.round.v8f32(<8 x float> %op)
  store <8 x float> %res, <8 x float>* %a
  ret void
}

define void @frinta_v16f32(<16 x float>* %a) #0 {
; VBITS_GE_256-LABEL: frinta_v16f32:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    mov x8, #8
; VBITS_GE_256-NEXT:    ptrue p0.s, vl8
; 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:    frinta z0.s, p0/m, z0.s
; VBITS_GE_256-NEXT:    frinta z1.s, p0/m, z1.s
; VBITS_GE_256-NEXT:    st1w { z0.s }, p0, [x0, x8, lsl #2]
; VBITS_GE_256-NEXT:    st1w { z1.s }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: frinta_v16f32:
; 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:    frinta z0.s, p0/m, z0.s
; VBITS_GE_512-NEXT:    st1w { z0.s }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op = load <16 x float>, <16 x float>* %a
  %res = call <16 x float> @llvm.round.v16f32(<16 x float> %op)
  store <16 x float> %res, <16 x float>* %a
  ret void
}

define void @frinta_v32f32(<32 x float>* %a) vscale_range(8,0) #0 {
; CHECK-LABEL: frinta_v32f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl32
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    frinta z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <32 x float>, <32 x float>* %a
  %res = call <32 x float> @llvm.round.v32f32(<32 x float> %op)
  store <32 x float> %res, <32 x float>* %a
  ret void
}

define void @frinta_v64f32(<64 x float>* %a) vscale_range(16,0) #0 {
; CHECK-LABEL: frinta_v64f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl64
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    frinta z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <64 x float>, <64 x float>* %a
  %res = call <64 x float> @llvm.round.v64f32(<64 x float> %op)
  store <64 x float> %res, <64 x float>* %a
  ret void
}

; Don't use SVE for 64-bit vectors.
define <1 x double> @frinta_v1f64(<1 x double> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frinta_v1f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frinta d0, d0
; CHECK-NEXT:    ret
  %res = call <1 x double> @llvm.round.v1f64(<1 x double> %op)
  ret <1 x double> %res
}

; Don't use SVE for 128-bit vectors.
define <2 x double> @frinta_v2f64(<2 x double> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frinta_v2f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frinta v0.2d, v0.2d
; CHECK-NEXT:    ret
  %res = call <2 x double> @llvm.round.v2f64(<2 x double> %op)
  ret <2 x double> %res
}

define void @frinta_v4f64(<4 x double>* %a) vscale_range(2,0) #0 {
; CHECK-LABEL: frinta_v4f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    frinta z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <4 x double>, <4 x double>* %a
  %res = call <4 x double> @llvm.round.v4f64(<4 x double> %op)
  store <4 x double> %res, <4 x double>* %a
  ret void
}

define void @frinta_v8f64(<8 x double>* %a) #0 {
; VBITS_GE_256-LABEL: frinta_v8f64:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    mov x8, #4
; VBITS_GE_256-NEXT:    ptrue p0.d, vl4
; 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:    frinta z0.d, p0/m, z0.d
; VBITS_GE_256-NEXT:    frinta z1.d, p0/m, z1.d
; VBITS_GE_256-NEXT:    st1d { z0.d }, p0, [x0, x8, lsl #3]
; VBITS_GE_256-NEXT:    st1d { z1.d }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: frinta_v8f64:
; 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:    frinta z0.d, p0/m, z0.d
; VBITS_GE_512-NEXT:    st1d { z0.d }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op = load <8 x double>, <8 x double>* %a
  %res = call <8 x double> @llvm.round.v8f64(<8 x double> %op)
  store <8 x double> %res, <8 x double>* %a
  ret void
}

define void @frinta_v16f64(<16 x double>* %a) vscale_range(8,0) #0 {
; CHECK-LABEL: frinta_v16f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl16
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    frinta z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <16 x double>, <16 x double>* %a
  %res = call <16 x double> @llvm.round.v16f64(<16 x double> %op)
  store <16 x double> %res, <16 x double>* %a
  ret void
}

define void @frinta_v32f64(<32 x double>* %a) vscale_range(16,0) #0 {
; CHECK-LABEL: frinta_v32f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl32
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    frinta z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <32 x double>, <32 x double>* %a
  %res = call <32 x double> @llvm.round.v32f64(<32 x double> %op)
  store <32 x double> %res, <32 x double>* %a
  ret void
}

;
; ROUNDEVEN -> FRINTN
;

; Don't use SVE for 64-bit vectors.
define <4 x half> @frintn_v4f16(<4 x half> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintn_v4f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintn v0.4h, v0.4h
; CHECK-NEXT:    ret
  %res = call <4 x half> @llvm.roundeven.v4f16(<4 x half> %op)
  ret <4 x half> %res
}

; Don't use SVE for 128-bit vectors.
define <8 x half> @frintn_v8f16(<8 x half> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintn_v8f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintn v0.8h, v0.8h
; CHECK-NEXT:    ret
  %res = call <8 x half> @llvm.roundeven.v8f16(<8 x half> %op)
  ret <8 x half> %res
}

define void @frintn_v16f16(<16 x half>* %a) vscale_range(2,0) #0 {
; CHECK-LABEL: frintn_v16f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl16
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    frintn z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <16 x half>, <16 x half>* %a
  %res = call <16 x half> @llvm.roundeven.v16f16(<16 x half> %op)
  store <16 x half> %res, <16 x half>* %a
  ret void
}

define void @frintn_v32f16(<32 x half>* %a) #0 {
; VBITS_GE_256-LABEL: frintn_v32f16:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    mov x8, #16
; VBITS_GE_256-NEXT:    ptrue p0.h, vl16
; 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:    frintn z0.h, p0/m, z0.h
; VBITS_GE_256-NEXT:    frintn z1.h, p0/m, z1.h
; VBITS_GE_256-NEXT:    st1h { z0.h }, p0, [x0, x8, lsl #1]
; VBITS_GE_256-NEXT:    st1h { z1.h }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: frintn_v32f16:
; 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:    frintn z0.h, p0/m, z0.h
; VBITS_GE_512-NEXT:    st1h { z0.h }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op = load <32 x half>, <32 x half>* %a
  %res = call <32 x half> @llvm.roundeven.v32f16(<32 x half> %op)
  store <32 x half> %res, <32 x half>* %a
  ret void
}

define void @frintn_v64f16(<64 x half>* %a) vscale_range(8,0) #0 {
; CHECK-LABEL: frintn_v64f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl64
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    frintn z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <64 x half>, <64 x half>* %a
  %res = call <64 x half> @llvm.roundeven.v64f16(<64 x half> %op)
  store <64 x half> %res, <64 x half>* %a
  ret void
}

define void @frintn_v128f16(<128 x half>* %a) vscale_range(16,0) #0 {
; CHECK-LABEL: frintn_v128f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl128
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    frintn z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <128 x half>, <128 x half>* %a
  %res = call <128 x half> @llvm.roundeven.v128f16(<128 x half> %op)
  store <128 x half> %res, <128 x half>* %a
  ret void
}

; Don't use SVE for 64-bit vectors.
define <2 x float> @frintn_v2f32(<2 x float> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintn_v2f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintn v0.2s, v0.2s
; CHECK-NEXT:    ret
  %res = call <2 x float> @llvm.roundeven.v2f32(<2 x float> %op)
  ret <2 x float> %res
}

; Don't use SVE for 128-bit vectors.
define <4 x float> @frintn_v4f32(<4 x float> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintn_v4f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintn v0.4s, v0.4s
; CHECK-NEXT:    ret
  %res = call <4 x float> @llvm.roundeven.v4f32(<4 x float> %op)
  ret <4 x float> %res
}

define void @frintn_v8f32(<8 x float>* %a) vscale_range(2,0) #0 {
; CHECK-LABEL: frintn_v8f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl8
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    frintn z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <8 x float>, <8 x float>* %a
  %res = call <8 x float> @llvm.roundeven.v8f32(<8 x float> %op)
  store <8 x float> %res, <8 x float>* %a
  ret void
}

define void @frintn_v16f32(<16 x float>* %a) #0 {
; VBITS_GE_256-LABEL: frintn_v16f32:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    mov x8, #8
; VBITS_GE_256-NEXT:    ptrue p0.s, vl8
; 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:    frintn z0.s, p0/m, z0.s
; VBITS_GE_256-NEXT:    frintn z1.s, p0/m, z1.s
; VBITS_GE_256-NEXT:    st1w { z0.s }, p0, [x0, x8, lsl #2]
; VBITS_GE_256-NEXT:    st1w { z1.s }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: frintn_v16f32:
; 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:    frintn z0.s, p0/m, z0.s
; VBITS_GE_512-NEXT:    st1w { z0.s }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op = load <16 x float>, <16 x float>* %a
  %res = call <16 x float> @llvm.roundeven.v16f32(<16 x float> %op)
  store <16 x float> %res, <16 x float>* %a
  ret void
}

define void @frintn_v32f32(<32 x float>* %a) vscale_range(8,0) #0 {
; CHECK-LABEL: frintn_v32f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl32
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    frintn z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <32 x float>, <32 x float>* %a
  %res = call <32 x float> @llvm.roundeven.v32f32(<32 x float> %op)
  store <32 x float> %res, <32 x float>* %a
  ret void
}

define void @frintn_v64f32(<64 x float>* %a) vscale_range(16,0) #0 {
; CHECK-LABEL: frintn_v64f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl64
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    frintn z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <64 x float>, <64 x float>* %a
  %res = call <64 x float> @llvm.roundeven.v64f32(<64 x float> %op)
  store <64 x float> %res, <64 x float>* %a
  ret void
}

; Don't use SVE for 64-bit vectors.
define <1 x double> @frintn_v1f64(<1 x double> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintn_v1f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintn d0, d0
; CHECK-NEXT:    ret
  %res = call <1 x double> @llvm.roundeven.v1f64(<1 x double> %op)
  ret <1 x double> %res
}

; Don't use SVE for 128-bit vectors.
define <2 x double> @frintn_v2f64(<2 x double> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintn_v2f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintn v0.2d, v0.2d
; CHECK-NEXT:    ret
  %res = call <2 x double> @llvm.roundeven.v2f64(<2 x double> %op)
  ret <2 x double> %res
}

define void @frintn_v4f64(<4 x double>* %a) vscale_range(2,0) #0 {
; CHECK-LABEL: frintn_v4f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    frintn z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <4 x double>, <4 x double>* %a
  %res = call <4 x double> @llvm.roundeven.v4f64(<4 x double> %op)
  store <4 x double> %res, <4 x double>* %a
  ret void
}

define void @frintn_v8f64(<8 x double>* %a) #0 {
; VBITS_GE_256-LABEL: frintn_v8f64:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    mov x8, #4
; VBITS_GE_256-NEXT:    ptrue p0.d, vl4
; 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:    frintn z0.d, p0/m, z0.d
; VBITS_GE_256-NEXT:    frintn z1.d, p0/m, z1.d
; VBITS_GE_256-NEXT:    st1d { z0.d }, p0, [x0, x8, lsl #3]
; VBITS_GE_256-NEXT:    st1d { z1.d }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: frintn_v8f64:
; 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:    frintn z0.d, p0/m, z0.d
; VBITS_GE_512-NEXT:    st1d { z0.d }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op = load <8 x double>, <8 x double>* %a
  %res = call <8 x double> @llvm.roundeven.v8f64(<8 x double> %op)
  store <8 x double> %res, <8 x double>* %a
  ret void
}

define void @frintn_v16f64(<16 x double>* %a) vscale_range(8,0) #0 {
; CHECK-LABEL: frintn_v16f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl16
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    frintn z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <16 x double>, <16 x double>* %a
  %res = call <16 x double> @llvm.roundeven.v16f64(<16 x double> %op)
  store <16 x double> %res, <16 x double>* %a
  ret void
}

define void @frintn_v32f64(<32 x double>* %a) vscale_range(16,0) #0 {
; CHECK-LABEL: frintn_v32f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl32
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    frintn z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <32 x double>, <32 x double>* %a
  %res = call <32 x double> @llvm.roundeven.v32f64(<32 x double> %op)
  store <32 x double> %res, <32 x double>* %a
  ret void
}

;
; TRUNC -> FRINTZ
;

; Don't use SVE for 64-bit vectors.
define <4 x half> @frintz_v4f16(<4 x half> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintz_v4f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintz v0.4h, v0.4h
; CHECK-NEXT:    ret
  %res = call <4 x half> @llvm.trunc.v4f16(<4 x half> %op)
  ret <4 x half> %res
}

; Don't use SVE for 128-bit vectors.
define <8 x half> @frintz_v8f16(<8 x half> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintz_v8f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintz v0.8h, v0.8h
; CHECK-NEXT:    ret
  %res = call <8 x half> @llvm.trunc.v8f16(<8 x half> %op)
  ret <8 x half> %res
}

define void @frintz_v16f16(<16 x half>* %a) vscale_range(2,0) #0 {
; CHECK-LABEL: frintz_v16f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl16
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    frintz z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <16 x half>, <16 x half>* %a
  %res = call <16 x half> @llvm.trunc.v16f16(<16 x half> %op)
  store <16 x half> %res, <16 x half>* %a
  ret void
}

define void @frintz_v32f16(<32 x half>* %a) #0 {
; VBITS_GE_256-LABEL: frintz_v32f16:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    mov x8, #16
; VBITS_GE_256-NEXT:    ptrue p0.h, vl16
; 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:    frintz z0.h, p0/m, z0.h
; VBITS_GE_256-NEXT:    frintz z1.h, p0/m, z1.h
; VBITS_GE_256-NEXT:    st1h { z0.h }, p0, [x0, x8, lsl #1]
; VBITS_GE_256-NEXT:    st1h { z1.h }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: frintz_v32f16:
; 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:    frintz z0.h, p0/m, z0.h
; VBITS_GE_512-NEXT:    st1h { z0.h }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op = load <32 x half>, <32 x half>* %a
  %res = call <32 x half> @llvm.trunc.v32f16(<32 x half> %op)
  store <32 x half> %res, <32 x half>* %a
  ret void
}

define void @frintz_v64f16(<64 x half>* %a) vscale_range(8,0) #0 {
; CHECK-LABEL: frintz_v64f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl64
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    frintz z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <64 x half>, <64 x half>* %a
  %res = call <64 x half> @llvm.trunc.v64f16(<64 x half> %op)
  store <64 x half> %res, <64 x half>* %a
  ret void
}

define void @frintz_v128f16(<128 x half>* %a) vscale_range(16,0) #0 {
; CHECK-LABEL: frintz_v128f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl128
; CHECK-NEXT:    ld1h { z0.h }, p0/z, [x0]
; CHECK-NEXT:    frintz z0.h, p0/m, z0.h
; CHECK-NEXT:    st1h { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <128 x half>, <128 x half>* %a
  %res = call <128 x half> @llvm.trunc.v128f16(<128 x half> %op)
  store <128 x half> %res, <128 x half>* %a
  ret void
}

; Don't use SVE for 64-bit vectors.
define <2 x float> @frintz_v2f32(<2 x float> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintz_v2f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintz v0.2s, v0.2s
; CHECK-NEXT:    ret
  %res = call <2 x float> @llvm.trunc.v2f32(<2 x float> %op)
  ret <2 x float> %res
}

; Don't use SVE for 128-bit vectors.
define <4 x float> @frintz_v4f32(<4 x float> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintz_v4f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintz v0.4s, v0.4s
; CHECK-NEXT:    ret
  %res = call <4 x float> @llvm.trunc.v4f32(<4 x float> %op)
  ret <4 x float> %res
}

define void @frintz_v8f32(<8 x float>* %a) vscale_range(2,0) #0 {
; CHECK-LABEL: frintz_v8f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl8
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    frintz z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <8 x float>, <8 x float>* %a
  %res = call <8 x float> @llvm.trunc.v8f32(<8 x float> %op)
  store <8 x float> %res, <8 x float>* %a
  ret void
}

define void @frintz_v16f32(<16 x float>* %a) #0 {
; VBITS_GE_256-LABEL: frintz_v16f32:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    mov x8, #8
; VBITS_GE_256-NEXT:    ptrue p0.s, vl8
; 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:    frintz z0.s, p0/m, z0.s
; VBITS_GE_256-NEXT:    frintz z1.s, p0/m, z1.s
; VBITS_GE_256-NEXT:    st1w { z0.s }, p0, [x0, x8, lsl #2]
; VBITS_GE_256-NEXT:    st1w { z1.s }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: frintz_v16f32:
; 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:    frintz z0.s, p0/m, z0.s
; VBITS_GE_512-NEXT:    st1w { z0.s }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op = load <16 x float>, <16 x float>* %a
  %res = call <16 x float> @llvm.trunc.v16f32(<16 x float> %op)
  store <16 x float> %res, <16 x float>* %a
  ret void
}

define void @frintz_v32f32(<32 x float>* %a) vscale_range(8,0) #0 {
; CHECK-LABEL: frintz_v32f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl32
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    frintz z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <32 x float>, <32 x float>* %a
  %res = call <32 x float> @llvm.trunc.v32f32(<32 x float> %op)
  store <32 x float> %res, <32 x float>* %a
  ret void
}

define void @frintz_v64f32(<64 x float>* %a) vscale_range(16,0) #0 {
; CHECK-LABEL: frintz_v64f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl64
; CHECK-NEXT:    ld1w { z0.s }, p0/z, [x0]
; CHECK-NEXT:    frintz z0.s, p0/m, z0.s
; CHECK-NEXT:    st1w { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <64 x float>, <64 x float>* %a
  %res = call <64 x float> @llvm.trunc.v64f32(<64 x float> %op)
  store <64 x float> %res, <64 x float>* %a
  ret void
}

; Don't use SVE for 64-bit vectors.
define <1 x double> @frintz_v1f64(<1 x double> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintz_v1f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintz d0, d0
; CHECK-NEXT:    ret
  %res = call <1 x double> @llvm.trunc.v1f64(<1 x double> %op)
  ret <1 x double> %res
}

; Don't use SVE for 128-bit vectors.
define <2 x double> @frintz_v2f64(<2 x double> %op) vscale_range(2,0) #0 {
; CHECK-LABEL: frintz_v2f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    frintz v0.2d, v0.2d
; CHECK-NEXT:    ret
  %res = call <2 x double> @llvm.trunc.v2f64(<2 x double> %op)
  ret <2 x double> %res
}

define void @frintz_v4f64(<4 x double>* %a) vscale_range(2,0) #0 {
; CHECK-LABEL: frintz_v4f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl4
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    frintz z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <4 x double>, <4 x double>* %a
  %res = call <4 x double> @llvm.trunc.v4f64(<4 x double> %op)
  store <4 x double> %res, <4 x double>* %a
  ret void
}

define void @frintz_v8f64(<8 x double>* %a) #0 {
; VBITS_GE_256-LABEL: frintz_v8f64:
; VBITS_GE_256:       // %bb.0:
; VBITS_GE_256-NEXT:    mov x8, #4
; VBITS_GE_256-NEXT:    ptrue p0.d, vl4
; 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:    frintz z0.d, p0/m, z0.d
; VBITS_GE_256-NEXT:    frintz z1.d, p0/m, z1.d
; VBITS_GE_256-NEXT:    st1d { z0.d }, p0, [x0, x8, lsl #3]
; VBITS_GE_256-NEXT:    st1d { z1.d }, p0, [x0]
; VBITS_GE_256-NEXT:    ret
;
; VBITS_GE_512-LABEL: frintz_v8f64:
; 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:    frintz z0.d, p0/m, z0.d
; VBITS_GE_512-NEXT:    st1d { z0.d }, p0, [x0]
; VBITS_GE_512-NEXT:    ret
  %op = load <8 x double>, <8 x double>* %a
  %res = call <8 x double> @llvm.trunc.v8f64(<8 x double> %op)
  store <8 x double> %res, <8 x double>* %a
  ret void
}

define void @frintz_v16f64(<16 x double>* %a) vscale_range(8,0) #0 {
; CHECK-LABEL: frintz_v16f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl16
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    frintz z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <16 x double>, <16 x double>* %a
  %res = call <16 x double> @llvm.trunc.v16f64(<16 x double> %op)
  store <16 x double> %res, <16 x double>* %a
  ret void
}

define void @frintz_v32f64(<32 x double>* %a) vscale_range(16,0) #0 {
; CHECK-LABEL: frintz_v32f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl32
; CHECK-NEXT:    ld1d { z0.d }, p0/z, [x0]
; CHECK-NEXT:    frintz z0.d, p0/m, z0.d
; CHECK-NEXT:    st1d { z0.d }, p0, [x0]
; CHECK-NEXT:    ret
  %op = load <32 x double>, <32 x double>* %a
  %res = call <32 x double> @llvm.trunc.v32f64(<32 x double> %op)
  store <32 x double> %res, <32 x double>* %a
  ret void
}

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

declare <4 x half> @llvm.ceil.v4f16(<4 x half>)
declare <8 x half> @llvm.ceil.v8f16(<8 x half>)
declare <16 x half> @llvm.ceil.v16f16(<16 x half>)
declare <32 x half> @llvm.ceil.v32f16(<32 x half>)
declare <64 x half> @llvm.ceil.v64f16(<64 x half>)
declare <128 x half> @llvm.ceil.v128f16(<128 x half>)
declare <2 x float> @llvm.ceil.v2f32(<2 x float>)
declare <4 x float> @llvm.ceil.v4f32(<4 x float>)
declare <8 x float> @llvm.ceil.v8f32(<8 x float>)
declare <16 x float> @llvm.ceil.v16f32(<16 x float>)
declare <32 x float> @llvm.ceil.v32f32(<32 x float>)
declare <64 x float> @llvm.ceil.v64f32(<64 x float>)
declare <1 x double> @llvm.ceil.v1f64(<1 x double>)
declare <2 x double> @llvm.ceil.v2f64(<2 x double>)
declare <4 x double> @llvm.ceil.v4f64(<4 x double>)
declare <8 x double> @llvm.ceil.v8f64(<8 x double>)
declare <16 x double> @llvm.ceil.v16f64(<16 x double>)
declare <32 x double> @llvm.ceil.v32f64(<32 x double>)

declare <4 x half> @llvm.floor.v4f16(<4 x half>)
declare <8 x half> @llvm.floor.v8f16(<8 x half>)
declare <16 x half> @llvm.floor.v16f16(<16 x half>)
declare <32 x half> @llvm.floor.v32f16(<32 x half>)
declare <64 x half> @llvm.floor.v64f16(<64 x half>)
declare <128 x half> @llvm.floor.v128f16(<128 x half>)
declare <2 x float> @llvm.floor.v2f32(<2 x float>)
declare <4 x float> @llvm.floor.v4f32(<4 x float>)
declare <8 x float> @llvm.floor.v8f32(<8 x float>)
declare <16 x float> @llvm.floor.v16f32(<16 x float>)
declare <32 x float> @llvm.floor.v32f32(<32 x float>)
declare <64 x float> @llvm.floor.v64f32(<64 x float>)
declare <1 x double> @llvm.floor.v1f64(<1 x double>)
declare <2 x double> @llvm.floor.v2f64(<2 x double>)
declare <4 x double> @llvm.floor.v4f64(<4 x double>)
declare <8 x double> @llvm.floor.v8f64(<8 x double>)
declare <16 x double> @llvm.floor.v16f64(<16 x double>)
declare <32 x double> @llvm.floor.v32f64(<32 x double>)

declare <4 x half> @llvm.nearbyint.v4f16(<4 x half>)
declare <8 x half> @llvm.nearbyint.v8f16(<8 x half>)
declare <16 x half> @llvm.nearbyint.v16f16(<16 x half>)
declare <32 x half> @llvm.nearbyint.v32f16(<32 x half>)
declare <64 x half> @llvm.nearbyint.v64f16(<64 x half>)
declare <128 x half> @llvm.nearbyint.v128f16(<128 x half>)
declare <2 x float> @llvm.nearbyint.v2f32(<2 x float>)
declare <4 x float> @llvm.nearbyint.v4f32(<4 x float>)
declare <8 x float> @llvm.nearbyint.v8f32(<8 x float>)
declare <16 x float> @llvm.nearbyint.v16f32(<16 x float>)
declare <32 x float> @llvm.nearbyint.v32f32(<32 x float>)
declare <64 x float> @llvm.nearbyint.v64f32(<64 x float>)
declare <1 x double> @llvm.nearbyint.v1f64(<1 x double>)
declare <2 x double> @llvm.nearbyint.v2f64(<2 x double>)
declare <4 x double> @llvm.nearbyint.v4f64(<4 x double>)
declare <8 x double> @llvm.nearbyint.v8f64(<8 x double>)
declare <16 x double> @llvm.nearbyint.v16f64(<16 x double>)
declare <32 x double> @llvm.nearbyint.v32f64(<32 x double>)

declare <4 x half> @llvm.rint.v4f16(<4 x half>)
declare <8 x half> @llvm.rint.v8f16(<8 x half>)
declare <16 x half> @llvm.rint.v16f16(<16 x half>)
declare <32 x half> @llvm.rint.v32f16(<32 x half>)
declare <64 x half> @llvm.rint.v64f16(<64 x half>)
declare <128 x half> @llvm.rint.v128f16(<128 x half>)
declare <2 x float> @llvm.rint.v2f32(<2 x float>)
declare <4 x float> @llvm.rint.v4f32(<4 x float>)
declare <8 x float> @llvm.rint.v8f32(<8 x float>)
declare <16 x float> @llvm.rint.v16f32(<16 x float>)
declare <32 x float> @llvm.rint.v32f32(<32 x float>)
declare <64 x float> @llvm.rint.v64f32(<64 x float>)
declare <1 x double> @llvm.rint.v1f64(<1 x double>)
declare <2 x double> @llvm.rint.v2f64(<2 x double>)
declare <4 x double> @llvm.rint.v4f64(<4 x double>)
declare <8 x double> @llvm.rint.v8f64(<8 x double>)
declare <16 x double> @llvm.rint.v16f64(<16 x double>)
declare <32 x double> @llvm.rint.v32f64(<32 x double>)

declare <4 x half> @llvm.round.v4f16(<4 x half>)
declare <8 x half> @llvm.round.v8f16(<8 x half>)
declare <16 x half> @llvm.round.v16f16(<16 x half>)
declare <32 x half> @llvm.round.v32f16(<32 x half>)
declare <64 x half> @llvm.round.v64f16(<64 x half>)
declare <128 x half> @llvm.round.v128f16(<128 x half>)
declare <2 x float> @llvm.round.v2f32(<2 x float>)
declare <4 x float> @llvm.round.v4f32(<4 x float>)
declare <8 x float> @llvm.round.v8f32(<8 x float>)
declare <16 x float> @llvm.round.v16f32(<16 x float>)
declare <32 x float> @llvm.round.v32f32(<32 x float>)
declare <64 x float> @llvm.round.v64f32(<64 x float>)
declare <1 x double> @llvm.round.v1f64(<1 x double>)
declare <2 x double> @llvm.round.v2f64(<2 x double>)
declare <4 x double> @llvm.round.v4f64(<4 x double>)
declare <8 x double> @llvm.round.v8f64(<8 x double>)
declare <16 x double> @llvm.round.v16f64(<16 x double>)
declare <32 x double> @llvm.round.v32f64(<32 x double>)

declare <4 x half> @llvm.roundeven.v4f16(<4 x half>)
declare <8 x half> @llvm.roundeven.v8f16(<8 x half>)
declare <16 x half> @llvm.roundeven.v16f16(<16 x half>)
declare <32 x half> @llvm.roundeven.v32f16(<32 x half>)
declare <64 x half> @llvm.roundeven.v64f16(<64 x half>)
declare <128 x half> @llvm.roundeven.v128f16(<128 x half>)
declare <2 x float> @llvm.roundeven.v2f32(<2 x float>)
declare <4 x float> @llvm.roundeven.v4f32(<4 x float>)
declare <8 x float> @llvm.roundeven.v8f32(<8 x float>)
declare <16 x float> @llvm.roundeven.v16f32(<16 x float>)
declare <32 x float> @llvm.roundeven.v32f32(<32 x float>)
declare <64 x float> @llvm.roundeven.v64f32(<64 x float>)
declare <1 x double> @llvm.roundeven.v1f64(<1 x double>)
declare <2 x double> @llvm.roundeven.v2f64(<2 x double>)
declare <4 x double> @llvm.roundeven.v4f64(<4 x double>)
declare <8 x double> @llvm.roundeven.v8f64(<8 x double>)
declare <16 x double> @llvm.roundeven.v16f64(<16 x double>)
declare <32 x double> @llvm.roundeven.v32f64(<32 x double>)

declare <4 x half> @llvm.trunc.v4f16(<4 x half>)
declare <8 x half> @llvm.trunc.v8f16(<8 x half>)
declare <16 x half> @llvm.trunc.v16f16(<16 x half>)
declare <32 x half> @llvm.trunc.v32f16(<32 x half>)
declare <64 x half> @llvm.trunc.v64f16(<64 x half>)
declare <128 x half> @llvm.trunc.v128f16(<128 x half>)
declare <2 x float> @llvm.trunc.v2f32(<2 x float>)
declare <4 x float> @llvm.trunc.v4f32(<4 x float>)
declare <8 x float> @llvm.trunc.v8f32(<8 x float>)
declare <16 x float> @llvm.trunc.v16f32(<16 x float>)
declare <32 x float> @llvm.trunc.v32f32(<32 x float>)
declare <64 x float> @llvm.trunc.v64f32(<64 x float>)
declare <1 x double> @llvm.trunc.v1f64(<1 x double>)
declare <2 x double> @llvm.trunc.v2f64(<2 x double>)
declare <4 x double> @llvm.trunc.v4f64(<4 x double>)
declare <8 x double> @llvm.trunc.v8f64(<8 x double>)
declare <16 x double> @llvm.trunc.v16f64(<16 x double>)
declare <32 x double> @llvm.trunc.v32f64(<32 x double>)