; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sme2 -force-streaming -enable-subreg-liveness -verify-machineinstrs < %s | FileCheck %s

;
; SQCVT
;

; x2
define <vscale x 8 x i16 > @multi_vector_qcvt_x2_s16_s32(<vscale x 4 x i32> %unused, <vscale x 4 x i32> %zn1, <vscale x 4 x i32> %zn2) {
; CHECK-LABEL: multi_vector_qcvt_x2_s16_s32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z3.d, z2.d
; CHECK-NEXT:    mov z2.d, z1.d
; CHECK-NEXT:    sqcvt z0.h, { z2.s, z3.s }
; CHECK-NEXT:    ret
  %res = call <vscale x 8 x i16> @llvm.aarch64.sve.sqcvt.x2.nxv4i32(<vscale x 4 x i32> %zn1, <vscale x 4 x i32> %zn2)
  ret <vscale x 8 x i16> %res
}

; x4
define <vscale x 16 x i8 > @multi_vector_qcvt_x4_s8_s32(<vscale x 4 x i32> %unused, <vscale x 4 x i32> %zn1, <vscale x 4 x i32> %zn2, <vscale x 4 x i32> %zn3, <vscale x 4 x i32> %zn4) {
; CHECK-LABEL: multi_vector_qcvt_x4_s8_s32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z7.d, z4.d
; CHECK-NEXT:    mov z6.d, z3.d
; CHECK-NEXT:    mov z5.d, z2.d
; CHECK-NEXT:    mov z4.d, z1.d
; CHECK-NEXT:    sqcvt z0.b, { z4.s - z7.s }
; CHECK-NEXT:    ret
  %res = call <vscale x 16 x i8> @llvm.aarch64.sve.sqcvt.x4.nxv4i32(<vscale x 4 x i32> %zn1, <vscale x 4 x i32> %zn2, <vscale x 4 x i32> %zn3, <vscale x 4 x i32> %zn4)
  ret <vscale x 16 x i8> %res
}

define <vscale x 8 x i16> @multi_vector_qcvt_x4_s16_s64(<vscale x 2 x i64> %unused, <vscale x 2 x i64> %zn1, <vscale x 2 x i64> %zn2, <vscale x 2 x i64> %zn3, <vscale x 2 x i64> %zn4) {
; CHECK-LABEL: multi_vector_qcvt_x4_s16_s64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z7.d, z4.d
; CHECK-NEXT:    mov z6.d, z3.d
; CHECK-NEXT:    mov z5.d, z2.d
; CHECK-NEXT:    mov z4.d, z1.d
; CHECK-NEXT:    sqcvt z0.h, { z4.d - z7.d }
; CHECK-NEXT:    ret
  %res = call <vscale x 8 x i16> @llvm.aarch64.sve.sqcvt.x4.nxv2i64(<vscale x 2 x i64> %zn1, <vscale x 2 x i64> %zn2, <vscale x 2 x i64> %zn3, <vscale x 2 x i64> %zn4)
  ret <vscale x 8 x i16> %res
}

define { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @multi_vector_qcvt_x4_s16_s64_tuple(i64 %stride, ptr %ptr) {
; CHECK-LABEL: multi_vector_qcvt_x4_s16_s64_tuple:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    str x29, [sp, #-16]! // 8-byte Folded Spill
; CHECK-NEXT:    addvl sp, sp, #-9
; CHECK-NEXT:    str p8, [sp, #7, mul vl] // 2-byte Spill
; CHECK-NEXT:    str z23, [sp, #1, mul vl] // 16-byte Folded Spill
; CHECK-NEXT:    str z22, [sp, #2, mul vl] // 16-byte Folded Spill
; CHECK-NEXT:    str z21, [sp, #3, mul vl] // 16-byte Folded Spill
; CHECK-NEXT:    str z20, [sp, #4, mul vl] // 16-byte Folded Spill
; CHECK-NEXT:    str z19, [sp, #5, mul vl] // 16-byte Folded Spill
; CHECK-NEXT:    str z18, [sp, #6, mul vl] // 16-byte Folded Spill
; CHECK-NEXT:    str z17, [sp, #7, mul vl] // 16-byte Folded Spill
; CHECK-NEXT:    str z16, [sp, #8, mul vl] // 16-byte Folded Spill
; CHECK-NEXT:    .cfi_escape 0x0f, 0x0a, 0x8f, 0x10, 0x92, 0x2e, 0x00, 0x11, 0xc8, 0x00, 0x1e, 0x22 // sp + 16 + 72 * VG
; CHECK-NEXT:    .cfi_offset w29, -16
; CHECK-NEXT:    lsl x8, x0, #1
; CHECK-NEXT:    add x9, x1, x0
; CHECK-NEXT:    ptrue pn8.b
; CHECK-NEXT:    ld1d { z16.d, z20.d, z24.d, z28.d }, pn8/z, [x1]
; CHECK-NEXT:    ld1d { z17.d, z21.d, z25.d, z29.d }, pn8/z, [x9]
; CHECK-NEXT:    add x10, x1, x8
; CHECK-NEXT:    add x8, x9, x8
; CHECK-NEXT:    ld1d { z18.d, z22.d, z26.d, z30.d }, pn8/z, [x10]
; CHECK-NEXT:    ld1d { z19.d, z23.d, z27.d, z31.d }, pn8/z, [x8]
; CHECK-NEXT:    sqcvt z0.h, { z16.d - z19.d }
; CHECK-NEXT:    sqcvt z1.h, { z20.d - z23.d }
; CHECK-NEXT:    sqcvt z2.h, { z24.d - z27.d }
; CHECK-NEXT:    sqcvt z3.h, { z28.d - z31.d }
; CHECK-NEXT:    ldr z23, [sp, #1, mul vl] // 16-byte Folded Reload
; CHECK-NEXT:    ldr z22, [sp, #2, mul vl] // 16-byte Folded Reload
; CHECK-NEXT:    ldr z21, [sp, #3, mul vl] // 16-byte Folded Reload
; CHECK-NEXT:    ldr z20, [sp, #4, mul vl] // 16-byte Folded Reload
; CHECK-NEXT:    ldr z19, [sp, #5, mul vl] // 16-byte Folded Reload
; CHECK-NEXT:    ldr z18, [sp, #6, mul vl] // 16-byte Folded Reload
; CHECK-NEXT:    ldr z17, [sp, #7, mul vl] // 16-byte Folded Reload
; CHECK-NEXT:    ldr z16, [sp, #8, mul vl] // 16-byte Folded Reload
; CHECK-NEXT:    ldr p8, [sp, #7, mul vl] // 2-byte Reload
; CHECK-NEXT:    addvl sp, sp, #9
; CHECK-NEXT:    ldr x29, [sp], #16 // 8-byte Folded Reload
; CHECK-NEXT:    ret
entry:
  %0 = tail call target("aarch64.svcount") @llvm.aarch64.sve.ptrue.c8()
  %1 = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld1.pn.x4.nxv2i64(target("aarch64.svcount") %0, ptr %ptr)
  %2 = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } %1, 0
  %3 = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } %1, 1
  %4 = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } %1, 2
  %5 = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } %1, 3
  %arrayidx2 = getelementptr inbounds i8, ptr %ptr, i64 %stride
  %6 = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld1.pn.x4.nxv2i64(target("aarch64.svcount") %0, ptr %arrayidx2)
  %7 = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } %6, 0
  %8 = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } %6, 1
  %9 = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } %6, 2
  %10 = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } %6, 3
  %mul3 = shl i64 %stride, 1
  %arrayidx4 = getelementptr inbounds i8, ptr %ptr, i64 %mul3
  %11 = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld1.pn.x4.nxv2i64(target("aarch64.svcount") %0, ptr %arrayidx4)
  %12 = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } %11, 0
  %13 = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } %11, 1
  %14 = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } %11, 2
  %15 = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } %11, 3
  %mul5 = mul i64 %stride, 3
  %arrayidx6 = getelementptr inbounds i8, ptr %ptr, i64 %mul5
  %16 = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sve.ld1.pn.x4.nxv2i64(target("aarch64.svcount") %0, ptr %arrayidx6)
  %17 = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } %16, 0
  %18 = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } %16, 1
  %19 = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } %16, 2
  %20 = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } %16, 3
  %res1 = call <vscale x 8 x i16> @llvm.aarch64.sve.sqcvt.x4.nxv2i64(<vscale x 2 x i64> %2, <vscale x 2 x i64> %7, <vscale x 2 x i64> %12, <vscale x 2 x i64> %17)
  %res2 = call <vscale x 8 x i16> @llvm.aarch64.sve.sqcvt.x4.nxv2i64(<vscale x 2 x i64> %3, <vscale x 2 x i64> %8, <vscale x 2 x i64> %13, <vscale x 2 x i64> %18)
  %res3 = call <vscale x 8 x i16> @llvm.aarch64.sve.sqcvt.x4.nxv2i64(<vscale x 2 x i64> %4, <vscale x 2 x i64> %9, <vscale x 2 x i64> %14, <vscale x 2 x i64> %19)
  %res4 = call <vscale x 8 x i16> @llvm.aarch64.sve.sqcvt.x4.nxv2i64(<vscale x 2 x i64> %5, <vscale x 2 x i64> %10, <vscale x 2 x i64> %15, <vscale x 2 x i64> %20)
  %ins1 = insertvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } poison, <vscale x 8 x i16> %res1, 0
  %ins2 = insertvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } %ins1, <vscale x 8 x i16> %res2, 1
  %ins3 = insertvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } %ins2, <vscale x 8 x i16> %res3, 2
  %ins4 = insertvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } %ins3, <vscale x 8 x i16> %res4, 3
  ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } %ins4
}

;
; UQCVT
;

; x2
define <vscale x 8 x i16> @multi_vector_qcvt_x2_u16_u32(<vscale x 4 x i32> %unused, <vscale x 4 x i32> %zn0, <vscale x 4 x i32> %zn1) {
; CHECK-LABEL: multi_vector_qcvt_x2_u16_u32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z3.d, z2.d
; CHECK-NEXT:    mov z2.d, z1.d
; CHECK-NEXT:    uqcvt z0.h, { z2.s, z3.s }
; CHECK-NEXT:    ret
  %res = call <vscale x 8 x i16> @llvm.aarch64.sve.uqcvt.x2.nxv4i32(<vscale x 4 x i32> %zn0, <vscale x 4 x i32> %zn1)
  ret<vscale x 8 x i16> %res
}

; x4
define <vscale x 16 x i8> @multi_vector_qcvt_x4_u8_u32(<vscale x 4 x i32> %unused, <vscale x 4 x i32> %zn1, <vscale x 4 x i32> %zn2, <vscale x 4 x i32> %zn3, <vscale x 4 x i32> %zn4) {
; CHECK-LABEL: multi_vector_qcvt_x4_u8_u32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z7.d, z4.d
; CHECK-NEXT:    mov z6.d, z3.d
; CHECK-NEXT:    mov z5.d, z2.d
; CHECK-NEXT:    mov z4.d, z1.d
; CHECK-NEXT:    uqcvt z0.b, { z4.s - z7.s }
; CHECK-NEXT:    ret
  %res = call <vscale x 16 x i8> @llvm.aarch64.sve.uqcvt.x4.nxv4i32(<vscale x 4 x i32> %zn1, <vscale x 4 x i32> %zn2, <vscale x 4 x i32> %zn3, <vscale x 4 x i32> %zn4)
  ret <vscale x 16 x i8> %res
}

define <vscale x 8 x i16> @multi_vector_qcvt_x4_u16_u64(<vscale x 2 x i64> %unused, <vscale x 2 x i64> %zn1, <vscale x 2 x i64> %zn2, <vscale x 2 x i64> %zn3, <vscale x 2 x i64> %zn4) {
; CHECK-LABEL: multi_vector_qcvt_x4_u16_u64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z7.d, z4.d
; CHECK-NEXT:    mov z6.d, z3.d
; CHECK-NEXT:    mov z5.d, z2.d
; CHECK-NEXT:    mov z4.d, z1.d
; CHECK-NEXT:    uqcvt z0.h, { z4.d - z7.d }
; CHECK-NEXT:    ret
  %res = call <vscale x 8 x i16> @llvm.aarch64.sve.uqcvt.x4.nxv2i64(<vscale x 2 x i64> %zn1, <vscale x 2 x i64> %zn2, <vscale x 2 x i64> %zn3, <vscale x 2 x i64> %zn4)
  ret <vscale x 8 x i16> %res
}

;
; SQCVTU
;

; x2
define <vscale x 8 x i16 > @multi_vector_qcvt_x2_s16_u32(<vscale x 4 x i32> %unused, <vscale x 4 x i32> %zn1, <vscale x 4 x i32> %zn2) {
; CHECK-LABEL: multi_vector_qcvt_x2_s16_u32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z3.d, z2.d
; CHECK-NEXT:    mov z2.d, z1.d
; CHECK-NEXT:    sqcvtu z0.h, { z2.s, z3.s }
; CHECK-NEXT:    ret
  %res = call <vscale x 8 x i16> @llvm.aarch64.sve.sqcvtu.x2.nxv4i32(<vscale x 4 x i32> %zn1, <vscale x 4 x i32> %zn2)
  ret <vscale x 8 x i16> %res
}

; x4
define <vscale x 16 x i8> @multi_vector_qcvt_x4_u8_s32(<vscale x 4 x i32> %unused, <vscale x 4 x i32> %zn1, <vscale x 4 x i32> %zn2, <vscale x 4 x i32> %zn3, <vscale x 4 x i32> %zn4) {
; CHECK-LABEL: multi_vector_qcvt_x4_u8_s32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z7.d, z4.d
; CHECK-NEXT:    mov z6.d, z3.d
; CHECK-NEXT:    mov z5.d, z2.d
; CHECK-NEXT:    mov z4.d, z1.d
; CHECK-NEXT:    sqcvtu z0.b, { z4.s - z7.s }
; CHECK-NEXT:    ret
  %res = call <vscale x 16 x i8> @llvm.aarch64.sve.sqcvtu.x4.nxv4i32(<vscale x 4 x i32> %zn1, <vscale x 4 x i32> %zn2, <vscale x 4 x i32> %zn3, <vscale x 4 x i32> %zn4)
  ret <vscale x 16 x i8> %res
}

define <vscale x 8 x i16> @multi_vector_qcvt_x4_u16_s64(<vscale x 2 x i64> %unused, <vscale x 2 x i64> %zn1, <vscale x 2 x i64> %zn2, <vscale x 2 x i64> %zn3, <vscale x 2 x i64> %zn4) {
; CHECK-LABEL: multi_vector_qcvt_x4_u16_s64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    mov z7.d, z4.d
; CHECK-NEXT:    mov z6.d, z3.d
; CHECK-NEXT:    mov z5.d, z2.d
; CHECK-NEXT:    mov z4.d, z1.d
; CHECK-NEXT:    sqcvtu z0.h, { z4.d - z7.d }
; CHECK-NEXT:    ret
  %res = call <vscale x 8 x i16> @llvm.aarch64.sve.sqcvtu.x4.nxv2i64(<vscale x 2 x i64> %zn1, <vscale x 2 x i64> %zn2, <vscale x 2 x i64> %zn3, <vscale x 2 x i64> %zn4)
  ret <vscale x 8 x i16> %res
}

declare <vscale x 8 x i16> @llvm.aarch64.sve.sqcvt.x2.nxv4i32(<vscale x 4 x i32>, <vscale x 4 x i32>)
declare <vscale x 8 x i16> @llvm.aarch64.sve.uqcvt.x2.nxv4i32(<vscale x 4 x i32>, <vscale x 4 x i32>)
declare <vscale x 8 x i16> @llvm.aarch64.sve.sqcvtu.x2.nxv4i32(<vscale x 4 x i32>, <vscale x 4 x i32>)
declare <vscale x 16 x i8> @llvm.aarch64.sve.sqcvt.x4.nxv4i32(<vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>)
declare <vscale x 8 x i16> @llvm.aarch64.sve.sqcvt.x4.nxv2i64(<vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>)
declare <vscale x 16 x i8> @llvm.aarch64.sve.uqcvt.x4.nxv4i32(<vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>)
declare <vscale x 8 x i16> @llvm.aarch64.sve.uqcvt.x4.nxv2i64(<vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>)
declare <vscale x 16 x i8> @llvm.aarch64.sve.sqcvtu.x4.nxv4i32(<vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>)
declare <vscale x 8 x i16> @llvm.aarch64.sve.sqcvtu.x4.nxv2i64(<vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>)