llvm/test/CodeGen/AArch64/complex-deinterleaving-reductions-scalable.ll

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

target triple = "aarch64"

%"class.std::complex" = type { { double, double } }

; Zero initialized reduction
;
;   complex<double> x = 0.0 + 0.0i;
;   for (int i = 0; i < 100; ++i)
;       x += a[i] * b[i];
;
define %"class.std::complex" @complex_mul_v2f64(ptr %a, ptr %b) {
; CHECK-LABEL: complex_mul_v2f64:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    mov z1.d, #0 // =0x0
; CHECK-NEXT:    ptrue p1.b
; CHECK-NEXT:    cntd x9
; CHECK-NEXT:    ptrue p0.d
; CHECK-NEXT:    neg x9, x9
; CHECK-NEXT:    mov w10, #100 // =0x64
; CHECK-NEXT:    mov x8, xzr
; CHECK-NEXT:    and x10, x9, x10
; CHECK-NEXT:    rdvl x11, #2
; CHECK-NEXT:    zip2 z0.d, z1.d, z1.d
; CHECK-NEXT:    zip1 z1.d, z1.d, z1.d
; CHECK-NEXT:  .LBB0_1: // %vector.body
; CHECK-NEXT:    // =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    add x12, x0, x8
; CHECK-NEXT:    add x13, x1, x8
; CHECK-NEXT:    ld1b { z2.b }, p1/z, [x0, x8]
; CHECK-NEXT:    ld1d { z3.d }, p0/z, [x12, #1, mul vl]
; CHECK-NEXT:    ld1b { z4.b }, p1/z, [x1, x8]
; CHECK-NEXT:    ld1d { z5.d }, p0/z, [x13, #1, mul vl]
; CHECK-NEXT:    adds x10, x10, x9
; CHECK-NEXT:    add x8, x8, x11
; CHECK-NEXT:    fcmla z1.d, p0/m, z4.d, z2.d, #0
; CHECK-NEXT:    fcmla z0.d, p0/m, z5.d, z3.d, #0
; CHECK-NEXT:    fcmla z1.d, p0/m, z4.d, z2.d, #90
; CHECK-NEXT:    fcmla z0.d, p0/m, z5.d, z3.d, #90
; CHECK-NEXT:    b.ne .LBB0_1
; CHECK-NEXT:  // %bb.2: // %exit.block
; CHECK-NEXT:    uzp1 z2.d, z1.d, z0.d
; CHECK-NEXT:    uzp2 z1.d, z1.d, z0.d
; CHECK-NEXT:    faddv d0, p0, z2.d
; CHECK-NEXT:    faddv d1, p0, z1.d
; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $z0
; CHECK-NEXT:    // kill: def $d1 killed $d1 killed $z1
; CHECK-NEXT:    ret
entry:
  %0 = tail call i64 @llvm.vscale.i64()
  %1 = shl nuw nsw i64 %0, 1
  %n.mod.vf = urem i64 100, %1
  %n.vec = sub nuw nsw i64 100, %n.mod.vf
  %2 = shl nuw nsw i64 %0, 5
  br label %vector.body

vector.body:                                      ; preds = %vector.body, %entry
  %lsr.iv31 = phi i64 [ %lsr.iv.next32, %vector.body ], [ %n.vec, %entry ]
  %lsr.iv27 = phi i64 [ %lsr.iv.next28, %vector.body ], [ 0, %entry ]
  %vec.phi = phi <vscale x 2 x double> [ zeroinitializer, %entry ], [ %16, %vector.body ]
  %vec.phi12 = phi <vscale x 2 x double> [ zeroinitializer, %entry ], [ %14, %vector.body ]
  %scevgep46 = getelementptr i8, ptr %a, i64 %lsr.iv27
  %scevgep47 = getelementptr i8, ptr %b, i64 %lsr.iv27
  %wide.vec = load <vscale x 4 x double>, ptr %scevgep46, align 8
  %3 = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.experimental.vector.deinterleave2.nxv4f64(<vscale x 4 x double> %wide.vec)
  %4 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %3, 0
  %5 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %3, 1
  %wide.vec30 = load <vscale x 4 x double>, ptr %scevgep47, align 8
  %6 = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.experimental.vector.deinterleave2.nxv4f64(<vscale x 4 x double> %wide.vec30)
  %7 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %6, 0
  %8 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %6, 1
  %9 = fmul fast <vscale x 2 x double> %8, %4
  %10 = fmul fast <vscale x 2 x double> %7, %5
  %11 = fmul fast <vscale x 2 x double> %7, %4
  %12 = fadd fast <vscale x 2 x double> %11, %vec.phi12
  %13 = fmul fast <vscale x 2 x double> %8, %5
  %14 = fsub fast <vscale x 2 x double> %12, %13
  %15 = fadd fast <vscale x 2 x double> %10, %vec.phi
  %16 = fadd fast <vscale x 2 x double> %15, %9
  %lsr.iv.next28 = add i64 %lsr.iv27, %2
  %lsr.iv.next32 = sub i64 %lsr.iv31, %1
  %17 = icmp eq i64 %lsr.iv.next32, 0
  br i1 %17, label %exit.block, label %vector.body

exit.block:                                     ; preds = %vector.body
  %18 = tail call fast double @llvm.vector.reduce.fadd.nxv2f64(double -0.000000e+00, <vscale x 2 x double> %14)
  %19 = tail call fast double @llvm.vector.reduce.fadd.nxv2f64(double -0.000000e+00, <vscale x 2 x double> %16)
  %.fca.0.0.insert = insertvalue %"class.std::complex" poison, double %18, 0, 0
  %.fca.0.1.insert = insertvalue %"class.std::complex" %.fca.0.0.insert, double %19, 0, 1
  ret %"class.std::complex" %.fca.0.1.insert
}

; Fixed value initialized reduction
;
;   complex<double> x = 2.0 + 1.0i;
;   for (int i = 0; i < 100; ++i)
;       x += a[i] * b[i];
;
define %"class.std::complex" @complex_mul_nonzero_init_v2f64(ptr %a, ptr %b) {
; CHECK-LABEL: complex_mul_nonzero_init_v2f64:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    ptrue p0.d, vl1
; CHECK-NEXT:    fmov d0, #1.00000000
; CHECK-NEXT:    mov z1.d, #0 // =0x0
; CHECK-NEXT:    fmov d2, #2.00000000
; CHECK-NEXT:    cntd x9
; CHECK-NEXT:    mov w10, #100 // =0x64
; CHECK-NEXT:    ptrue p1.b
; CHECK-NEXT:    neg x9, x9
; CHECK-NEXT:    mov x8, xzr
; CHECK-NEXT:    and x10, x9, x10
; CHECK-NEXT:    rdvl x11, #2
; CHECK-NEXT:    sel z3.d, p0, z0.d, z1.d
; CHECK-NEXT:    mov z1.d, p0/m, z2.d
; CHECK-NEXT:    ptrue p0.d
; CHECK-NEXT:    zip2 z0.d, z1.d, z3.d
; CHECK-NEXT:    zip1 z1.d, z1.d, z3.d
; CHECK-NEXT:  .LBB1_1: // %vector.body
; CHECK-NEXT:    // =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    add x12, x0, x8
; CHECK-NEXT:    add x13, x1, x8
; CHECK-NEXT:    ld1b { z2.b }, p1/z, [x0, x8]
; CHECK-NEXT:    ld1d { z3.d }, p0/z, [x12, #1, mul vl]
; CHECK-NEXT:    ld1b { z4.b }, p1/z, [x1, x8]
; CHECK-NEXT:    ld1d { z5.d }, p0/z, [x13, #1, mul vl]
; CHECK-NEXT:    adds x10, x10, x9
; CHECK-NEXT:    add x8, x8, x11
; CHECK-NEXT:    fcmla z1.d, p0/m, z4.d, z2.d, #0
; CHECK-NEXT:    fcmla z0.d, p0/m, z5.d, z3.d, #0
; CHECK-NEXT:    fcmla z1.d, p0/m, z4.d, z2.d, #90
; CHECK-NEXT:    fcmla z0.d, p0/m, z5.d, z3.d, #90
; CHECK-NEXT:    b.ne .LBB1_1
; CHECK-NEXT:  // %bb.2: // %exit.block
; CHECK-NEXT:    uzp1 z2.d, z1.d, z0.d
; CHECK-NEXT:    uzp2 z1.d, z1.d, z0.d
; CHECK-NEXT:    faddv d0, p0, z2.d
; CHECK-NEXT:    faddv d1, p0, z1.d
; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $z0
; CHECK-NEXT:    // kill: def $d1 killed $d1 killed $z1
; CHECK-NEXT:    ret
entry:
  %0 = tail call i64 @llvm.vscale.i64()
  %1 = shl nuw nsw i64 %0, 1
  %n.mod.vf = urem i64 100, %1
  %n.vec = sub nuw nsw i64 100, %n.mod.vf
  %2 = shl nuw nsw i64 %0, 5
  br label %vector.body

vector.body:                                      ; preds = %vector.body, %entry
  %lsr.iv31 = phi i64 [ %lsr.iv.next32, %vector.body ], [ %n.vec, %entry ]
  %lsr.iv27 = phi i64 [ %lsr.iv.next28, %vector.body ], [ 0, %entry ]
  %vec.phi = phi <vscale x 2 x double> [ insertelement (<vscale x 2 x double> zeroinitializer, double 1.000000e+00, i32 0), %entry ], [ %16, %vector.body ]
  %vec.phi12 = phi <vscale x 2 x double> [ insertelement (<vscale x 2 x double> zeroinitializer, double 2.000000e+0, i32 0), %entry ], [ %14, %vector.body ]
  %scevgep46 = getelementptr i8, ptr %a, i64 %lsr.iv27
  %scevgep47 = getelementptr i8, ptr %b, i64 %lsr.iv27
  %wide.vec = load <vscale x 4 x double>, ptr %scevgep46, align 8
  %3 = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.experimental.vector.deinterleave2.nxv4f64(<vscale x 4 x double> %wide.vec)
  %4 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %3, 0
  %5 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %3, 1
  %wide.vec30 = load <vscale x 4 x double>, ptr %scevgep47, align 8
  %6 = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.experimental.vector.deinterleave2.nxv4f64(<vscale x 4 x double> %wide.vec30)
  %7 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %6, 0
  %8 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %6, 1
  %9 = fmul fast <vscale x 2 x double> %8, %4
  %10 = fmul fast <vscale x 2 x double> %7, %5
  %11 = fmul fast <vscale x 2 x double> %7, %4
  %12 = fadd fast <vscale x 2 x double> %11, %vec.phi12
  %13 = fmul fast <vscale x 2 x double> %8, %5
  %14 = fsub fast <vscale x 2 x double> %12, %13
  %15 = fadd fast <vscale x 2 x double> %10, %vec.phi
  %16 = fadd fast <vscale x 2 x double> %15, %9
  %lsr.iv.next28 = add i64 %lsr.iv27, %2
  %lsr.iv.next32 = sub i64 %lsr.iv31, %1
  %17 = icmp eq i64 %lsr.iv.next32, 0
  br i1 %17, label %exit.block, label %vector.body

exit.block:                                     ; preds = %vector.body
  %18 = tail call fast double @llvm.vector.reduce.fadd.nxv2f64(double -0.000000e+00, <vscale x 2 x double> %14)
  %19 = tail call fast double @llvm.vector.reduce.fadd.nxv2f64(double -0.000000e+00, <vscale x 2 x double> %16)
  %.fca.0.0.insert = insertvalue %"class.std::complex" poison, double %18, 0, 0
  %.fca.0.1.insert = insertvalue %"class.std::complex" %.fca.0.0.insert, double %19, 0, 1
  ret %"class.std::complex" %.fca.0.1.insert
}

; Loop unrolled with factor 2
;
define %"class.std::complex" @complex_mul_v2f64_unrolled(ptr %a, ptr %b) {
; CHECK-LABEL: complex_mul_v2f64_unrolled:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    mov z1.d, #0 // =0x0
; CHECK-NEXT:    ptrue p1.b
; CHECK-NEXT:    cntw x9
; CHECK-NEXT:    ptrue p0.d
; CHECK-NEXT:    neg x9, x9
; CHECK-NEXT:    mov w10, #1000 // =0x3e8
; CHECK-NEXT:    rdvl x12, #2
; CHECK-NEXT:    mov x8, xzr
; CHECK-NEXT:    and x10, x9, x10
; CHECK-NEXT:    zip2 z0.d, z1.d, z1.d
; CHECK-NEXT:    zip1 z1.d, z1.d, z1.d
; CHECK-NEXT:    add x11, x1, x12
; CHECK-NEXT:    add x12, x0, x12
; CHECK-NEXT:    rdvl x13, #4
; CHECK-NEXT:    mov z2.d, z1.d
; CHECK-NEXT:    mov z3.d, z0.d
; CHECK-NEXT:  .LBB2_1: // %vector.body
; CHECK-NEXT:    // =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    add x14, x0, x8
; CHECK-NEXT:    add x15, x12, x8
; CHECK-NEXT:    add x16, x1, x8
; CHECK-NEXT:    add x17, x11, x8
; CHECK-NEXT:    ld1b { z4.b }, p1/z, [x0, x8]
; CHECK-NEXT:    ld1d { z5.d }, p0/z, [x14, #1, mul vl]
; CHECK-NEXT:    ld1b { z6.b }, p1/z, [x12, x8]
; CHECK-NEXT:    ld1b { z7.b }, p1/z, [x1, x8]
; CHECK-NEXT:    ld1d { z16.d }, p0/z, [x16, #1, mul vl]
; CHECK-NEXT:    ld1d { z17.d }, p0/z, [x15, #1, mul vl]
; CHECK-NEXT:    ld1b { z18.b }, p1/z, [x11, x8]
; CHECK-NEXT:    ld1d { z19.d }, p0/z, [x17, #1, mul vl]
; CHECK-NEXT:    adds x10, x10, x9
; CHECK-NEXT:    add x8, x8, x13
; CHECK-NEXT:    fcmla z1.d, p0/m, z7.d, z4.d, #0
; CHECK-NEXT:    fcmla z0.d, p0/m, z16.d, z5.d, #0
; CHECK-NEXT:    fcmla z2.d, p0/m, z18.d, z6.d, #0
; CHECK-NEXT:    fcmla z3.d, p0/m, z19.d, z17.d, #0
; CHECK-NEXT:    fcmla z1.d, p0/m, z7.d, z4.d, #90
; CHECK-NEXT:    fcmla z0.d, p0/m, z16.d, z5.d, #90
; CHECK-NEXT:    fcmla z2.d, p0/m, z18.d, z6.d, #90
; CHECK-NEXT:    fcmla z3.d, p0/m, z19.d, z17.d, #90
; CHECK-NEXT:    b.ne .LBB2_1
; CHECK-NEXT:  // %bb.2: // %exit.block
; CHECK-NEXT:    uzp1 z4.d, z2.d, z3.d
; CHECK-NEXT:    uzp1 z5.d, z1.d, z0.d
; CHECK-NEXT:    uzp2 z2.d, z2.d, z3.d
; CHECK-NEXT:    uzp2 z0.d, z1.d, z0.d
; CHECK-NEXT:    fadd z1.d, z4.d, z5.d
; CHECK-NEXT:    fadd z2.d, z2.d, z0.d
; CHECK-NEXT:    faddv d0, p0, z1.d
; CHECK-NEXT:    faddv d1, p0, z2.d
; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $z0
; CHECK-NEXT:    // kill: def $d1 killed $d1 killed $z1
; CHECK-NEXT:    ret
entry:
  %0 = tail call i64 @llvm.vscale.i64()
  %1 = shl nuw nsw i64 %0, 2
  %n.mod.vf = urem i64 1000, %1
  %n.vec = sub i64 1000, %n.mod.vf
  %2 = shl nuw nsw i64 %0, 6
  %3 = shl nuw nsw i64 %0, 5
  %scevgep61 = getelementptr i8, ptr %b, i64 %3
  %scevgep63 = getelementptr i8, ptr %a, i64 %3
  br label %vector.body

vector.body:                                      ; preds = %vector.body, %entry
  %lsr.iv38 = phi i64 [ %lsr.iv.next39, %vector.body ], [ %n.vec, %entry ]
  %lsr.iv34 = phi i64 [ %lsr.iv.next35, %vector.body ], [ 0, %entry ]
  %vec.phi = phi <vscale x 2 x double> [ zeroinitializer, %entry ], [ %30, %vector.body ]
  %vec.phi12 = phi <vscale x 2 x double> [ zeroinitializer, %entry ], [ %31, %vector.body ]
  %vec.phi13 = phi <vscale x 2 x double> [ zeroinitializer, %entry ], [ %26, %vector.body ]
  %vec.phi14 = phi <vscale x 2 x double> [ zeroinitializer, %entry ], [ %27, %vector.body ]
  %scevgep57 = getelementptr i8, ptr %a, i64 %lsr.iv34
  %scevgep64 = getelementptr i8, ptr %scevgep63, i64 %lsr.iv34
  %scevgep58 = getelementptr i8, ptr %b, i64 %lsr.iv34
  %scevgep62 = getelementptr i8, ptr %scevgep61, i64 %lsr.iv34
  %wide.vec = load <vscale x 4 x double>, ptr %scevgep57, align 8
  %wide.vec32 = load <vscale x 4 x double>, ptr %scevgep64, align 8
  %4 = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.experimental.vector.deinterleave2.nxv4f64(<vscale x 4 x double> %wide.vec)
  %5 = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.experimental.vector.deinterleave2.nxv4f64(<vscale x 4 x double> %wide.vec32)
  %6 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %4, 0
  %7 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %5, 0
  %8 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %4, 1
  %9 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %5, 1
  %wide.vec34 = load <vscale x 4 x double>, ptr %scevgep58, align 8
  %wide.vec35 = load <vscale x 4 x double>, ptr %scevgep62, align 8
  %10 = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.experimental.vector.deinterleave2.nxv4f64(<vscale x 4 x double> %wide.vec34)
  %11 = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.experimental.vector.deinterleave2.nxv4f64(<vscale x 4 x double> %wide.vec35)
  %12 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %10, 0
  %13 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %11, 0
  %14 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %10, 1
  %15 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %11, 1
  %16 = fmul fast <vscale x 2 x double> %14, %6
  %17 = fmul fast <vscale x 2 x double> %15, %7
  %18 = fmul fast <vscale x 2 x double> %12, %8
  %19 = fmul fast <vscale x 2 x double> %13, %9
  %20 = fmul fast <vscale x 2 x double> %12, %6
  %21 = fmul fast <vscale x 2 x double> %13, %7
  %22 = fadd fast <vscale x 2 x double> %20, %vec.phi13
  %23 = fadd fast <vscale x 2 x double> %21, %vec.phi14
  %24 = fmul fast <vscale x 2 x double> %14, %8
  %25 = fmul fast <vscale x 2 x double> %15, %9
  %26 = fsub fast <vscale x 2 x double> %22, %24
  %27 = fsub fast <vscale x 2 x double> %23, %25
  %28 = fadd fast <vscale x 2 x double> %18, %vec.phi
  %29 = fadd fast <vscale x 2 x double> %19, %vec.phi12
  %30 = fadd fast <vscale x 2 x double> %28, %16
  %31 = fadd fast <vscale x 2 x double> %29, %17
  %lsr.iv.next35 = add i64 %lsr.iv34, %2
  %lsr.iv.next39 = sub i64 %lsr.iv38, %1
  %32 = icmp eq i64 %lsr.iv.next39, 0
  br i1 %32, label %exit.block, label %vector.body

exit.block:                                     ; preds = %vector.body
  %bin.rdx15 = fadd fast <vscale x 2 x double> %27, %26
  %33 = tail call fast double @llvm.vector.reduce.fadd.nxv2f64(double -0.000000e+00, <vscale x 2 x double> %bin.rdx15)
  %bin.rdx = fadd fast <vscale x 2 x double> %31, %30
  %34 = tail call fast double @llvm.vector.reduce.fadd.nxv2f64(double -0.000000e+00, <vscale x 2 x double> %bin.rdx)
  %.fca.0.0.insert = insertvalue %"class.std::complex" poison, double %33, 0, 0
  %.fca.0.1.insert = insertvalue %"class.std::complex" %.fca.0.0.insert, double %34, 0, 1
  ret %"class.std::complex" %.fca.0.1.insert
}

; Integer and floating point complex number reduction in the same loop:
;   complex<double> *s = ...;
;   int *a = ...;
;
;   for (int i = 0; i < N; ++i) {
;     sum += s[i];
;     int_sum += a[i];
;   }
;
define dso_local %"class.std::complex" @reduction_mix(ptr %a, ptr %b, ptr noalias nocapture noundef readnone %c, [2 x double] %d.coerce, ptr nocapture noundef readonly %s, ptr nocapture noundef writeonly %outs) local_unnamed_addr #0 {
; CHECK-LABEL: reduction_mix:
; CHECK:       // %bb.0: // %entry
; CHECK-NEXT:    mov z2.d, #0 // =0x0
; CHECK-NEXT:    ptrue p0.d
; CHECK-NEXT:    cntd x9
; CHECK-NEXT:    neg x10, x9
; CHECK-NEXT:    mov w11, #100 // =0x64
; CHECK-NEXT:    mov x8, xzr
; CHECK-NEXT:    and x10, x10, x11
; CHECK-NEXT:    rdvl x11, #2
; CHECK-NEXT:    zip2 z0.d, z2.d, z2.d
; CHECK-NEXT:    zip1 z1.d, z2.d, z2.d
; CHECK-NEXT:  .LBB3_1: // %vector.body
; CHECK-NEXT:    // =>This Inner Loop Header: Depth=1
; CHECK-NEXT:    ld1d { z3.d }, p0/z, [x0]
; CHECK-NEXT:    ld1d { z4.d }, p0/z, [x0, #1, mul vl]
; CHECK-NEXT:    add x0, x0, x11
; CHECK-NEXT:    ld1w { z5.d }, p0/z, [x3, x8, lsl #2]
; CHECK-NEXT:    add x8, x8, x9
; CHECK-NEXT:    cmp x10, x8
; CHECK-NEXT:    fadd z0.d, z4.d, z0.d
; CHECK-NEXT:    fadd z1.d, z3.d, z1.d
; CHECK-NEXT:    add z2.d, z5.d, z2.d
; CHECK-NEXT:    b.ne .LBB3_1
; CHECK-NEXT:  // %bb.2: // %middle.block
; CHECK-NEXT:    uaddv d2, p0, z2.d
; CHECK-NEXT:    uzp2 z3.d, z1.d, z0.d
; CHECK-NEXT:    uzp1 z1.d, z1.d, z0.d
; CHECK-NEXT:    fmov x8, d2
; CHECK-NEXT:    faddv d0, p0, z3.d
; CHECK-NEXT:    faddv d1, p0, z1.d
; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $z0
; CHECK-NEXT:    // kill: def $d1 killed $d1 killed $z1
; CHECK-NEXT:    str w8, [x4]
; CHECK-NEXT:    ret
entry:
  %0 = tail call i64 @llvm.vscale.i64()
  %1 = shl nuw nsw i64 %0, 1
  %n.mod.vf = urem i64 100, %1
  %n.vec = sub nuw nsw i64 100, %n.mod.vf
  %2 = tail call i64 @llvm.vscale.i64()
  %3 = shl nuw nsw i64 %2, 1
  br label %vector.body

vector.body:                                      ; preds = %vector.body, %entry
  %index = phi i64 [ 0, %entry ], [ %index.next, %vector.body ]
  %vec.phi = phi <vscale x 2 x i32> [ zeroinitializer, %entry ], [ %5, %vector.body ]
  %vec.phi13 = phi <vscale x 2 x double> [ zeroinitializer, %entry ], [ %9, %vector.body ]
  %vec.phi14 = phi <vscale x 2 x double> [ zeroinitializer, %entry ], [ %10, %vector.body ]
  %4 = getelementptr inbounds i32, ptr %s, i64 %index
  %wide.load = load <vscale x 2 x i32>, ptr %4, align 4
  %5 = add <vscale x 2 x i32> %wide.load, %vec.phi
  %6 = getelementptr inbounds %"class.std::complex", ptr %a, i64 %index
  %wide.vec = load <vscale x 4 x double>, ptr %6, align 8
  %strided.vec = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.experimental.vector.deinterleave2.nxv4f64(<vscale x 4 x double> %wide.vec)
  %7 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %strided.vec, 0
  %8 = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } %strided.vec, 1
  %9 = fadd fast <vscale x 2 x double> %7, %vec.phi13
  %10 = fadd fast <vscale x 2 x double> %8, %vec.phi14
  %index.next = add nuw i64 %index, %3
  %11 = icmp eq i64 %index.next, %n.vec
  br i1 %11, label %middle.block, label %vector.body

middle.block:                                     ; preds = %vector.body
  %12 = tail call fast double @llvm.vector.reduce.fadd.nxv2f64(double -0.000000e+00, <vscale x 2 x double> %10)
  %13 = tail call fast double @llvm.vector.reduce.fadd.nxv2f64(double -0.000000e+00, <vscale x 2 x double> %9)
  %14 = tail call i32 @llvm.vector.reduce.add.nxv2i32(<vscale x 2 x i32> %5)
  store i32 %14, ptr %outs, align 4
  %.fca.0.0.insert = insertvalue %"class.std::complex" poison, double %12, 0, 0
  %.fca.0.1.insert = insertvalue %"class.std::complex" %.fca.0.0.insert, double %13, 0, 1
  ret %"class.std::complex" %.fca.0.1.insert
}


declare i64 @llvm.vscale.i64()
declare { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.experimental.vector.deinterleave2.nxv4f64(<vscale x 4 x double>)
declare double @llvm.vector.reduce.fadd.nxv2f64(double, <vscale x 2 x double>)
declare i32 @llvm.vector.reduce.add.nxv2i32(<vscale x 2 x i32>)