llvm/test/Transforms/LoopVectorize/VPlan/vplan-printing-reductions-tail-folded.ll

; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --filter-out-after "scalar.ph:" --version 6
; RUN: opt -passes=loop-vectorize -force-target-supports-masked-memory-ops -tail-folding-policy=must-fold-tail -vplan-print-after=printOptimizedVPlan -force-vector-interleave=1 -force-vector-width=4 -prefer-inloop-reductions -disable-output %s 2>&1 | FileCheck %s

; Tests for printing VPlans with reductions.

define float @print_reduction(i64 %n, ptr noalias %y) {
; CHECK-LABEL: VPlan for loop in 'print_reduction'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in vp<[[VP3:%[0-9]+]]> = backedge-taken count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:    EMIT vp<[[VP4:%[0-9]+]]> = reduction-start-vector fast ir<0.000000e+00>, ir<0.000000e+00>, ir<1>
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP5:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%red> = phi (fadd) fast vp<[[VP4]]>, ir<%red.next>
; CHECK-NEXT:      vp<[[VP6:%[0-9]+]]> = SCALAR-STEPS vp<[[VP5]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      EMIT vp<[[VP7:%[0-9]+]]> = WIDEN-CANONICAL-INDUCTION nuw vp<[[VP5]]>
; CHECK-NEXT:      EMIT vp<[[VP8:%[0-9]+]]> = icmp ule vp<[[VP7]]>, vp<[[VP3]]>
; CHECK-NEXT:      CLONE ir<%arrayidx> = getelementptr inbounds ir<%y>, vp<[[VP6]]>
; CHECK-NEXT:      vp<[[VP9:%[0-9]+]]> = vector-pointer inbounds ir<%arrayidx>, ir<1>
; CHECK-NEXT:      WIDEN ir<%lv> = load vp<[[VP9]]>, vp<[[VP8]]>
; CHECK-NEXT:      REDUCE ir<%red.next> = ir<%red> + fast  reduce.fadd (ir<%lv>, vp<[[VP8]]>)
; CHECK-NEXT:      EMIT vp<%index.next> = add vp<[[VP5]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP11:%[0-9]+]]> = compute-reduction-result (fadd, in-loop) fast ir<%red.next>
; CHECK-NEXT:  Successor(s): ir-bb<exit>
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:    IR   %red.next.lcssa = phi float [ %red.next, %loop ] (extra operand: vp<[[VP11]]> from middle.block)
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ %iv.next, %loop ], [ 0, %entry ]
  %red = phi float [ %red.next, %loop ], [ 0.0, %entry ]
  %arrayidx = getelementptr inbounds float, ptr %y, i64 %iv
  %lv = load float, ptr %arrayidx, align 4
  %red.next = fadd fast float %lv, %red
  %iv.next = add i64 %iv, 1
  %exitcond = icmp eq i64 %iv.next, %n
  br i1 %exitcond, label %exit, label %loop

exit:
  ret float %red.next
}

define void @print_reduction_with_invariant_store(i64 %n, ptr noalias %y, ptr noalias %dst) {
; CHECK-LABEL: VPlan for loop in 'print_reduction_with_invariant_store'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in vp<[[VP3:%[0-9]+]]> = backedge-taken count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:    EMIT vp<[[VP4:%[0-9]+]]> = reduction-start-vector fast ir<0.000000e+00>, ir<0.000000e+00>, ir<1>
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP5:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%red> = phi (fadd) fast vp<[[VP4]]>, ir<%red.next>
; CHECK-NEXT:      vp<[[VP6:%[0-9]+]]> = SCALAR-STEPS vp<[[VP5]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      EMIT vp<[[VP7:%[0-9]+]]> = WIDEN-CANONICAL-INDUCTION nuw vp<[[VP5]]>
; CHECK-NEXT:      EMIT vp<[[VP8:%[0-9]+]]> = icmp ule vp<[[VP7]]>, vp<[[VP3]]>
; CHECK-NEXT:      CLONE ir<%arrayidx> = getelementptr inbounds ir<%y>, vp<[[VP6]]>
; CHECK-NEXT:      vp<[[VP9:%[0-9]+]]> = vector-pointer inbounds ir<%arrayidx>, ir<1>
; CHECK-NEXT:      WIDEN ir<%lv> = load vp<[[VP9]]>, vp<[[VP8]]>
; CHECK-NEXT:      REDUCE ir<%red.next> = ir<%red> + fast  reduce.fadd (ir<%lv>, vp<[[VP8]]>)
; CHECK-NEXT:      EMIT vp<%index.next> = add vp<[[VP5]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP11:%[0-9]+]]> = compute-reduction-result (fadd, in-loop) fast ir<%red.next>
; CHECK-NEXT:    CLONE store vp<[[VP11]]>, ir<%dst>
; CHECK-NEXT:  Successor(s): ir-bb<exit>
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ %iv.next, %loop ], [ 0, %entry ]
  %red = phi float [ %red.next, %loop ], [ 0.0, %entry ]
  %arrayidx = getelementptr inbounds float, ptr %y, i64 %iv
  %lv = load float, ptr %arrayidx, align 4
  %red.next = fadd fast float %lv, %red
  store float %red.next, ptr %dst, align 4
  %iv.next = add i64 %iv, 1
  %exitcond = icmp eq i64 %iv.next, %n
  br i1 %exitcond, label %exit, label %loop

exit:
  ret void
}

define float @print_fmuladd_strict(ptr %a, ptr %b, i64 %n) {
; CHECK-LABEL: VPlan for loop in 'print_fmuladd_strict'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in vp<[[VP3:%[0-9]+]]> = backedge-taken count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:    EMIT vp<[[VP4:%[0-9]+]]> = reduction-start-vector nnan ninf nsz ir<0.000000e+00>, ir<0.000000e+00>, ir<1>
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP5:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%sum.07> = phi (fmuladd) nnan ninf nsz vp<[[VP4]]>, ir<%muladd>
; CHECK-NEXT:      vp<[[VP6:%[0-9]+]]> = SCALAR-STEPS vp<[[VP5]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      EMIT vp<[[VP7:%[0-9]+]]> = WIDEN-CANONICAL-INDUCTION nuw vp<[[VP5]]>
; CHECK-NEXT:      EMIT vp<[[VP8:%[0-9]+]]> = icmp ule vp<[[VP7]]>, vp<[[VP3]]>
; CHECK-NEXT:      CLONE ir<%arrayidx> = getelementptr inbounds ir<%a>, vp<[[VP6]]>
; CHECK-NEXT:      vp<[[VP9:%[0-9]+]]> = vector-pointer inbounds ir<%arrayidx>, ir<1>
; CHECK-NEXT:      WIDEN ir<%l.a> = load vp<[[VP9]]>, vp<[[VP8]]>
; CHECK-NEXT:      CLONE ir<%arrayidx2> = getelementptr inbounds ir<%b>, vp<[[VP6]]>
; CHECK-NEXT:      vp<[[VP10:%[0-9]+]]> = vector-pointer inbounds ir<%arrayidx2>, ir<1>
; CHECK-NEXT:      WIDEN ir<%l.b> = load vp<[[VP10]]>, vp<[[VP8]]>
; CHECK-NEXT:      EMIT vp<[[VP11:%[0-9]+]]> = fmul nnan ninf nsz ir<%l.a>, ir<%l.b>
; CHECK-NEXT:      REDUCE ir<%muladd> = ir<%sum.07> + nnan ninf nsz  reduce.fmuladd (vp<[[VP11]]>, vp<[[VP8]]>)
; CHECK-NEXT:      EMIT vp<%index.next> = add vp<[[VP5]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP13:%[0-9]+]]> = compute-reduction-result (fmuladd, in-loop) nnan ninf nsz ir<%muladd>
; CHECK-NEXT:  Successor(s): ir-bb<exit>
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:    IR   %muladd.lcssa = phi float [ %muladd, %loop ] (extra operand: vp<[[VP13]]> from middle.block)
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;

entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %sum.07 = phi float [ 0.000000e+00, %entry ], [ %muladd, %loop ]
  %arrayidx = getelementptr inbounds float, ptr %a, i64 %iv
  %l.a = load float, ptr %arrayidx, align 4
  %arrayidx2 = getelementptr inbounds float, ptr %b, i64 %iv
  %l.b = load float, ptr %arrayidx2, align 4
  %muladd = tail call nnan ninf nsz float @llvm.fmuladd.f32(float %l.a, float %l.b, float %sum.07)
  %iv.next = add nuw nsw i64 %iv, 1
  %exitcond.not = icmp eq i64 %iv.next, %n
  br i1 %exitcond.not, label %exit, label %loop

exit:
  ret float %muladd
}

define i64 @find_last_iv(ptr %a, i64 %n, i64 %start) {
; CHECK-LABEL: VPlan for loop in 'find_last_iv'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in vp<[[VP3:%[0-9]+]]> = backedge-taken count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP4:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      ir<%iv> = WIDEN-INDUCTION nuw nsw ir<0>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%rdx> = phi (find-iv) ir<-9223372036854775808>, vp<[[VP9:%[0-9]+]]>
; CHECK-NEXT:      vp<[[VP5:%[0-9]+]]> = SCALAR-STEPS vp<[[VP4]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      EMIT vp<[[VP6:%[0-9]+]]> = WIDEN-CANONICAL-INDUCTION nuw vp<[[VP4]]>
; CHECK-NEXT:      EMIT vp<[[VP7:%[0-9]+]]> = icmp ule vp<[[VP6]]>, vp<[[VP3]]>
; CHECK-NEXT:      CLONE ir<%gep.a> = getelementptr inbounds ir<%a>, vp<[[VP5]]>
; CHECK-NEXT:      vp<[[VP8:%[0-9]+]]> = vector-pointer inbounds ir<%gep.a>, ir<1>
; CHECK-NEXT:      WIDEN ir<%l.a> = load vp<[[VP8]]>, vp<[[VP7]]>
; CHECK-NEXT:      WIDEN ir<%cmp2> = icmp eq ir<%l.a>, ir<%start>
; CHECK-NEXT:      WIDEN ir<%cond> = select ir<%cmp2>, ir<%iv>, ir<%rdx>
; CHECK-NEXT:      EMIT vp<[[VP9]]> = select vp<[[VP7]]>, ir<%cond>, ir<%rdx>
; CHECK-NEXT:      EMIT vp<%index.next> = add vp<[[VP4]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP11:%[0-9]+]]> = compute-reduction-result (smax) vp<[[VP9]]>
; CHECK-NEXT:    EMIT vp<[[VP12:%[0-9]+]]> = icmp ne vp<[[VP11]]>, ir<-9223372036854775808>
; CHECK-NEXT:    EMIT vp<[[VP13:%[0-9]+]]> = select vp<[[VP12]]>, vp<[[VP11]]>, ir<%start>
; CHECK-NEXT:  Successor(s): ir-bb<exit>
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:    IR   %cond.lcssa = phi i64 [ %cond, %loop ] (extra operand: vp<[[VP13]]> from middle.block)
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %inc, %loop ]
  %rdx = phi i64 [ %start, %entry ], [ %cond, %loop ]
  %gep.a = getelementptr inbounds i64, ptr %a, i64 %iv
  %l.a = load i64, ptr %gep.a, align 8
  %cmp2 = icmp eq i64 %l.a, %start
  %cond = select i1 %cmp2, i64 %iv, i64 %rdx
  %inc = add nuw nsw i64 %iv, 1
  %exitcond.not = icmp eq i64 %inc, %n
  br i1 %exitcond.not, label %exit, label %loop

exit:
  ret i64 %cond
}

define i64 @print_extended_reduction(ptr nocapture readonly %x, ptr nocapture readonly %y, i32 %n) {
; CHECK-LABEL: VPlan for loop in 'print_extended_reduction'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in vp<[[VP3:%[0-9]+]]> = backedge-taken count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:    EMIT vp<[[VP4:%[0-9]+]]> = reduction-start-vector ir<0>, ir<0>, ir<1>
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP5:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%rdx> = phi (add) vp<[[VP4]]>, vp<[[VP10:%[0-9]+]]>
; CHECK-NEXT:      vp<[[VP6:%[0-9]+]]> = SCALAR-STEPS vp<[[VP5]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      EMIT vp<[[VP7:%[0-9]+]]> = WIDEN-CANONICAL-INDUCTION nuw vp<[[VP5]]>
; CHECK-NEXT:      EMIT vp<[[VP8:%[0-9]+]]> = icmp ule vp<[[VP7]]>, vp<[[VP3]]>
; CHECK-NEXT:      CLONE ir<%arrayidx> = getelementptr inbounds ir<%x>, vp<[[VP6]]>
; CHECK-NEXT:      vp<[[VP9:%[0-9]+]]> = vector-pointer inbounds ir<%arrayidx>, ir<1>
; CHECK-NEXT:      WIDEN ir<%load0> = load vp<[[VP9]]>, vp<[[VP8]]>
; CHECK-NEXT:      EXPRESSION vp<[[VP10]]> = vp<[[VP8]]> + reduce.add (ir<%load0> zext to i64, vp<[[VP8]]>)
; CHECK-NEXT:      EMIT vp<%index.next> = add vp<[[VP5]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP12:%[0-9]+]]> = compute-reduction-result (add, in-loop) vp<[[VP10]]>
; CHECK-NEXT:  Successor(s): ir-bb<exit>
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:    IR   %r.0.lcssa = phi i64 [ %rdx.next, %loop ] (extra operand: vp<[[VP12]]> from middle.block)
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;
entry:
  br label %loop

loop:
  %iv = phi i32 [ %iv.next, %loop ], [ 0, %entry ]
  %rdx = phi i64 [ %rdx.next, %loop ], [ 0, %entry ]
  %arrayidx = getelementptr inbounds i32, ptr %x, i32 %iv
  %load0 = load i32, ptr %arrayidx, align 4
  %conv0 = zext i32 %load0 to i64
  %rdx.next = add nsw i64 %rdx, %conv0
  %iv.next = add nuw nsw i32 %iv, 1
  %exitcond = icmp eq i32 %iv.next, %n
  br i1 %exitcond, label %exit, label %loop

exit:
  %r.0.lcssa = phi i64 [ %rdx.next, %loop ]
  ret i64 %r.0.lcssa
}

define i64 @print_mulacc(ptr nocapture readonly %x, ptr nocapture readonly %y, i32 %n) {
; CHECK-LABEL: VPlan for loop in 'print_mulacc'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in vp<[[VP3:%[0-9]+]]> = backedge-taken count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:    EMIT vp<[[VP4:%[0-9]+]]> = reduction-start-vector ir<0>, ir<0>, ir<1>
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP5:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%rdx> = phi (add) vp<[[VP4]]>, vp<[[VP11:%[0-9]+]]>
; CHECK-NEXT:      vp<[[VP6:%[0-9]+]]> = SCALAR-STEPS vp<[[VP5]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      EMIT vp<[[VP7:%[0-9]+]]> = WIDEN-CANONICAL-INDUCTION nuw vp<[[VP5]]>
; CHECK-NEXT:      EMIT vp<[[VP8:%[0-9]+]]> = icmp ule vp<[[VP7]]>, vp<[[VP3]]>
; CHECK-NEXT:      CLONE ir<%arrayidx> = getelementptr inbounds ir<%x>, vp<[[VP6]]>
; CHECK-NEXT:      vp<[[VP9:%[0-9]+]]> = vector-pointer inbounds ir<%arrayidx>, ir<1>
; CHECK-NEXT:      WIDEN ir<%load0> = load vp<[[VP9]]>, vp<[[VP8]]>
; CHECK-NEXT:      CLONE ir<%arrayidx1> = getelementptr inbounds ir<%y>, vp<[[VP6]]>
; CHECK-NEXT:      vp<[[VP10:%[0-9]+]]> = vector-pointer inbounds ir<%arrayidx1>, ir<1>
; CHECK-NEXT:      WIDEN ir<%load1> = load vp<[[VP10]]>, vp<[[VP8]]>
; CHECK-NEXT:      EXPRESSION vp<[[VP11]]> = ir<%rdx> + reduce.add (mul nsw ir<%load0>, ir<%load1>, vp<[[VP8]]>)
; CHECK-NEXT:      EMIT vp<%index.next> = add vp<[[VP5]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP13:%[0-9]+]]> = compute-reduction-result (add, in-loop) vp<[[VP11]]>
; CHECK-NEXT:  Successor(s): ir-bb<exit>
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:    IR   %r.0.lcssa = phi i64 [ %rdx.next, %loop ] (extra operand: vp<[[VP13]]> from middle.block)
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;
entry:
  br label %loop

loop:
  %iv = phi i32 [ %iv.next, %loop ], [ 0, %entry ]
  %rdx = phi i64 [ %rdx.next, %loop ], [ 0, %entry ]
  %arrayidx = getelementptr inbounds i64, ptr %x, i32 %iv
  %load0 = load i64, ptr %arrayidx, align 4
  %arrayidx1 = getelementptr inbounds i64, ptr %y, i32 %iv
  %load1 = load i64, ptr %arrayidx1, align 4
  %mul = mul nsw i64 %load0, %load1
  %rdx.next = add nsw i64 %rdx, %mul
  %iv.next = add nuw nsw i32 %iv, 1
  %exitcond = icmp eq i32 %iv.next, %n
  br i1 %exitcond, label %exit, label %loop

exit:
  %r.0.lcssa = phi i64 [ %rdx.next, %loop ]
  ret i64 %r.0.lcssa
}

define i64 @print_mulacc_extended(ptr nocapture readonly %x, ptr nocapture readonly %y, i32 %n) {
; CHECK-LABEL: VPlan for loop in 'print_mulacc_extended'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in vp<[[VP3:%[0-9]+]]> = backedge-taken count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:    EMIT vp<[[VP4:%[0-9]+]]> = reduction-start-vector ir<0>, ir<0>, ir<1>
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP5:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%rdx> = phi (add) vp<[[VP4]]>, vp<[[VP11:%[0-9]+]]>
; CHECK-NEXT:      vp<[[VP6:%[0-9]+]]> = SCALAR-STEPS vp<[[VP5]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      EMIT vp<[[VP7:%[0-9]+]]> = WIDEN-CANONICAL-INDUCTION nuw vp<[[VP5]]>
; CHECK-NEXT:      EMIT vp<[[VP8:%[0-9]+]]> = icmp ule vp<[[VP7]]>, vp<[[VP3]]>
; CHECK-NEXT:      CLONE ir<%arrayidx> = getelementptr inbounds ir<%x>, vp<[[VP6]]>
; CHECK-NEXT:      vp<[[VP9:%[0-9]+]]> = vector-pointer inbounds ir<%arrayidx>, ir<1>
; CHECK-NEXT:      WIDEN ir<%load0> = load vp<[[VP9]]>, vp<[[VP8]]>
; CHECK-NEXT:      CLONE ir<%arrayidx1> = getelementptr inbounds ir<%y>, vp<[[VP6]]>
; CHECK-NEXT:      vp<[[VP10:%[0-9]+]]> = vector-pointer inbounds ir<%arrayidx1>, ir<1>
; CHECK-NEXT:      WIDEN ir<%load1> = load vp<[[VP10]]>, vp<[[VP8]]>
; CHECK-NEXT:      EXPRESSION vp<[[VP11]]> = ir<%rdx> + reduce.add (mul nsw (ir<%load0> sext to i64), (ir<%load1> sext to i64), vp<[[VP8]]>)
; CHECK-NEXT:      EMIT vp<%index.next> = add vp<[[VP5]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP13:%[0-9]+]]> = compute-reduction-result (add, in-loop) vp<[[VP11]]>
; CHECK-NEXT:  Successor(s): ir-bb<exit>
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:    IR   %r.0.lcssa = phi i64 [ %rdx.next, %loop ] (extra operand: vp<[[VP13]]> from middle.block)
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;
entry:
  br label %loop

loop:
  %iv = phi i32 [ %iv.next, %loop ], [ 0, %entry ]
  %rdx = phi i64 [ %rdx.next, %loop ], [ 0, %entry ]
  %arrayidx = getelementptr inbounds i16, ptr %x, i32 %iv
  %load0 = load i16, ptr %arrayidx, align 4
  %arrayidx1 = getelementptr inbounds i16, ptr %y, i32 %iv
  %load1 = load i16, ptr %arrayidx1, align 4
  %conv0 = sext i16 %load0 to i32
  %conv1 = sext i16 %load1 to i32
  %mul = mul nsw i32 %conv0, %conv1
  %conv = sext i32 %mul to i64
  %rdx.next = add nsw i64 %rdx, %conv
  %iv.next = add nuw nsw i32 %iv, 1
  %exitcond = icmp eq i32 %iv.next, %n
  br i1 %exitcond, label %exit, label %loop

exit:
  %r.0.lcssa = phi i64 [ %rdx.next, %loop ]
  ret i64 %r.0.lcssa
}

define i64 @print_extended_sub_reduction(ptr nocapture readonly %x, ptr nocapture readonly %y, i32 %n) {
; CHECK-LABEL: VPlan for loop in 'print_extended_sub_reduction'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in vp<[[VP3:%[0-9]+]]> = backedge-taken count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:    EMIT vp<[[VP4:%[0-9]+]]> = reduction-start-vector ir<0>, ir<0>, ir<1>
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP5:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%rdx> = phi (sub) vp<[[VP4]]>, vp<[[VP10:%[0-9]+]]>
; CHECK-NEXT:      vp<[[VP6:%[0-9]+]]> = SCALAR-STEPS vp<[[VP5]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      EMIT vp<[[VP7:%[0-9]+]]> = WIDEN-CANONICAL-INDUCTION nuw vp<[[VP5]]>
; CHECK-NEXT:      EMIT vp<[[VP8:%[0-9]+]]> = icmp ule vp<[[VP7]]>, vp<[[VP3]]>
; CHECK-NEXT:      CLONE ir<%arrayidx> = getelementptr inbounds ir<%x>, vp<[[VP6]]>
; CHECK-NEXT:      vp<[[VP9:%[0-9]+]]> = vector-pointer inbounds ir<%arrayidx>, ir<1>
; CHECK-NEXT:      WIDEN ir<%load0> = load vp<[[VP9]]>, vp<[[VP8]]>
; CHECK-NEXT:      EXPRESSION vp<[[VP10]]> = vp<[[VP8]]> + reduce.sub (ir<%load0> zext to i64, vp<[[VP8]]>)
; CHECK-NEXT:      EMIT vp<%index.next> = add vp<[[VP5]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP12:%[0-9]+]]> = compute-reduction-result (sub, in-loop) vp<[[VP10]]>
; CHECK-NEXT:  Successor(s): ir-bb<exit>
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:    IR   %r.0.lcssa = phi i64 [ %rdx.next, %loop ] (extra operand: vp<[[VP12]]> from middle.block)
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;
entry:
  br label %loop

loop:
  %iv = phi i32 [ %iv.next, %loop ], [ 0, %entry ]
  %rdx = phi i64 [ %rdx.next, %loop ], [ 0, %entry ]
  %arrayidx = getelementptr inbounds i32, ptr %x, i32 %iv
  %load0 = load i32, ptr %arrayidx, align 4
  %conv0 = zext i32 %load0 to i64
  %rdx.next = sub nsw i64 %rdx, %conv0
  %iv.next = add nuw nsw i32 %iv, 1
  %exitcond = icmp eq i32 %iv.next, %n
  br i1 %exitcond, label %exit, label %loop

exit:
  %r.0.lcssa = phi i64 [ %rdx.next, %loop ]
  ret i64 %r.0.lcssa
}

define i32 @print_mulacc_sub(ptr %a, ptr %b) {
; CHECK-LABEL: VPlan for loop in 'print_mulacc_sub'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in ir<1024> = original trip-count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:    EMIT vp<[[VP3:%[0-9]+]]> = reduction-start-vector ir<0>, ir<0>, ir<1>
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP4:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%accum> = phi (sub) vp<[[VP3]]>, vp<[[VP8:%[0-9]+]]>
; CHECK-NEXT:      vp<[[VP5:%[0-9]+]]> = SCALAR-STEPS vp<[[VP4]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      CLONE ir<%gep.a> = getelementptr ir<%a>, vp<[[VP5]]>
; CHECK-NEXT:      vp<[[VP6:%[0-9]+]]> = vector-pointer ir<%gep.a>, ir<1>
; CHECK-NEXT:      WIDEN ir<%load.a> = load vp<[[VP6]]>
; CHECK-NEXT:      CLONE ir<%gep.b> = getelementptr ir<%b>, vp<[[VP5]]>
; CHECK-NEXT:      vp<[[VP7:%[0-9]+]]> = vector-pointer ir<%gep.b>, ir<1>
; CHECK-NEXT:      WIDEN ir<%load.b> = load vp<[[VP7]]>
; CHECK-NEXT:      EXPRESSION vp<[[VP8]]> = ir<%accum> + reduce.sub (mul (ir<%load.b> zext to i32), (ir<%load.a> zext to i32))
; CHECK-NEXT:      EMIT vp<%index.next> = add nuw vp<[[VP4]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP10:%[0-9]+]]> = compute-reduction-result (sub, in-loop) vp<[[VP8]]>
; CHECK-NEXT:    EMIT vp<%cmp.n> = icmp eq ir<1024>, vp<[[VP2]]>
; CHECK-NEXT:    EMIT branch-on-cond vp<%cmp.n>
; CHECK-NEXT:  Successor(s): ir-bb<exit>, scalar.ph
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:    IR   %add.lcssa = phi i32 [ %add, %loop ] (extra operand: vp<[[VP10]]> from middle.block)
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %accum = phi i32 [ 0, %entry ], [ %add, %loop ]
  %gep.a = getelementptr i8, ptr %a, i64 %iv
  %load.a = load i8, ptr %gep.a, align 1
  %ext.a = zext i8 %load.a to i32
  %gep.b = getelementptr i8, ptr %b, i64 %iv
  %load.b = load i8, ptr %gep.b, align 1
  %ext.b = zext i8 %load.b to i32
  %mul = mul i32 %ext.b, %ext.a
  %add = sub i32 %accum, %mul
  %iv.next = add i64 %iv, 1
  %exitcond.not = icmp eq i64 %iv.next, 1024
  br i1 %exitcond.not, label %exit, label %loop

exit:
  ret i32 %add
}

define i32 @print_mulacc_negated(ptr %a, ptr %b) {
; CHECK-LABEL: VPlan for loop in 'print_mulacc_negated'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in ir<1024> = original trip-count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:    EMIT vp<[[VP3:%[0-9]+]]> = reduction-start-vector ir<0>, ir<0>, ir<1>
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP4:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%accum> = phi (add) vp<[[VP3]]>, vp<[[VP8:%[0-9]+]]>
; CHECK-NEXT:      vp<[[VP5:%[0-9]+]]> = SCALAR-STEPS vp<[[VP4]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      CLONE ir<%gep.a> = getelementptr ir<%a>, vp<[[VP5]]>
; CHECK-NEXT:      vp<[[VP6:%[0-9]+]]> = vector-pointer ir<%gep.a>, ir<1>
; CHECK-NEXT:      WIDEN ir<%load.a> = load vp<[[VP6]]>
; CHECK-NEXT:      CLONE ir<%gep.b> = getelementptr ir<%b>, vp<[[VP5]]>
; CHECK-NEXT:      vp<[[VP7:%[0-9]+]]> = vector-pointer ir<%gep.b>, ir<1>
; CHECK-NEXT:      WIDEN ir<%load.b> = load vp<[[VP7]]>
; CHECK-NEXT:      EXPRESSION vp<[[VP8]]> = ir<%accum> + reduce.add (sub (0, mul (ir<%load.b> zext to i32), (ir<%load.a> zext to i32)))
; CHECK-NEXT:      EMIT vp<%index.next> = add nuw vp<[[VP4]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP10:%[0-9]+]]> = compute-reduction-result (add, in-loop) vp<[[VP8]]>
; CHECK-NEXT:    EMIT vp<%cmp.n> = icmp eq ir<1024>, vp<[[VP2]]>
; CHECK-NEXT:    EMIT branch-on-cond vp<%cmp.n>
; CHECK-NEXT:  Successor(s): ir-bb<exit>, scalar.ph
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:    IR   %add.lcssa = phi i32 [ %add, %loop ] (extra operand: vp<[[VP10]]> from middle.block)
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %accum = phi i32 [ 0, %entry ], [ %add, %loop ]
  %gep.a = getelementptr i8, ptr %a, i64 %iv
  %load.a = load i8, ptr %gep.a, align 1
  %ext.a = zext i8 %load.a to i32
  %gep.b = getelementptr i8, ptr %b, i64 %iv
  %load.b = load i8, ptr %gep.b, align 1
  %ext.b = zext i8 %load.b to i32
  %mul = mul i32 %ext.b, %ext.a
  %sub = sub i32 0, %mul
  %add = add i32 %accum, %sub
  %iv.next = add i64 %iv, 1
  %exitcond.not = icmp eq i64 %iv.next, 1024
  br i1 %exitcond.not, label %exit, label %loop

exit:
  ret i32 %add
}

define i64 @print_mulacc_sub_extended(ptr nocapture readonly %x, ptr nocapture readonly %y, i32 %n) {
; CHECK-LABEL: VPlan for loop in 'print_mulacc_sub_extended'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in vp<[[VP3:%[0-9]+]]> = backedge-taken count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:    EMIT vp<[[VP4:%[0-9]+]]> = reduction-start-vector ir<0>, ir<0>, ir<1>
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP5:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%rdx> = phi (sub) vp<[[VP4]]>, vp<[[VP11:%[0-9]+]]>
; CHECK-NEXT:      vp<[[VP6:%[0-9]+]]> = SCALAR-STEPS vp<[[VP5]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      EMIT vp<[[VP7:%[0-9]+]]> = WIDEN-CANONICAL-INDUCTION nuw vp<[[VP5]]>
; CHECK-NEXT:      EMIT vp<[[VP8:%[0-9]+]]> = icmp ule vp<[[VP7]]>, vp<[[VP3]]>
; CHECK-NEXT:      CLONE ir<%arrayidx> = getelementptr inbounds ir<%x>, vp<[[VP6]]>
; CHECK-NEXT:      vp<[[VP9:%[0-9]+]]> = vector-pointer inbounds ir<%arrayidx>, ir<1>
; CHECK-NEXT:      WIDEN ir<%load0> = load vp<[[VP9]]>, vp<[[VP8]]>
; CHECK-NEXT:      CLONE ir<%arrayidx1> = getelementptr inbounds ir<%y>, vp<[[VP6]]>
; CHECK-NEXT:      vp<[[VP10:%[0-9]+]]> = vector-pointer inbounds ir<%arrayidx1>, ir<1>
; CHECK-NEXT:      WIDEN ir<%load1> = load vp<[[VP10]]>, vp<[[VP8]]>
; CHECK-NEXT:      EXPRESSION vp<[[VP11]]> = ir<%rdx> + reduce.sub (mul nsw (ir<%load0> sext to i64), (ir<%load1> sext to i64), vp<[[VP8]]>)
; CHECK-NEXT:      EMIT vp<%index.next> = add vp<[[VP5]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP13:%[0-9]+]]> = compute-reduction-result (sub, in-loop) vp<[[VP11]]>
; CHECK-NEXT:  Successor(s): ir-bb<exit>
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:    IR   %r.0.lcssa = phi i64 [ %rdx.next, %loop ] (extra operand: vp<[[VP13]]> from middle.block)
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;
entry:
  br label %loop

loop:
  %iv = phi i32 [ %iv.next, %loop ], [ 0, %entry ]
  %rdx = phi i64 [ %rdx.next, %loop ], [ 0, %entry ]
  %arrayidx = getelementptr inbounds i16, ptr %x, i32 %iv
  %load0 = load i16, ptr %arrayidx, align 4
  %arrayidx1 = getelementptr inbounds i16, ptr %y, i32 %iv
  %load1 = load i16, ptr %arrayidx1, align 4
  %conv0 = sext i16 %load0 to i32
  %conv1 = sext i16 %load1 to i32
  %mul = mul nsw i32 %conv0, %conv1
  %conv = sext i32 %mul to i64
  %rdx.next = sub nsw i64 %rdx, %conv
  %iv.next = add nuw nsw i32 %iv, 1
  %exitcond = icmp eq i32 %iv.next, %n
  br i1 %exitcond, label %exit, label %loop

exit:
  %r.0.lcssa = phi i64 [ %rdx.next, %loop ]
  ret i64 %r.0.lcssa
}

define i64 @print_mulacc_duplicate_extends(ptr nocapture readonly %x, ptr nocapture readonly %y, i32 %n) {
; CHECK-LABEL: VPlan for loop in 'print_mulacc_duplicate_extends'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in vp<[[VP3:%[0-9]+]]> = backedge-taken count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:    EMIT vp<[[VP4:%[0-9]+]]> = reduction-start-vector ir<0>, ir<0>, ir<1>
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP5:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%rdx> = phi (sub) vp<[[VP4]]>, vp<[[VP10:%[0-9]+]]>
; CHECK-NEXT:      vp<[[VP6:%[0-9]+]]> = SCALAR-STEPS vp<[[VP5]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      EMIT vp<[[VP7:%[0-9]+]]> = WIDEN-CANONICAL-INDUCTION nuw vp<[[VP5]]>
; CHECK-NEXT:      EMIT vp<[[VP8:%[0-9]+]]> = icmp ule vp<[[VP7]]>, vp<[[VP3]]>
; CHECK-NEXT:      CLONE ir<%arrayidx> = getelementptr inbounds ir<%x>, vp<[[VP6]]>
; CHECK-NEXT:      vp<[[VP9:%[0-9]+]]> = vector-pointer inbounds ir<%arrayidx>, ir<1>
; CHECK-NEXT:      WIDEN ir<%load0> = load vp<[[VP9]]>, vp<[[VP8]]>
; CHECK-NEXT:      EXPRESSION vp<[[VP10]]> = ir<%rdx> + reduce.sub (mul nsw (ir<%load0> sext to i64), (ir<%load0> sext to i64), vp<[[VP8]]>)
; CHECK-NEXT:      EMIT vp<%index.next> = add vp<[[VP5]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP12:%[0-9]+]]> = compute-reduction-result (sub, in-loop) vp<[[VP10]]>
; CHECK-NEXT:  Successor(s): ir-bb<exit>
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:    IR   %r.0.lcssa = phi i64 [ %rdx.next, %loop ] (extra operand: vp<[[VP12]]> from middle.block)
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;
entry:
  br label %loop

loop:
  %iv = phi i32 [ %iv.next, %loop ], [ 0, %entry ]
  %rdx = phi i64 [ %rdx.next, %loop ], [ 0, %entry ]
  %arrayidx = getelementptr inbounds i16, ptr %x, i32 %iv
  %load0 = load i16, ptr %arrayidx, align 4
  %conv0 = sext i16 %load0 to i32
  %mul = mul nsw i32 %conv0, %conv0
  %conv = sext i32 %mul to i64
  %rdx.next = sub nsw i64 %rdx, %conv
  %iv.next = add nuw nsw i32 %iv, 1
  %exitcond = icmp eq i32 %iv.next, %n
  br i1 %exitcond, label %exit, label %loop

exit:
  %r.0.lcssa = phi i64 [ %rdx.next, %loop ]
  ret i64 %r.0.lcssa
}

define i32 @print_mulacc_extended_const(ptr %start, ptr %end) {
; CHECK-LABEL: VPlan for loop in 'print_mulacc_extended_const'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in vp<[[VP3:%[0-9]+]]> = backedge-taken count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:    EMIT vp<[[VP4:%[0-9]+]]> = EXPAND SCEV (1 + (-1 * (ptrtoint ptr %start to i64)) + (ptrtoint ptr %end to i64))
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:    EMIT vp<[[VP5:%[0-9]+]]> = reduction-start-vector ir<0>, ir<0>, ir<1>
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP6:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%red> = phi (add) vp<[[VP5]]>, vp<[[VP11:%[0-9]+]]>
; CHECK-NEXT:      vp<[[VP7:%[0-9]+]]> = SCALAR-STEPS vp<[[VP6]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      EMIT vp<%next.gep> = ptradd ir<%start>, vp<[[VP7]]>
; CHECK-NEXT:      EMIT vp<[[VP8:%[0-9]+]]> = WIDEN-CANONICAL-INDUCTION nuw vp<[[VP6]]>
; CHECK-NEXT:      EMIT vp<[[VP9:%[0-9]+]]> = icmp ule vp<[[VP8]]>, vp<[[VP3]]>
; CHECK-NEXT:      vp<[[VP10:%[0-9]+]]> = vector-pointer vp<%next.gep>, ir<1>
; CHECK-NEXT:      WIDEN ir<%l> = load vp<[[VP10]]>, vp<[[VP9]]>
; CHECK-NEXT:      EXPRESSION vp<[[VP11]]> = ir<%red> + reduce.add (mul (ir<%l> zext to i32), (ir<63> zext to i32), vp<[[VP9]]>)
; CHECK-NEXT:      EMIT vp<%index.next> = add vp<[[VP6]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP13:%[0-9]+]]> = compute-reduction-result (add, in-loop) vp<[[VP11]]>
; CHECK-NEXT:  Successor(s): ir-bb<exit>
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:    IR   %red.next.lcssa = phi i32 [ %red.next, %loop ] (extra operand: vp<[[VP13]]> from middle.block)
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;
entry:
  br label %loop

loop:
  %ptr.iv = phi ptr [ %start, %entry ], [ %gep.iv.next, %loop ]
  %red = phi i32 [ 0, %entry ], [ %red.next, %loop ]
  %l = load i8, ptr %ptr.iv, align 1
  %l.ext = zext i8 %l to i32
  %mul = mul i32 %l.ext, 63
  %red.next = add i32 %red, %mul
  %gep.iv.next = getelementptr i8, ptr %ptr.iv, i64 1
  %ec = icmp eq ptr %ptr.iv, %end
  br i1 %ec, label %exit, label %loop

exit:
  ret i32 %red.next
}

define i32 @print_mulacc_extended_const_lhs(ptr %start, ptr %end) {
; CHECK-LABEL: VPlan for loop in 'print_mulacc_extended_const_lhs'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in vp<[[VP3:%[0-9]+]]> = backedge-taken count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:    EMIT vp<[[VP4:%[0-9]+]]> = EXPAND SCEV (1 + (-1 * (ptrtoint ptr %start to i64)) + (ptrtoint ptr %end to i64))
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:    EMIT vp<[[VP5:%[0-9]+]]> = reduction-start-vector ir<0>, ir<0>, ir<1>
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP6:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%red> = phi (add) vp<[[VP5]]>, vp<[[VP11:%[0-9]+]]>
; CHECK-NEXT:      vp<[[VP7:%[0-9]+]]> = SCALAR-STEPS vp<[[VP6]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      EMIT vp<%next.gep> = ptradd ir<%start>, vp<[[VP7]]>
; CHECK-NEXT:      EMIT vp<[[VP8:%[0-9]+]]> = WIDEN-CANONICAL-INDUCTION nuw vp<[[VP6]]>
; CHECK-NEXT:      EMIT vp<[[VP9:%[0-9]+]]> = icmp ule vp<[[VP8]]>, vp<[[VP3]]>
; CHECK-NEXT:      vp<[[VP10:%[0-9]+]]> = vector-pointer vp<%next.gep>, ir<1>
; CHECK-NEXT:      WIDEN ir<%l> = load vp<[[VP10]]>, vp<[[VP9]]>
; CHECK-NEXT:      WIDEN-CAST ir<%l.ext> = zext ir<%l> to i32
; CHECK-NEXT:      EXPRESSION vp<[[VP11]]> = ir<%red> + reduce.add (mul ir<63>, ir<%l.ext>, vp<[[VP9]]>)
; CHECK-NEXT:      EMIT vp<%index.next> = add vp<[[VP6]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP13:%[0-9]+]]> = compute-reduction-result (add, in-loop) vp<[[VP11]]>
; CHECK-NEXT:  Successor(s): ir-bb<exit>
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:    IR   %red.next.lcssa = phi i32 [ %red.next, %loop ] (extra operand: vp<[[VP13]]> from middle.block)
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;
entry:
  br label %loop

loop:
  %ptr.iv = phi ptr [ %start, %entry ], [ %gep.iv.next, %loop ]
  %red = phi i32 [ 0, %entry ], [ %red.next, %loop ]
  %l = load i8, ptr %ptr.iv, align 1
  %l.ext = zext i8 %l to i32
  %mul = mul i32 63, %l.ext
  %red.next = add i32 %red, %mul
  %gep.iv.next = getelementptr i8, ptr %ptr.iv, i64 1
  %ec = icmp eq ptr %ptr.iv, %end
  br i1 %ec, label %exit, label %loop

exit:
  ret i32 %red.next
}

; Constants >= 128 cannot be treated as sign-extended, so the expression shouldn't extend 128
define i32 @print_mulacc_not_extended_const(ptr %start, ptr %end) {
; CHECK-LABEL: VPlan for loop in 'print_mulacc_not_extended_const'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in vp<[[VP3:%[0-9]+]]> = backedge-taken count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:    EMIT vp<[[VP4:%[0-9]+]]> = EXPAND SCEV (1 + (-1 * (ptrtoint ptr %start to i64)) + (ptrtoint ptr %end to i64))
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:    EMIT vp<[[VP5:%[0-9]+]]> = reduction-start-vector ir<0>, ir<0>, ir<1>
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP6:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%red> = phi (add) vp<[[VP5]]>, vp<[[VP11:%[0-9]+]]>
; CHECK-NEXT:      vp<[[VP7:%[0-9]+]]> = SCALAR-STEPS vp<[[VP6]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      EMIT vp<%next.gep> = ptradd ir<%start>, vp<[[VP7]]>
; CHECK-NEXT:      EMIT vp<[[VP8:%[0-9]+]]> = WIDEN-CANONICAL-INDUCTION nuw vp<[[VP6]]>
; CHECK-NEXT:      EMIT vp<[[VP9:%[0-9]+]]> = icmp ule vp<[[VP8]]>, vp<[[VP3]]>
; CHECK-NEXT:      vp<[[VP10:%[0-9]+]]> = vector-pointer vp<%next.gep>, ir<1>
; CHECK-NEXT:      WIDEN ir<%l> = load vp<[[VP10]]>, vp<[[VP9]]>
; CHECK-NEXT:      WIDEN-CAST ir<%l.ext> = sext ir<%l> to i32
; CHECK-NEXT:      EXPRESSION vp<[[VP11]]> = ir<%red> + reduce.add (mul ir<%l.ext>, ir<128>, vp<[[VP9]]>)
; CHECK-NEXT:      EMIT vp<%index.next> = add vp<[[VP6]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP13:%[0-9]+]]> = compute-reduction-result (add, in-loop) vp<[[VP11]]>
; CHECK-NEXT:  Successor(s): ir-bb<exit>
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:    IR   %red.next.lcssa = phi i32 [ %red.next, %loop ] (extra operand: vp<[[VP13]]> from middle.block)
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;
entry:
  br label %loop

loop:
  %ptr.iv = phi ptr [ %start, %entry ], [ %gep.iv.next, %loop ]
  %red = phi i32 [ 0, %entry ], [ %red.next, %loop ]
  %l = load i8, ptr %ptr.iv, align 1
  %l.ext = sext i8 %l to i32
  %mul = mul i32 %l.ext, 128
  %red.next = add i32 %red, %mul
  %gep.iv.next = getelementptr i8, ptr %ptr.iv, i64 1
  %ec = icmp eq ptr %ptr.iv, %end
  br i1 %ec, label %exit, label %loop

exit:
  %red.next.lcssa = phi i32 [ %red.next, %loop ]
  ret i32 %red.next.lcssa
}

define i64 @print_ext_mulacc_extended_const(ptr %start, ptr %end) {
; CHECK-LABEL: VPlan for loop in 'print_ext_mulacc_extended_const'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in vp<[[VP3:%[0-9]+]]> = backedge-taken count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:    EMIT vp<[[VP4:%[0-9]+]]> = EXPAND SCEV (1 + (-1 * (ptrtoint ptr %start to i64)) + (ptrtoint ptr %end to i64))
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:    EMIT vp<[[VP5:%[0-9]+]]> = reduction-start-vector ir<0>, ir<0>, ir<1>
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP6:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%red> = phi (add) vp<[[VP5]]>, vp<[[VP11:%[0-9]+]]>
; CHECK-NEXT:      vp<[[VP7:%[0-9]+]]> = SCALAR-STEPS vp<[[VP6]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      EMIT vp<%next.gep> = ptradd ir<%start>, vp<[[VP7]]>
; CHECK-NEXT:      EMIT vp<[[VP8:%[0-9]+]]> = WIDEN-CANONICAL-INDUCTION nuw vp<[[VP6]]>
; CHECK-NEXT:      EMIT vp<[[VP9:%[0-9]+]]> = icmp ule vp<[[VP8]]>, vp<[[VP3]]>
; CHECK-NEXT:      vp<[[VP10:%[0-9]+]]> = vector-pointer vp<%next.gep>, ir<1>
; CHECK-NEXT:      WIDEN ir<%l> = load vp<[[VP10]]>, vp<[[VP9]]>
; CHECK-NEXT:      EXPRESSION vp<[[VP11]]> = ir<%red> + reduce.add (mul (ir<%l> zext to i64), (ir<63> zext to i64), vp<[[VP9]]>)
; CHECK-NEXT:      EMIT vp<%index.next> = add vp<[[VP6]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP13:%[0-9]+]]> = compute-reduction-result (add, in-loop) vp<[[VP11]]>
; CHECK-NEXT:  Successor(s): ir-bb<exit>
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:    IR   %red.next.lcssa = phi i64 [ %red.next, %loop ] (extra operand: vp<[[VP13]]> from middle.block)
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;
entry:
  br label %loop

loop:
  %ptr.iv = phi ptr [ %start, %entry ], [ %gep.iv.next, %loop ]
  %red = phi i64 [ 0, %entry ], [ %red.next, %loop ]
  %l = load i8, ptr %ptr.iv, align 1
  %l.ext = zext i8 %l to i32
  %mul = mul i32 %l.ext, 63
  %mul.ext = zext i32 %mul to i64
  %red.next = add i64 %red, %mul.ext
  %gep.iv.next = getelementptr i8, ptr %ptr.iv, i64 1
  %ec = icmp eq ptr %ptr.iv, %end
  br i1 %ec, label %exit, label %loop

exit:
  ret i64 %red.next
}

; Constants >= 128 cannot be treated as sign-extended, so the expression shouldn't extend 128
define i64 @print_ext_mulacc_not_extended_const(ptr %start, ptr %end) {
; CHECK-LABEL: VPlan for loop in 'print_ext_mulacc_not_extended_const'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in vp<[[VP3:%[0-9]+]]> = backedge-taken count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:    EMIT vp<[[VP4:%[0-9]+]]> = EXPAND SCEV (1 + (-1 * (ptrtoint ptr %start to i64)) + (ptrtoint ptr %end to i64))
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:    EMIT vp<[[VP5:%[0-9]+]]> = reduction-start-vector ir<0>, ir<0>, ir<1>
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP6:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%red> = phi (add) vp<[[VP5]]>, vp<[[VP12:%[0-9]+]]>
; CHECK-NEXT:      vp<[[VP7:%[0-9]+]]> = SCALAR-STEPS vp<[[VP6]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      EMIT vp<%next.gep> = ptradd ir<%start>, vp<[[VP7]]>
; CHECK-NEXT:      EMIT vp<[[VP8:%[0-9]+]]> = WIDEN-CANONICAL-INDUCTION nuw vp<[[VP6]]>
; CHECK-NEXT:      EMIT vp<[[VP9:%[0-9]+]]> = icmp ule vp<[[VP8]]>, vp<[[VP3]]>
; CHECK-NEXT:      vp<[[VP10:%[0-9]+]]> = vector-pointer vp<%next.gep>, ir<1>
; CHECK-NEXT:      WIDEN ir<%l> = load vp<[[VP10]]>, vp<[[VP9]]>
; CHECK-NEXT:      WIDEN-CAST ir<%l.ext> = sext ir<%l> to i32
; CHECK-NEXT:      EMIT vp<[[VP11:%[0-9]+]]> = shl ir<%l.ext>, ir<7>
; CHECK-NEXT:      EXPRESSION vp<[[VP12]]> = vp<[[VP9]]> + reduce.add (vp<[[VP11]]> sext to i64, vp<[[VP9]]>)
; CHECK-NEXT:      EMIT vp<%index.next> = add vp<[[VP6]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP14:%[0-9]+]]> = compute-reduction-result (add, in-loop) vp<[[VP12]]>
; CHECK-NEXT:  Successor(s): ir-bb<exit>
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:    IR   %red.next.lcssa = phi i64 [ %red.next, %loop ] (extra operand: vp<[[VP14]]> from middle.block)
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;
entry:
  br label %loop

loop:
  %ptr.iv = phi ptr [ %start, %entry ], [ %gep.iv.next, %loop ]
  %red = phi i64 [ 0, %entry ], [ %red.next, %loop ]
  %l = load i8, ptr %ptr.iv, align 1
  %l.ext = sext i8 %l to i32
  %mul = mul i32 %l.ext, 128
  %mul.ext = sext i32 %mul to i64
  %red.next = add i64 %red, %mul.ext
  %gep.iv.next = getelementptr i8, ptr %ptr.iv, i64 1
  %ec = icmp eq ptr %ptr.iv, %end
  br i1 %ec, label %exit, label %loop

exit:
  %red.next.lcssa = phi i64 [ %red.next, %loop ]
  ret i64 %red.next.lcssa
}

; This reduce.add(ext(mul(ext(A), ext(B)))) can't be turned into an
; ExtMulAccReduction VPExpressionRecipe since the mul has two users.
; It can however be turned into an ExtendedReduction since that one doesn't
; modify the mul's operands.
define i64 @print_ext_mul_two_uses(i64 %n, ptr %a, i16 %b, i32 %c) {
; CHECK-LABEL: VPlan for loop in 'print_ext_mul_two_uses'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = backedge-taken count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:    EMIT vp<[[VP3:%[0-9]+]]> = EXPAND SCEV (1 + %n)
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:    EMIT vp<[[VP4:%[0-9]+]]> = reduction-start-vector ir<0>, ir<0>, ir<1>
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP5:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%res2> = phi (add) vp<[[VP4]]>, vp<[[VP8:%[0-9]+]]>
; CHECK-NEXT:      EMIT vp<[[VP6:%[0-9]+]]> = WIDEN-CANONICAL-INDUCTION nuw vp<[[VP5]]>
; CHECK-NEXT:      EMIT vp<[[VP7:%[0-9]+]]> = icmp ule vp<[[VP6]]>, vp<[[VP2]]>
; CHECK-NEXT:      EXPRESSION vp<[[VP8]]> = ir<%res2> + reduce.add (mul (ir<%b> sext to i64), (ir<%b> sext to i64), vp<[[VP7]]>)
; CHECK-NEXT:      EMIT vp<%index.next> = add vp<[[VP5]]>, vp<[[VP0]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP1]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP10:%[0-9]+]]> = compute-reduction-result (add, in-loop) vp<[[VP8]]>
; CHECK-NEXT:  Successor(s): ir-bb<exit>
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:    IR   %add.lcssa = phi i64 [ %add, %loop ] (extra operand: vp<[[VP10]]> from middle.block)
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %res1 = phi i64 [ 0, %entry ], [ %load.ext.ext, %loop ]
  %res2 = phi i64 [ 0, %entry ], [ %add, %loop ]
  %load = load i16, ptr %a, align 2
  %iv.next = add i64 %iv, 1
  %conv = sext i16 %b to i32
  %mul = mul i32 %conv, %conv
  %mul.ext = zext i32 %mul to i64
  %add = add i64 %res2, %mul.ext
  %second_use = or i32 %mul, %c ; this value is otherwise unused, but that's sufficient for the test
  %load.ext = sext i16 %load to i32
  %load.ext.ext = sext i32 %load.ext to i64
  %exitcond740.not = icmp eq i64 %iv, %n
  br i1 %exitcond740.not, label %exit, label %loop

exit:
  ret i64 %add
}

define i32 @print_umax_reduction_out_of_loop(ptr %y) {
; CHECK-LABEL: VPlan for loop in 'print_umax_reduction_out_of_loop'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in ir<100> = original trip-count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP3:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%red> = phi (umax) ir<0>, ir<%red.next>
; CHECK-NEXT:      vp<[[VP4:%[0-9]+]]> = SCALAR-STEPS vp<[[VP3]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      CLONE ir<%gep> = getelementptr inbounds ir<%y>, vp<[[VP4]]>
; CHECK-NEXT:      vp<[[VP5:%[0-9]+]]> = vector-pointer inbounds ir<%gep>, ir<1>
; CHECK-NEXT:      WIDEN ir<%lv> = load vp<[[VP5]]>
; CHECK-NEXT:      WIDEN-INTRINSIC ir<%red.next> = call llvm.umax(ir<%lv>, ir<%red>)
; CHECK-NEXT:      EMIT vp<%index.next> = add nuw vp<[[VP3]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP7:%[0-9]+]]> = compute-reduction-result (umax) ir<%red.next>
; CHECK-NEXT:    EMIT vp<%cmp.n> = icmp eq ir<100>, vp<[[VP2]]>
; CHECK-NEXT:    EMIT branch-on-cond vp<%cmp.n>
; CHECK-NEXT:  Successor(s): ir-bb<exit>, scalar.ph
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:    IR   %red.next.lcssa = phi i32 [ %red.next, %loop ] (extra operand: vp<[[VP7]]> from middle.block)
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %red = phi i32 [ 0, %entry ], [ %red.next, %loop ]
  %gep = getelementptr inbounds i32, ptr %y, i64 %iv
  %lv = load i32, ptr %gep, align 4
  %red.next = call i32 @llvm.umax(i32 %lv, i32 %red)
  %iv.next = add i64 %iv, 1
  %ec = icmp eq i64 %iv.next, 100
  br i1 %ec, label %exit, label %loop

exit:
  ret i32 %red.next
}

define i32 @print_umax_reduction(ptr %y) {
; CHECK-LABEL: VPlan for loop in 'print_umax_reduction'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in ir<100> = original trip-count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP3:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%red> = phi (umax) ir<0>, ir<%red.next>
; CHECK-NEXT:      vp<[[VP4:%[0-9]+]]> = SCALAR-STEPS vp<[[VP3]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      CLONE ir<%gep> = getelementptr inbounds ir<%y>, vp<[[VP4]]>
; CHECK-NEXT:      vp<[[VP5:%[0-9]+]]> = vector-pointer inbounds ir<%gep>, ir<1>
; CHECK-NEXT:      WIDEN ir<%lv> = load vp<[[VP5]]>
; CHECK-NEXT:      WIDEN-INTRINSIC ir<%red.next> = call llvm.umax(ir<%red>, ir<%lv>)
; CHECK-NEXT:      EMIT vp<%index.next> = add nuw vp<[[VP3]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP7:%[0-9]+]]> = compute-reduction-result (umax) ir<%red.next>
; CHECK-NEXT:    EMIT vp<%cmp.n> = icmp eq ir<100>, vp<[[VP2]]>
; CHECK-NEXT:    EMIT branch-on-cond vp<%cmp.n>
; CHECK-NEXT:  Successor(s): ir-bb<exit>, scalar.ph
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:    IR   %red.next.lcssa = phi i32 [ %red.next, %loop ] (extra operand: vp<[[VP7]]> from middle.block)
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %red = phi i32 [ 0, %entry ], [ %red.next, %loop ]
  %gep = getelementptr inbounds i32, ptr %y, i64 %iv
  %lv = load i32, ptr %gep, align 4
  %red.next = call i32 @llvm.umax(i32 %red, i32 %lv)
  %iv.next = add i64 %iv, 1
  %ec = icmp eq i64 %iv.next, 100
  br i1 %ec, label %exit, label %loop

exit:
  ret i32 %red.next
}

define i32 @print_umin_reduction(ptr %y) {
; CHECK-LABEL: VPlan for loop in 'print_umin_reduction'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in ir<100> = original trip-count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP3:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%red> = phi (umin) ir<0>, ir<%red.next>
; CHECK-NEXT:      vp<[[VP4:%[0-9]+]]> = SCALAR-STEPS vp<[[VP3]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      CLONE ir<%gep> = getelementptr inbounds ir<%y>, vp<[[VP4]]>
; CHECK-NEXT:      vp<[[VP5:%[0-9]+]]> = vector-pointer inbounds ir<%gep>, ir<1>
; CHECK-NEXT:      WIDEN ir<%lv> = load vp<[[VP5]]>
; CHECK-NEXT:      WIDEN-INTRINSIC ir<%red.next> = call llvm.umin(ir<%lv>, ir<%red>)
; CHECK-NEXT:      EMIT vp<%index.next> = add nuw vp<[[VP3]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP7:%[0-9]+]]> = compute-reduction-result (umin) ir<%red.next>
; CHECK-NEXT:    EMIT vp<%cmp.n> = icmp eq ir<100>, vp<[[VP2]]>
; CHECK-NEXT:    EMIT branch-on-cond vp<%cmp.n>
; CHECK-NEXT:  Successor(s): ir-bb<exit>, scalar.ph
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:    IR   %red.next.lcssa = phi i32 [ %red.next, %loop ] (extra operand: vp<[[VP7]]> from middle.block)
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %red = phi i32 [ 0, %entry ], [ %red.next, %loop ]
  %gep = getelementptr inbounds i32, ptr %y, i64 %iv
  %lv = load i32, ptr %gep, align 4
  %red.next = call i32 @llvm.umin(i32 %lv, i32 %red)
  %iv.next = add i64 %iv, 1
  %ec = icmp eq i64 %iv.next, 100
  br i1 %ec, label %exit, label %loop

exit:
  ret i32 %red.next
}

define i32 @print_smax_reduction(ptr %y) {
; CHECK-LABEL: VPlan for loop in 'print_smax_reduction'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in ir<100> = original trip-count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP3:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%red> = phi (smax) ir<0>, ir<%red.next>
; CHECK-NEXT:      vp<[[VP4:%[0-9]+]]> = SCALAR-STEPS vp<[[VP3]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      CLONE ir<%gep> = getelementptr inbounds ir<%y>, vp<[[VP4]]>
; CHECK-NEXT:      vp<[[VP5:%[0-9]+]]> = vector-pointer inbounds ir<%gep>, ir<1>
; CHECK-NEXT:      WIDEN ir<%lv> = load vp<[[VP5]]>
; CHECK-NEXT:      WIDEN-INTRINSIC ir<%red.next> = call llvm.smax(ir<%red>, ir<%lv>)
; CHECK-NEXT:      EMIT vp<%index.next> = add nuw vp<[[VP3]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP7:%[0-9]+]]> = compute-reduction-result (smax) ir<%red.next>
; CHECK-NEXT:    EMIT vp<%cmp.n> = icmp eq ir<100>, vp<[[VP2]]>
; CHECK-NEXT:    EMIT branch-on-cond vp<%cmp.n>
; CHECK-NEXT:  Successor(s): ir-bb<exit>, scalar.ph
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:    IR   %red.next.lcssa = phi i32 [ %red.next, %loop ] (extra operand: vp<[[VP7]]> from middle.block)
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %red = phi i32 [ 0, %entry ], [ %red.next, %loop ]
  %gep = getelementptr inbounds i32, ptr %y, i64 %iv
  %lv = load i32, ptr %gep, align 4
  %red.next = call i32 @llvm.smax(i32 %red, i32 %lv)
  %iv.next = add i64 %iv, 1
  %ec = icmp eq i64 %iv.next, 100
  br i1 %ec, label %exit, label %loop

exit:
  ret i32 %red.next
}

define i32 @print_smin_reduction(ptr %y) {
; CHECK-LABEL: VPlan for loop in 'print_smin_reduction'
; CHECK:  VPlan 'Initial VPlan for VF={4},UF>=1' {
; CHECK-NEXT:  Live-in vp<[[VP0:%[0-9]+]]> = VF
; CHECK-NEXT:  Live-in vp<[[VP1:%[0-9]+]]> = VF * UF
; CHECK-NEXT:  Live-in vp<[[VP2:%[0-9]+]]> = vector-trip-count
; CHECK-NEXT:  Live-in ir<100> = original trip-count
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<entry>:
; CHECK-NEXT:  Successor(s): scalar.ph, vector.ph
; CHECK-EMPTY:
; CHECK-NEXT:  vector.ph:
; CHECK-NEXT:  Successor(s): vector loop
; CHECK-EMPTY:
; CHECK-NEXT:  <x1> vector loop: {
; CHECK-NEXT:  vp<[[VP3:%[0-9]+]]> = CANONICAL-IV
; CHECK-EMPTY:
; CHECK-NEXT:    vector.body:
; CHECK-NEXT:      WIDEN-REDUCTION-PHI ir<%red> = phi (smin) ir<0>, ir<%red.next>
; CHECK-NEXT:      vp<[[VP4:%[0-9]+]]> = SCALAR-STEPS vp<[[VP3]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      CLONE ir<%gep> = getelementptr inbounds ir<%y>, vp<[[VP4]]>
; CHECK-NEXT:      vp<[[VP5:%[0-9]+]]> = vector-pointer inbounds ir<%gep>, ir<1>
; CHECK-NEXT:      WIDEN ir<%lv> = load vp<[[VP5]]>
; CHECK-NEXT:      WIDEN-INTRINSIC ir<%red.next> = call llvm.smin(ir<%lv>, ir<%red>)
; CHECK-NEXT:      EMIT vp<%index.next> = add nuw vp<[[VP3]]>, vp<[[VP1]]>
; CHECK-NEXT:      EMIT branch-on-count vp<%index.next>, vp<[[VP2]]>
; CHECK-NEXT:    No successors
; CHECK-NEXT:  }
; CHECK-NEXT:  Successor(s): middle.block
; CHECK-EMPTY:
; CHECK-NEXT:  middle.block:
; CHECK-NEXT:    EMIT vp<[[VP7:%[0-9]+]]> = compute-reduction-result (smin) ir<%red.next>
; CHECK-NEXT:    EMIT vp<%cmp.n> = icmp eq ir<100>, vp<[[VP2]]>
; CHECK-NEXT:    EMIT branch-on-cond vp<%cmp.n>
; CHECK-NEXT:  Successor(s): ir-bb<exit>, scalar.ph
; CHECK-EMPTY:
; CHECK-NEXT:  ir-bb<exit>:
; CHECK-NEXT:    IR   %red.next.lcssa = phi i32 [ %red.next, %loop ] (extra operand: vp<[[VP7]]> from middle.block)
; CHECK-NEXT:  No successors
; CHECK-EMPTY:
; CHECK-NEXT:  scalar.ph:
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %red = phi i32 [ 0, %entry ], [ %red.next, %loop ]
  %gep = getelementptr inbounds i32, ptr %y, i64 %iv
  %lv = load i32, ptr %gep, align 4
  %red.next = call i32 @llvm.smin(i32 %lv, i32 %red)
  %iv.next = add i64 %iv, 1
  %ec = icmp eq i64 %iv.next, 100
  br i1 %ec, label %exit, label %loop

exit:
  ret i32 %red.next
}