test/Transforms/LoopVectorize/VPlan/vplan-narrow-iv-users.ll

; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --filter-out-after "middle.block:" --version 6
; RUN: opt -passes=loop-vectorize -force-vector-interleave=1 -force-vector-width=4 -disable-output -vplan-print-after=legalizeAndOptimizeInductions %s 2>&1 | FileCheck %s

define void @narrow_iv_user_chain(ptr noalias %A, ptr noalias %B) {
; CHECK-LABEL: VPlan for loop in 'narrow_iv_user_chain'
; 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<1000> = 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:      ir<%iv> = WIDEN-INDUCTION nsw ir<0>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      vp<[[VP4:%[0-9]+]]> = DERIVED-IV ir<0> + vp<[[VP3]]> * ir<1>
; CHECK-NEXT:      vp<[[VP5:%[0-9]+]]> = SCALAR-STEPS vp<[[VP4]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      CLONE ir<%and> = and vp<[[VP5]]>, ir<-4>
; CHECK-NEXT:      CLONE ir<%gep.ld> = getelementptr inbounds ir<%A>, ir<%and>
; CHECK-NEXT:      CLONE ir<%ld> = load ir<%gep.ld>
; CHECK-NEXT:      CLONE ir<%calc> = add nsw ir<%ld>, ir<42>
; CHECK-NEXT:      CLONE ir<%gep.st> = getelementptr inbounds ir<%B>, ir<%and>
; CHECK-NEXT:      REPLICATE store ir<%calc>, ir<%gep.st>
; 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:
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %and = and i64 %iv, -4
  %gep.ld = getelementptr inbounds i64, ptr %A, i64 %and
  %ld = load i64, ptr %gep.ld, align 8
  %calc = add nsw i64 %ld, 42
  %gep.st = getelementptr inbounds i64, ptr %B, i64 %and
  store i64 %calc, ptr %gep.st, align 8
  %iv.next = add nsw i64 %iv, 1
  %ec = icmp eq i64 %iv.next, 1000
  br i1 %ec, label %exit, label %loop

exit:
  ret void
}

; Same as above, but with a chain of 2 narrowable users.
define void @narrow_iv_user_chain_multiple_levels(ptr noalias %A, ptr noalias %B) {
; CHECK-LABEL: VPlan for loop in 'narrow_iv_user_chain_multiple_levels'
; 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<1000> = 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:      ir<%iv> = WIDEN-INDUCTION nsw ir<0>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      vp<[[VP4:%[0-9]+]]> = DERIVED-IV ir<0> + vp<[[VP3]]> * ir<1>
; CHECK-NEXT:      vp<[[VP5:%[0-9]+]]> = SCALAR-STEPS vp<[[VP4]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      CLONE ir<%and> = and vp<[[VP5]]>, ir<-4>
; CHECK-NEXT:      CLONE ir<%gep.ld> = getelementptr inbounds ir<%A>, ir<%and>
; CHECK-NEXT:      CLONE ir<%ld> = load ir<%gep.ld>
; CHECK-NEXT:      CLONE ir<%calc> = add nsw ir<%ld>, ir<42>
; CHECK-NEXT:      CLONE ir<%calc2> = mul nsw ir<%calc>, ir<3>
; CHECK-NEXT:      CLONE ir<%gep.st> = getelementptr inbounds ir<%B>, ir<%and>
; CHECK-NEXT:      REPLICATE store ir<%calc2>, ir<%gep.st>
; 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:
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %and = and i64 %iv, -4
  %gep.ld = getelementptr inbounds i64, ptr %A, i64 %and
  %ld = load i64, ptr %gep.ld, align 8
  %calc = add nsw i64 %ld, 42
  %calc2 = mul nsw i64 %calc, 3
  %gep.st = getelementptr inbounds i64, ptr %B, i64 %and
  store i64 %calc2, ptr %gep.st, align 8
  %iv.next = add nsw i64 %iv, 1
  %ec = icmp eq i64 %iv.next, 1000
  br i1 %ec, label %exit, label %loop

exit:
  ret void
}

define void @no_narrowing_if_all_lanes_used(ptr noalias %A, ptr noalias %B) {
; CHECK-LABEL: VPlan for loop in 'no_narrowing_if_all_lanes_used'
; 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<1000> = 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:      ir<%iv> = WIDEN-INDUCTION nsw ir<0>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      vp<[[VP4:%[0-9]+]]> = DERIVED-IV ir<0> + vp<[[VP3]]> * ir<1>
; CHECK-NEXT:      vp<[[VP5:%[0-9]+]]> = SCALAR-STEPS vp<[[VP4]]>, ir<1>, vp<[[VP0]]>
; CHECK-NEXT:      CLONE ir<%gep.ld> = getelementptr inbounds ir<%A>, vp<[[VP5]]>
; CHECK-NEXT:      vp<[[VP6:%[0-9]+]]> = vector-pointer inbounds i64, ir<%gep.ld>, ir<1>
; CHECK-NEXT:      WIDEN ir<%ld> = load vp<[[VP6]]>
; CHECK-NEXT:      WIDEN ir<%calc> = add nsw ir<%ld>, ir<42>
; CHECK-NEXT:      CLONE ir<%gep.st> = getelementptr inbounds ir<%B>, vp<[[VP5]]>
; CHECK-NEXT:      vp<[[VP7:%[0-9]+]]> = vector-pointer inbounds i64, ir<%gep.st>, ir<1>
; CHECK-NEXT:      WIDEN store vp<[[VP7]]>, ir<%calc>
; 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:
;
entry:
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %gep.ld = getelementptr inbounds i64, ptr %A, i64 %iv
  %ld = load i64, ptr %gep.ld, align 8
  %calc = add nsw i64 %ld, 42
  %gep.st = getelementptr inbounds i64, ptr %B, i64 %iv
  store i64 %calc, ptr %gep.st, align 8
  %iv.next = add nsw i64 %iv, 1
  %ec = icmp eq i64 %iv.next, 1000
  br i1 %ec, label %exit, label %loop

exit:
  ret void
}