llvm/test/Transforms/LoopVectorize/X86/masked-store-cost.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 4
; RUN: opt -p loop-vectorize -mtriple=x86_64-apple-macosx -S %s | FileCheck %s

target datalayout = "e-m:o-p270:32:32-p271:32:32-p272:64:64-i64:64-i128:128-f80:128-n8:16:32:64-S128"

define i32 @test_scalar_predicated_cost(i64 %x, i64 %y, ptr %A) #0 {
; CHECK-LABEL: define i32 @test_scalar_predicated_cost(
; CHECK-SAME: i64 [[X:%.*]], i64 [[Y:%.*]], ptr [[A:%.*]]) #[[ATTR0:[0-9]+]] {
; CHECK-NEXT:  iter.check:
; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK:       vector.main.loop.iter.check:
; CHECK-NEXT:    br i1 false, label [[VEC_EPILOG_PH:%.*]], label [[VECTOR_PH1:%.*]]
; CHECK:       vector.ph:
; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <8 x i64> poison, i64 [[Y]], i64 0
; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <8 x i64> [[BROADCAST_SPLATINSERT]], <8 x i64> poison, <8 x i32> zeroinitializer
; CHECK-NEXT:    [[BROADCAST_SPLATINSERT4:%.*]] = insertelement <8 x i64> poison, i64 [[X]], i64 0
; CHECK-NEXT:    [[BROADCAST_SPLAT5:%.*]] = shufflevector <8 x i64> [[BROADCAST_SPLATINSERT4]], <8 x i64> poison, <8 x i32> zeroinitializer
; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
; CHECK:       vector.body:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH1]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <8 x i64> [ <i64 0, i64 1, i64 2, i64 3, i64 4, i64 5, i64 6, i64 7>, [[VECTOR_PH1]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT:    [[STEP_ADD:%.*]] = add <8 x i64> [[VEC_IND]], splat (i64 8)
; CHECK-NEXT:    [[STEP_ADD1:%.*]] = add <8 x i64> [[STEP_ADD]], splat (i64 8)
; CHECK-NEXT:    [[STEP_ADD2:%.*]] = add <8 x i64> [[STEP_ADD1]], splat (i64 8)
; CHECK-NEXT:    [[TMP4:%.*]] = icmp ule <8 x i64> [[VEC_IND]], [[BROADCAST_SPLAT]]
; CHECK-NEXT:    [[TMP5:%.*]] = icmp ule <8 x i64> [[STEP_ADD]], [[BROADCAST_SPLAT]]
; CHECK-NEXT:    [[TMP6:%.*]] = icmp ule <8 x i64> [[STEP_ADD1]], [[BROADCAST_SPLAT]]
; CHECK-NEXT:    [[TMP7:%.*]] = icmp ule <8 x i64> [[STEP_ADD2]], [[BROADCAST_SPLAT]]
; CHECK-NEXT:    [[TMP8:%.*]] = xor <8 x i1> [[TMP4]], splat (i1 true)
; CHECK-NEXT:    [[TMP9:%.*]] = xor <8 x i1> [[TMP5]], splat (i1 true)
; CHECK-NEXT:    [[TMP10:%.*]] = xor <8 x i1> [[TMP6]], splat (i1 true)
; CHECK-NEXT:    [[TMP11:%.*]] = xor <8 x i1> [[TMP7]], splat (i1 true)
; CHECK-NEXT:    [[TMP12:%.*]] = or <8 x i64> [[BROADCAST_SPLAT5]], [[VEC_IND]]
; CHECK-NEXT:    [[TMP13:%.*]] = or <8 x i64> [[BROADCAST_SPLAT5]], [[STEP_ADD]]
; CHECK-NEXT:    [[TMP14:%.*]] = or <8 x i64> [[BROADCAST_SPLAT5]], [[STEP_ADD1]]
; CHECK-NEXT:    [[TMP15:%.*]] = or <8 x i64> [[BROADCAST_SPLAT5]], [[STEP_ADD2]]
; CHECK-NEXT:    [[TMP16:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX]]
; CHECK-NEXT:    [[TMP20:%.*]] = trunc <8 x i64> [[TMP12]] to <8 x i32>
; CHECK-NEXT:    [[TMP21:%.*]] = trunc <8 x i64> [[TMP13]] to <8 x i32>
; CHECK-NEXT:    [[TMP22:%.*]] = trunc <8 x i64> [[TMP14]] to <8 x i32>
; CHECK-NEXT:    [[TMP23:%.*]] = trunc <8 x i64> [[TMP15]] to <8 x i32>
; CHECK-NEXT:    [[TMP24:%.*]] = getelementptr i32, ptr [[TMP16]], i32 0
; CHECK-NEXT:    [[TMP25:%.*]] = getelementptr i32, ptr [[TMP16]], i32 8
; CHECK-NEXT:    [[TMP26:%.*]] = getelementptr i32, ptr [[TMP16]], i32 16
; CHECK-NEXT:    [[TMP27:%.*]] = getelementptr i32, ptr [[TMP16]], i32 24
; CHECK-NEXT:    call void @llvm.masked.store.v8i32.p0(<8 x i32> [[TMP20]], ptr [[TMP24]], i32 4, <8 x i1> [[TMP8]])
; CHECK-NEXT:    call void @llvm.masked.store.v8i32.p0(<8 x i32> [[TMP21]], ptr [[TMP25]], i32 4, <8 x i1> [[TMP9]])
; CHECK-NEXT:    call void @llvm.masked.store.v8i32.p0(<8 x i32> [[TMP22]], ptr [[TMP26]], i32 4, <8 x i1> [[TMP10]])
; CHECK-NEXT:    call void @llvm.masked.store.v8i32.p0(<8 x i32> [[TMP23]], ptr [[TMP27]], i32 4, <8 x i1> [[TMP11]])
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 32
; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <8 x i64> [[STEP_ADD2]], splat (i64 8)
; CHECK-NEXT:    [[TMP28:%.*]] = icmp eq i64 [[INDEX_NEXT]], 96
; CHECK-NEXT:    br i1 [[TMP28]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; CHECK:       middle.block:
; CHECK-NEXT:    br i1 false, label [[EXIT:%.*]], label [[VEC_EPILOG_ITER_CHECK:%.*]]
; CHECK:       vec.epilog.iter.check:
; CHECK-NEXT:    br i1 false, label [[SCALAR_PH]], label [[VEC_EPILOG_PH]]
; CHECK:       vec.epilog.ph:
; CHECK-NEXT:    [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ 96, [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_PH]] ]
; CHECK-NEXT:    [[BROADCAST_SPLATINSERT7:%.*]] = insertelement <4 x i64> poison, i64 [[Y]], i64 0
; CHECK-NEXT:    [[BROADCAST_SPLAT8:%.*]] = shufflevector <4 x i64> [[BROADCAST_SPLATINSERT7]], <4 x i64> poison, <4 x i32> zeroinitializer
; CHECK-NEXT:    [[BROADCAST_SPLATINSERT9:%.*]] = insertelement <4 x i64> poison, i64 [[X]], i64 0
; CHECK-NEXT:    [[BROADCAST_SPLAT10:%.*]] = shufflevector <4 x i64> [[BROADCAST_SPLATINSERT9]], <4 x i64> poison, <4 x i32> zeroinitializer
; CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <4 x i64> poison, i64 [[VEC_EPILOG_RESUME_VAL]], i64 0
; CHECK-NEXT:    [[DOTSPLAT:%.*]] = shufflevector <4 x i64> [[DOTSPLATINSERT]], <4 x i64> poison, <4 x i32> zeroinitializer
; CHECK-NEXT:    [[INDUCTION:%.*]] = add <4 x i64> [[DOTSPLAT]], <i64 0, i64 1, i64 2, i64 3>
; CHECK-NEXT:    br label [[LOOP_HEADER:%.*]]
; CHECK:       vec.epilog.vector.body:
; CHECK-NEXT:    [[INDEX4:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT11:%.*]], [[LOOP_HEADER]] ]
; CHECK-NEXT:    [[VEC_IND5:%.*]] = phi <4 x i64> [ [[INDUCTION]], [[VEC_EPILOG_PH]] ], [ [[VEC_IND_NEXT6:%.*]], [[LOOP_HEADER]] ]
; CHECK-NEXT:    [[TMP32:%.*]] = icmp ule <4 x i64> [[VEC_IND5]], [[BROADCAST_SPLAT8]]
; CHECK-NEXT:    [[TMP33:%.*]] = xor <4 x i1> [[TMP32]], splat (i1 true)
; CHECK-NEXT:    [[TMP34:%.*]] = or <4 x i64> [[BROADCAST_SPLAT10]], [[VEC_IND5]]
; CHECK-NEXT:    [[TMP35:%.*]] = getelementptr i32, ptr [[A]], i64 [[INDEX4]]
; CHECK-NEXT:    [[TMP36:%.*]] = trunc <4 x i64> [[TMP34]] to <4 x i32>
; CHECK-NEXT:    [[TMP29:%.*]] = getelementptr i32, ptr [[TMP35]], i32 0
; CHECK-NEXT:    call void @llvm.masked.store.v4i32.p0(<4 x i32> [[TMP36]], ptr [[TMP29]], i32 4, <4 x i1> [[TMP33]])
; CHECK-NEXT:    [[INDEX_NEXT11]] = add nuw i64 [[INDEX4]], 4
; CHECK-NEXT:    [[VEC_IND_NEXT6]] = add <4 x i64> [[VEC_IND5]], splat (i64 4)
; CHECK-NEXT:    [[TMP30:%.*]] = icmp eq i64 [[INDEX_NEXT11]], 100
; CHECK-NEXT:    br i1 [[TMP30]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[LOOP_HEADER]], !llvm.loop [[LOOP3:![0-9]+]]
; CHECK:       vec.epilog.middle.block:
; CHECK-NEXT:    br i1 false, label [[EXIT]], label [[SCALAR_PH]]
; CHECK:       vec.epilog.scalar.ph:
; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 100, [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ 96, [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[ITER_CHECK:%.*]] ]
; CHECK-NEXT:    br label [[LOOP_HEADER1:%.*]]
; CHECK:       loop.header:
; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ]
; CHECK-NEXT:    [[CMP9_NOT:%.*]] = icmp ule i64 [[IV]], [[Y]]
; CHECK-NEXT:    br i1 [[CMP9_NOT]], label [[LOOP_LATCH]], label [[IF_THEN:%.*]]
; CHECK:       if.then:
; CHECK-NEXT:    [[OR:%.*]] = or i64 [[X]], [[IV]]
; CHECK-NEXT:    [[GEP:%.*]] = getelementptr i32, ptr [[A]], i64 [[IV]]
; CHECK-NEXT:    [[T:%.*]] = trunc i64 [[OR]] to i32
; CHECK-NEXT:    store i32 [[T]], ptr [[GEP]], align 4
; CHECK-NEXT:    br label [[LOOP_LATCH]]
; CHECK:       loop.latch:
; CHECK-NEXT:    [[IV_NEXT]] = add i64 [[IV]], 1
; CHECK-NEXT:    [[EC:%.*]] = icmp eq i64 [[IV]], 100
; CHECK-NEXT:    br i1 [[EC]], label [[EXIT]], label [[LOOP_HEADER1]], !llvm.loop [[LOOP4:![0-9]+]]
; CHECK:       exit:
; CHECK-NEXT:    ret i32 0
;
entry:
  br label %loop.header

loop.header:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
  %cmp9.not = icmp ule i64 %iv, %y
  br i1 %cmp9.not, label %loop.latch, label %if.then

if.then:
  %or = or i64 %x, %iv
  %gep = getelementptr i32, ptr %A, i64 %iv
  %t = trunc i64 %or to i32
  store i32 %t, ptr %gep, align 4
  br label %loop.latch

loop.latch:
  %iv.next = add i64 %iv, 1
  %ec = icmp eq i64 %iv, 100
  br i1 %ec, label %exit, label %loop.header

exit:
  ret i32 0
}

define void @test_scalar_cost_single_store_loop_invariant_cond(ptr %dst, i1 %c) #0 {
; CHECK-LABEL: define void @test_scalar_cost_single_store_loop_invariant_cond(
; CHECK-SAME: ptr [[DST:%.*]], i1 [[C:%.*]]) #[[ATTR0]] {
; CHECK-NEXT:  entry:
; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK:       vector.ph:
; CHECK-NEXT:    [[IND_END:%.*]] = getelementptr i8, ptr [[DST]], i64 96
; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <8 x i1> poison, i1 [[C]], i64 0
; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <8 x i1> [[BROADCAST_SPLATINSERT]], <8 x i1> poison, <8 x i32> zeroinitializer
; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
; CHECK:       vector.body:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 4
; CHECK-NEXT:    [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[DST]], i64 [[OFFSET_IDX]]
; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr i32, ptr [[NEXT_GEP]], i32 0
; CHECK-NEXT:    call void @llvm.masked.store.v8i32.p0(<8 x i32> zeroinitializer, ptr [[TMP1]], i32 4, <8 x i1> [[BROADCAST_SPLAT]])
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
; CHECK-NEXT:    [[TMP2:%.*]] = icmp eq i64 [[INDEX_NEXT]], 24
; CHECK-NEXT:    br i1 [[TMP2]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
; CHECK:       middle.block:
; CHECK-NEXT:    br i1 false, label [[EXIT:%.*]], label [[SCALAR_PH]]
; CHECK:       scalar.ph:
; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi ptr [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ [[DST]], [[ENTRY:%.*]] ]
; CHECK-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 96, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
; CHECK-NEXT:    br label [[LOOP_HEADER:%.*]]
; CHECK:       loop.header:
; CHECK-NEXT:    [[PTR_IV:%.*]] = phi ptr [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[PTR_IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ]
; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP_LATCH]] ]
; CHECK-NEXT:    br i1 [[C]], label [[IF_THEN:%.*]], label [[LOOP_LATCH]]
; CHECK:       if.then:
; CHECK-NEXT:    store i32 0, ptr [[PTR_IV]], align 4
; CHECK-NEXT:    br label [[LOOP_LATCH]]
; CHECK:       loop.latch:
; CHECK-NEXT:    [[IV_NEXT]] = add i64 [[IV]], 4
; CHECK-NEXT:    [[PTR_IV_NEXT]] = getelementptr i8, ptr [[DST]], i64 [[IV_NEXT]]
; CHECK-NEXT:    [[CMP_NOT:%.*]] = icmp eq i64 [[IV]], 116
; CHECK-NEXT:    br i1 [[CMP_NOT]], label [[EXIT]], label [[LOOP_HEADER]], !llvm.loop [[LOOP6:![0-9]+]]
; CHECK:       exit:
; CHECK-NEXT:    ret void
;
entry:
  br label %loop.header

loop.header:
  %ptr.iv = phi ptr [ %dst, %entry ], [ %ptr.iv.next, %loop.latch ]
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
  br i1 %c, label %if.then, label %loop.latch

if.then:
  store i32 0, ptr %ptr.iv, align 4
  br label %loop.latch

loop.latch:
  %iv.next = add i64 %iv, 4
  %ptr.iv.next = getelementptr i8, ptr %dst, i64 %iv.next
  %cmp.not = icmp eq i64 %iv, 116
  br i1 %cmp.not, label %exit, label %loop.header

exit:
  ret void
}

define void @test_scalar_cost_single_store_loop_varying_cond(ptr %dst, ptr noalias %src) #0 {
; CHECK-LABEL: define void @test_scalar_cost_single_store_loop_varying_cond(
; CHECK-SAME: ptr [[DST:%.*]], ptr noalias [[SRC:%.*]]) #[[ATTR0]] {
; CHECK-NEXT:  entry:
; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK:       vector.ph:
; CHECK-NEXT:    [[IND_END:%.*]] = getelementptr i8, ptr [[DST]], i64 96
; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
; CHECK:       vector.body:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 4
; CHECK-NEXT:    [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[DST]], i64 [[OFFSET_IDX]]
; CHECK-NEXT:    [[OFFSET_IDX3:%.*]] = mul i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP3:%.*]] = add i64 [[OFFSET_IDX3]], 16
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i32, ptr [[SRC]], i64 [[OFFSET_IDX3]]
; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[SRC]], i64 [[TMP3]]
; CHECK-NEXT:    [[WIDE_VEC:%.*]] = load <16 x i32>, ptr [[TMP4]], align 4
; CHECK-NEXT:    [[STRIDED_VEC:%.*]] = shufflevector <16 x i32> [[WIDE_VEC]], <16 x i32> poison, <4 x i32> <i32 0, i32 4, i32 8, i32 12>
; CHECK-NEXT:    [[WIDE_VEC4:%.*]] = load <16 x i32>, ptr [[TMP5]], align 4
; CHECK-NEXT:    [[STRIDED_VEC5:%.*]] = shufflevector <16 x i32> [[WIDE_VEC4]], <16 x i32> poison, <4 x i32> <i32 0, i32 4, i32 8, i32 12>
; CHECK-NEXT:    [[TMP8:%.*]] = icmp eq <4 x i32> [[STRIDED_VEC]], splat (i32 123)
; CHECK-NEXT:    [[TMP9:%.*]] = icmp eq <4 x i32> [[STRIDED_VEC5]], splat (i32 123)
; CHECK-NEXT:    [[TMP10:%.*]] = getelementptr i32, ptr [[NEXT_GEP]], i32 0
; CHECK-NEXT:    [[TMP11:%.*]] = getelementptr i32, ptr [[NEXT_GEP]], i32 4
; CHECK-NEXT:    call void @llvm.masked.store.v4i32.p0(<4 x i32> zeroinitializer, ptr [[TMP10]], i32 4, <4 x i1> [[TMP8]])
; CHECK-NEXT:    call void @llvm.masked.store.v4i32.p0(<4 x i32> zeroinitializer, ptr [[TMP11]], i32 4, <4 x i1> [[TMP9]])
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
; CHECK-NEXT:    [[TMP12:%.*]] = icmp eq i64 [[INDEX_NEXT]], 24
; CHECK-NEXT:    br i1 [[TMP12]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
; CHECK:       middle.block:
; CHECK-NEXT:    br label [[SCALAR_PH]]
; CHECK:       scalar.ph:
; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi ptr [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ [[DST]], [[ENTRY:%.*]] ]
; CHECK-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 96, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
; CHECK-NEXT:    br label [[LOOP_HEADER:%.*]]
; CHECK:       loop.header:
; CHECK-NEXT:    [[PTR_IV:%.*]] = phi ptr [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[PTR_IV_NEXT:%.*]], [[LOOP_LATCH:%.*]] ]
; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP_LATCH]] ]
; CHECK-NEXT:    [[GEP_SRC:%.*]] = getelementptr inbounds i32, ptr [[SRC]], i64 [[IV]]
; CHECK-NEXT:    [[L:%.*]] = load i32, ptr [[GEP_SRC]], align 4
; CHECK-NEXT:    [[C:%.*]] = icmp eq i32 [[L]], 123
; CHECK-NEXT:    br i1 [[C]], label [[IF_THEN:%.*]], label [[LOOP_LATCH]]
; CHECK:       if.then:
; CHECK-NEXT:    store i32 0, ptr [[PTR_IV]], align 4
; CHECK-NEXT:    br label [[LOOP_LATCH]]
; CHECK:       loop.latch:
; CHECK-NEXT:    [[IV_NEXT]] = add i64 [[IV]], 4
; CHECK-NEXT:    [[PTR_IV_NEXT]] = getelementptr i8, ptr [[DST]], i64 [[IV_NEXT]]
; CHECK-NEXT:    [[CMP_NOT:%.*]] = icmp eq i64 [[IV]], 116
; CHECK-NEXT:    br i1 [[CMP_NOT]], label [[EXIT:%.*]], label [[LOOP_HEADER]], !llvm.loop [[LOOP8:![0-9]+]]
; CHECK:       exit:
; CHECK-NEXT:    ret void
;
entry:
  br label %loop.header

loop.header:
  %ptr.iv = phi ptr [ %dst, %entry ], [ %ptr.iv.next, %loop.latch ]
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop.latch ]
  %gep.src = getelementptr inbounds i32, ptr %src, i64 %iv
  %l = load i32, ptr %gep.src
  %c = icmp eq i32 %l, 123
  br i1 %c, label %if.then, label %loop.latch

if.then:
  store i32 0, ptr %ptr.iv, align 4
  br label %loop.latch

loop.latch:
  %iv.next = add i64 %iv, 4
  %ptr.iv.next = getelementptr i8, ptr %dst, i64 %iv.next
  %cmp.not = icmp eq i64 %iv, 116
  br i1 %cmp.not, label %exit, label %loop.header

exit:
  ret void
}

attributes #0 = { "min-legal-vector-width"="0" "target-cpu"="skylake-avx512" }
;.
; CHECK: [[LOOP0]] = distinct !{[[LOOP0]], [[META1:![0-9]+]], [[META2:![0-9]+]]}
; CHECK: [[META1]] = !{!"llvm.loop.isvectorized", i32 1}
; CHECK: [[META2]] = !{!"llvm.loop.unroll.runtime.disable"}
; CHECK: [[LOOP3]] = distinct !{[[LOOP3]], [[META1]], [[META2]]}
; CHECK: [[LOOP4]] = distinct !{[[LOOP4]], [[META2]], [[META1]]}
; CHECK: [[LOOP5]] = distinct !{[[LOOP5]], [[META1]], [[META2]]}
; CHECK: [[LOOP6]] = distinct !{[[LOOP6]], [[META2]], [[META1]]}
; CHECK: [[LOOP7]] = distinct !{[[LOOP7]], [[META1]], [[META2]]}
; CHECK: [[LOOP8]] = distinct !{[[LOOP8]], [[META2]], [[META1]]}
;.