[llvm] 6826047 - [LV] Remove redundant checks from recurrence test.

Wed Jul 6 15:32:16 PDT 2022

Author: Florian Hahn
Date: 2022-07-06T15:31:57-07:00
New Revision: 682604738122e06579c911b886511993a91541ea

URL: https://github.com/llvm/llvm-project/commit/682604738122e06579c911b886511993a91541ea
DIFF: https://github.com/llvm/llvm-project/commit/682604738122e06579c911b886511993a91541ea.diff

LOG: [LV] Remove redundant checks from recurrence test.

The removed CHECK configurations are tested as well below, modulo the
dce/instcombine runs. This makes them redundant, and removing them
removes a substantial amount of uneeded checks.

Added: 
    

Modified: 
    llvm/test/Transforms/LoopVectorize/first-order-recurrence.ll

Removed: 
    


################################################################################
diff  --git a/llvm/test/Transforms/LoopVectorize/first-order-recurrence.ll b/llvm/test/Transforms/LoopVectorize/first-order-recurrence.ll
index c308a37ab9085..7c8eb4c17c3cf 100644

--- a/llvm/test/Transforms/LoopVectorize/first-order-recurrence.ll
+++ b/llvm/test/Transforms/LoopVectorize/first-order-recurrence.ll
@@ -1,6 +1,4 @@
 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
-; RUN: opt < %s -loop-vectorize -force-vector-width=4 -force-vector-interleave=1 -dce -instcombine -S | FileCheck %s
-; RUN: opt < %s -loop-vectorize -force-vector-width=4 -force-vector-interleave=2 -dce -instcombine -S | FileCheck %s --check-prefix=UNROLL
 ; RUN: opt < %s -loop-vectorize -force-vector-width=4 -force-vector-interleave=2 -S | FileCheck %s --check-prefix=UNROLL-NO-IC
 ; RUN: opt < %s -loop-vectorize -force-vector-width=1 -force-vector-interleave=2 -S | FileCheck %s --check-prefix=UNROLL-NO-VF
 ; RUN: opt < %s -loop-vectorize -force-vector-width=4 -force-vector-interleave=1 -S | FileCheck %s --check-prefix=SINK-AFTER
@@ -15,118 +13,6 @@ target datalayout = "e-m:e-i64:64-i128:128-n32:64-S128"
 ;
 ;
 define void @recurrence_1(i32* readonly noalias %a, i32* noalias %b, i32 %n) {
-; CHECK-LABEL: @recurrence_1(
-; CHECK-NEXT:  entry:
-; CHECK-NEXT:    br label [[FOR_PREHEADER:%.*]]
-; CHECK:       for.preheader:
-; CHECK-NEXT:    [[PRE_LOAD:%.*]] = load i32, i32* [[A:%.*]], align 4
-; CHECK-NEXT:    [[TMP0:%.*]] = add i32 [[N:%.*]], -1
-; CHECK-NEXT:    [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
-; CHECK-NEXT:    [[TMP2:%.*]] = add nuw nsw i64 [[TMP1]], 1
-; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[TMP0]], 3
-; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; CHECK:       vector.ph:
-; CHECK-NEXT:    [[N_VEC:%.*]] = and i64 [[TMP2]], 8589934588
-; CHECK-NEXT:    [[VECTOR_RECUR_INIT:%.*]] = insertelement <4 x i32> poison, i32 [[PRE_LOAD]], i64 3
-; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
-; CHECK:       vector.body:
-; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VECTOR_RECUR:%.*]] = phi <4 x i32> [ [[VECTOR_RECUR_INIT]], [[VECTOR_PH]] ], [ [[WIDE_LOAD:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP3:%.*]] = or i64 [[INDEX]], 1
-; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP3]]
-; CHECK-NEXT:    [[TMP5:%.*]] = bitcast i32* [[TMP4]] to <4 x i32>*
-; CHECK-NEXT:    [[WIDE_LOAD]] = load <4 x i32>, <4 x i32>* [[TMP5]], align 4
-; CHECK-NEXT:    [[TMP6:%.*]] = shufflevector <4 x i32> [[VECTOR_RECUR]], <4 x i32> [[WIDE_LOAD]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i32, i32* [[B:%.*]], i64 [[INDEX]]
-; CHECK-NEXT:    [[TMP8:%.*]] = add <4 x i32> [[WIDE_LOAD]], [[TMP6]]
-; CHECK-NEXT:    [[TMP9:%.*]] = bitcast i32* [[TMP7]] to <4 x i32>*
-; CHECK-NEXT:    store <4 x i32> [[TMP8]], <4 x i32>* [[TMP9]], align 4
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT:    br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP2]], [[N_VEC]]
-; CHECK-NEXT:    [[VECTOR_RECUR_EXTRACT:%.*]] = extractelement <4 x i32> [[WIDE_LOAD]], i64 3
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[FOR_EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[SCALAR_RECUR_INIT:%.*]] = phi i32 [ [[PRE_LOAD]], [[FOR_PREHEADER]] ], [ [[VECTOR_RECUR_EXTRACT]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 0, [[FOR_PREHEADER]] ], [ [[N_VEC]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT:    br label [[SCALAR_BODY:%.*]]
-; CHECK:       scalar.body:
-; CHECK-NEXT:    [[SCALAR_RECUR:%.*]] = phi i32 [ [[SCALAR_RECUR_INIT]], [[SCALAR_PH]] ], [ [[TMP11:%.*]], [[SCALAR_BODY]] ]
-; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[SCALAR_BODY]] ]
-; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; CHECK-NEXT:    [[ARRAYIDX32:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[INDVARS_IV_NEXT]]
-; CHECK-NEXT:    [[TMP11]] = load i32, i32* [[ARRAYIDX32]], align 4
-; CHECK-NEXT:    [[ARRAYIDX34:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    [[ADD35:%.*]] = add i32 [[TMP11]], [[SCALAR_RECUR]]
-; CHECK-NEXT:    store i32 [[ADD35]], i32* [[ARRAYIDX34]], align 4
-; CHECK-NEXT:    [[LFTR_WIDEIV:%.*]] = trunc i64 [[INDVARS_IV_NEXT]] to i32
-; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp eq i32 [[LFTR_WIDEIV]], [[N]]
-; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_EXIT]], label [[SCALAR_BODY]], !llvm.loop [[LOOP2:![0-9]+]]
-; CHECK:       for.exit:
-; CHECK-NEXT:    ret void
-;
-; UNROLL-LABEL: @recurrence_1(
-; UNROLL-NEXT:  entry:
-; UNROLL-NEXT:    br label [[FOR_PREHEADER:%.*]]
-; UNROLL:       for.preheader:
-; UNROLL-NEXT:    [[PRE_LOAD:%.*]] = load i32, i32* [[A:%.*]], align 4
-; UNROLL-NEXT:    [[TMP0:%.*]] = add i32 [[N:%.*]], -1
-; UNROLL-NEXT:    [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
-; UNROLL-NEXT:    [[TMP2:%.*]] = add nuw nsw i64 [[TMP1]], 1
-; UNROLL-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[TMP0]], 7
-; UNROLL-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; UNROLL:       vector.ph:
-; UNROLL-NEXT:    [[N_VEC:%.*]] = and i64 [[TMP2]], 8589934584
-; UNROLL-NEXT:    [[VECTOR_RECUR_INIT:%.*]] = insertelement <4 x i32> poison, i32 [[PRE_LOAD]], i64 3
-; UNROLL-NEXT:    br label [[VECTOR_BODY:%.*]]
-; UNROLL:       vector.body:
-; UNROLL-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[VECTOR_RECUR:%.*]] = phi <4 x i32> [ [[VECTOR_RECUR_INIT]], [[VECTOR_PH]] ], [ [[WIDE_LOAD1:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[TMP3:%.*]] = or i64 [[INDEX]], 1
-; UNROLL-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[TMP3]]
-; UNROLL-NEXT:    [[TMP5:%.*]] = bitcast i32* [[TMP4]] to <4 x i32>*
-; UNROLL-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i32>, <4 x i32>* [[TMP5]], align 4
-; UNROLL-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i32, i32* [[TMP4]], i64 4
-; UNROLL-NEXT:    [[TMP7:%.*]] = bitcast i32* [[TMP6]] to <4 x i32>*
-; UNROLL-NEXT:    [[WIDE_LOAD1]] = load <4 x i32>, <4 x i32>* [[TMP7]], align 4
-; UNROLL-NEXT:    [[TMP8:%.*]] = shufflevector <4 x i32> [[VECTOR_RECUR]], <4 x i32> [[WIDE_LOAD]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; UNROLL-NEXT:    [[TMP9:%.*]] = shufflevector <4 x i32> [[WIDE_LOAD]], <4 x i32> [[WIDE_LOAD1]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; UNROLL-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i32, i32* [[B:%.*]], i64 [[INDEX]]
-; UNROLL-NEXT:    [[TMP11:%.*]] = add <4 x i32> [[WIDE_LOAD]], [[TMP8]]
-; UNROLL-NEXT:    [[TMP12:%.*]] = add <4 x i32> [[WIDE_LOAD1]], [[TMP9]]
-; UNROLL-NEXT:    [[TMP13:%.*]] = bitcast i32* [[TMP10]] to <4 x i32>*
-; UNROLL-NEXT:    store <4 x i32> [[TMP11]], <4 x i32>* [[TMP13]], align 4
-; UNROLL-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i32, i32* [[TMP10]], i64 4
-; UNROLL-NEXT:    [[TMP15:%.*]] = bitcast i32* [[TMP14]] to <4 x i32>*
-; UNROLL-NEXT:    store <4 x i32> [[TMP12]], <4 x i32>* [[TMP15]], align 4
-; UNROLL-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
-; UNROLL-NEXT:    [[TMP16:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; UNROLL-NEXT:    br i1 [[TMP16]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
-; UNROLL:       middle.block:
-; UNROLL-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP2]], [[N_VEC]]
-; UNROLL-NEXT:    [[VECTOR_RECUR_EXTRACT:%.*]] = extractelement <4 x i32> [[WIDE_LOAD1]], i64 3
-; UNROLL-NEXT:    br i1 [[CMP_N]], label [[FOR_EXIT:%.*]], label [[SCALAR_PH]]
-; UNROLL:       scalar.ph:
-; UNROLL-NEXT:    [[SCALAR_RECUR_INIT:%.*]] = phi i32 [ [[PRE_LOAD]], [[FOR_PREHEADER]] ], [ [[VECTOR_RECUR_EXTRACT]], [[MIDDLE_BLOCK]] ]
-; UNROLL-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 0, [[FOR_PREHEADER]] ], [ [[N_VEC]], [[MIDDLE_BLOCK]] ]
-; UNROLL-NEXT:    br label [[SCALAR_BODY:%.*]]
-; UNROLL:       scalar.body:
-; UNROLL-NEXT:    [[SCALAR_RECUR:%.*]] = phi i32 [ [[SCALAR_RECUR_INIT]], [[SCALAR_PH]] ], [ [[TMP17:%.*]], [[SCALAR_BODY]] ]
-; UNROLL-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[SCALAR_BODY]] ]
-; UNROLL-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; UNROLL-NEXT:    [[ARRAYIDX32:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[INDVARS_IV_NEXT]]
-; UNROLL-NEXT:    [[TMP17]] = load i32, i32* [[ARRAYIDX32]], align 4
-; UNROLL-NEXT:    [[ARRAYIDX34:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 [[INDVARS_IV]]
-; UNROLL-NEXT:    [[ADD35:%.*]] = add i32 [[TMP17]], [[SCALAR_RECUR]]
-; UNROLL-NEXT:    store i32 [[ADD35]], i32* [[ARRAYIDX34]], align 4
-; UNROLL-NEXT:    [[LFTR_WIDEIV:%.*]] = trunc i64 [[INDVARS_IV_NEXT]] to i32
-; UNROLL-NEXT:    [[EXITCOND:%.*]] = icmp eq i32 [[LFTR_WIDEIV]], [[N]]
-; UNROLL-NEXT:    br i1 [[EXITCOND]], label [[FOR_EXIT]], label [[SCALAR_BODY]], !llvm.loop [[LOOP2:![0-9]+]]
-; UNROLL:       for.exit:
-; UNROLL-NEXT:    ret void
-;
 ; UNROLL-NO-IC-LABEL: @recurrence_1(
 ; UNROLL-NO-IC-NEXT:  entry:
 ; UNROLL-NO-IC-NEXT:    br label [[FOR_PREHEADER:%.*]]
@@ -347,135 +233,6 @@ for.exit:
 ;
 ;
 define i32 @recurrence_2(i32* nocapture readonly %a, i32 %n) {
-; CHECK-LABEL: @recurrence_2(
-; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[CMP27:%.*]] = icmp sgt i32 [[N:%.*]], 0
-; CHECK-NEXT:    br i1 [[CMP27]], label [[FOR_PREHEADER:%.*]], label [[FOR_COND_CLEANUP:%.*]]
-; CHECK:       for.preheader:
-; CHECK-NEXT:    [[ARRAYIDX2_PHI_TRANS_INSERT:%.*]] = getelementptr inbounds i32, i32* [[A:%.*]], i64 -1
-; CHECK-NEXT:    [[DOTPRE:%.*]] = load i32, i32* [[ARRAYIDX2_PHI_TRANS_INSERT]], align 4
-; CHECK-NEXT:    [[TMP0:%.*]] = add i32 [[N]], -1
-; CHECK-NEXT:    [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
-; CHECK-NEXT:    [[TMP2:%.*]] = add nuw nsw i64 [[TMP1]], 1
-; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[TMP0]], 3
-; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; CHECK:       vector.ph:
-; CHECK-NEXT:    [[N_VEC:%.*]] = and i64 [[TMP2]], 8589934588
-; CHECK-NEXT:    [[VECTOR_RECUR_INIT:%.*]] = insertelement <4 x i32> poison, i32 [[DOTPRE]], i64 3
-; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
-; CHECK:       vector.body:
-; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VECTOR_RECUR:%.*]] = phi <4 x i32> [ [[VECTOR_RECUR_INIT]], [[VECTOR_PH]] ], [ [[WIDE_LOAD:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ poison, [[VECTOR_PH]] ], [ [[TMP8:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[INDEX]]
-; CHECK-NEXT:    [[TMP4:%.*]] = bitcast i32* [[TMP3]] to <4 x i32>*
-; CHECK-NEXT:    [[WIDE_LOAD]] = load <4 x i32>, <4 x i32>* [[TMP4]], align 4
-; CHECK-NEXT:    [[TMP5:%.*]] = shufflevector <4 x i32> [[VECTOR_RECUR]], <4 x i32> [[WIDE_LOAD]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; CHECK-NEXT:    [[TMP6:%.*]] = sub nsw <4 x i32> [[WIDE_LOAD]], [[TMP5]]
-; CHECK-NEXT:    [[TMP7:%.*]] = call <4 x i32> @llvm.smax.v4i32(<4 x i32> [[TMP6]], <4 x i32> zeroinitializer)
-; CHECK-NEXT:    [[TMP8]] = call <4 x i32> @llvm.smin.v4i32(<4 x i32> [[VEC_PHI]], <4 x i32> [[TMP7]])
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-NEXT:    [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT:    br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[TMP10:%.*]] = call i32 @llvm.vector.reduce.smin.v4i32(<4 x i32> [[TMP8]])
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP2]], [[N_VEC]]
-; CHECK-NEXT:    [[VECTOR_RECUR_EXTRACT:%.*]] = extractelement <4 x i32> [[WIDE_LOAD]], i64 3
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[FOR_COND_CLEANUP_LOOPEXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[SCALAR_RECUR_INIT:%.*]] = phi i32 [ [[DOTPRE]], [[FOR_PREHEADER]] ], [ [[VECTOR_RECUR_EXTRACT]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 0, [[FOR_PREHEADER]] ], [ [[N_VEC]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT:    [[BC_MERGE_RDX:%.*]] = phi i32 [ poison, [[FOR_PREHEADER]] ], [ [[TMP10]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT:    br label [[SCALAR_BODY:%.*]]
-; CHECK:       for.cond.cleanup.loopexit:
-; CHECK-NEXT:    [[MINMAX_0_COND_LCSSA:%.*]] = phi i32 [ [[TMP13:%.*]], [[SCALAR_BODY]] ], [ [[TMP10]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT:    br label [[FOR_COND_CLEANUP]]
-; CHECK:       for.cond.cleanup:
-; CHECK-NEXT:    [[MINMAX_0_LCSSA:%.*]] = phi i32 [ poison, [[ENTRY:%.*]] ], [ [[MINMAX_0_COND_LCSSA]], [[FOR_COND_CLEANUP_LOOPEXIT]] ]
-; CHECK-NEXT:    ret i32 [[MINMAX_0_LCSSA]]
-; CHECK:       scalar.body:
-; CHECK-NEXT:    [[SCALAR_RECUR:%.*]] = phi i32 [ [[SCALAR_RECUR_INIT]], [[SCALAR_PH]] ], [ [[TMP11:%.*]], [[SCALAR_BODY]] ]
-; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[SCALAR_BODY]] ]
-; CHECK-NEXT:    [[MINMAX_028:%.*]] = phi i32 [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[TMP13]], [[SCALAR_BODY]] ]
-; CHECK-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    [[TMP11]] = load i32, i32* [[ARRAYIDX]], align 4
-; CHECK-NEXT:    [[SUB3:%.*]] = sub nsw i32 [[TMP11]], [[SCALAR_RECUR]]
-; CHECK-NEXT:    [[TMP12:%.*]] = call i32 @llvm.smax.i32(i32 [[SUB3]], i32 0)
-; CHECK-NEXT:    [[TMP13]] = call i32 @llvm.smin.i32(i32 [[MINMAX_028]], i32 [[TMP12]])
-; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; CHECK-NEXT:    [[LFTR_WIDEIV:%.*]] = trunc i64 [[INDVARS_IV_NEXT]] to i32
-; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp eq i32 [[LFTR_WIDEIV]], [[N]]
-; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP_LOOPEXIT]], label [[SCALAR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
-;
-; UNROLL-LABEL: @recurrence_2(
-; UNROLL-NEXT:  entry:
-; UNROLL-NEXT:    [[CMP27:%.*]] = icmp sgt i32 [[N:%.*]], 0
-; UNROLL-NEXT:    br i1 [[CMP27]], label [[FOR_PREHEADER:%.*]], label [[FOR_COND_CLEANUP:%.*]]
-; UNROLL:       for.preheader:
-; UNROLL-NEXT:    [[ARRAYIDX2_PHI_TRANS_INSERT:%.*]] = getelementptr inbounds i32, i32* [[A:%.*]], i64 -1
-; UNROLL-NEXT:    [[DOTPRE:%.*]] = load i32, i32* [[ARRAYIDX2_PHI_TRANS_INSERT]], align 4
-; UNROLL-NEXT:    [[TMP0:%.*]] = add i32 [[N]], -1
-; UNROLL-NEXT:    [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
-; UNROLL-NEXT:    [[TMP2:%.*]] = add nuw nsw i64 [[TMP1]], 1
-; UNROLL-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[TMP0]], 7
-; UNROLL-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; UNROLL:       vector.ph:
-; UNROLL-NEXT:    [[N_VEC:%.*]] = and i64 [[TMP2]], 8589934584
-; UNROLL-NEXT:    [[VECTOR_RECUR_INIT:%.*]] = insertelement <4 x i32> poison, i32 [[DOTPRE]], i64 3
-; UNROLL-NEXT:    br label [[VECTOR_BODY:%.*]]
-; UNROLL:       vector.body:
-; UNROLL-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[VECTOR_RECUR:%.*]] = phi <4 x i32> [ [[VECTOR_RECUR_INIT]], [[VECTOR_PH]] ], [ [[WIDE_LOAD2:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ poison, [[VECTOR_PH]] ], [ [[TMP13:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[VEC_PHI1:%.*]] = phi <4 x i32> [ poison, [[VECTOR_PH]] ], [ [[TMP14:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[INDEX]]
-; UNROLL-NEXT:    [[TMP4:%.*]] = bitcast i32* [[TMP3]] to <4 x i32>*
-; UNROLL-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i32>, <4 x i32>* [[TMP4]], align 4
-; UNROLL-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i32, i32* [[TMP3]], i64 4
-; UNROLL-NEXT:    [[TMP6:%.*]] = bitcast i32* [[TMP5]] to <4 x i32>*
-; UNROLL-NEXT:    [[WIDE_LOAD2]] = load <4 x i32>, <4 x i32>* [[TMP6]], align 4
-; UNROLL-NEXT:    [[TMP7:%.*]] = shufflevector <4 x i32> [[VECTOR_RECUR]], <4 x i32> [[WIDE_LOAD]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; UNROLL-NEXT:    [[TMP8:%.*]] = shufflevector <4 x i32> [[WIDE_LOAD]], <4 x i32> [[WIDE_LOAD2]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; UNROLL-NEXT:    [[TMP9:%.*]] = sub nsw <4 x i32> [[WIDE_LOAD]], [[TMP7]]
-; UNROLL-NEXT:    [[TMP10:%.*]] = sub nsw <4 x i32> [[WIDE_LOAD2]], [[TMP8]]
-; UNROLL-NEXT:    [[TMP11:%.*]] = call <4 x i32> @llvm.smax.v4i32(<4 x i32> [[TMP9]], <4 x i32> zeroinitializer)
-; UNROLL-NEXT:    [[TMP12:%.*]] = call <4 x i32> @llvm.smax.v4i32(<4 x i32> [[TMP10]], <4 x i32> zeroinitializer)
-; UNROLL-NEXT:    [[TMP13]] = call <4 x i32> @llvm.smin.v4i32(<4 x i32> [[VEC_PHI]], <4 x i32> [[TMP11]])
-; UNROLL-NEXT:    [[TMP14]] = call <4 x i32> @llvm.smin.v4i32(<4 x i32> [[VEC_PHI1]], <4 x i32> [[TMP12]])
-; UNROLL-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
-; UNROLL-NEXT:    [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; UNROLL-NEXT:    br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
-; UNROLL:       middle.block:
-; UNROLL-NEXT:    [[TMP16:%.*]] = call <4 x i32> @llvm.smin.v4i32(<4 x i32> [[TMP13]], <4 x i32> [[TMP14]])
-; UNROLL-NEXT:    [[TMP17:%.*]] = call i32 @llvm.vector.reduce.smin.v4i32(<4 x i32> [[TMP16]])
-; UNROLL-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP2]], [[N_VEC]]
-; UNROLL-NEXT:    [[VECTOR_RECUR_EXTRACT:%.*]] = extractelement <4 x i32> [[WIDE_LOAD2]], i64 3
-; UNROLL-NEXT:    br i1 [[CMP_N]], label [[FOR_COND_CLEANUP_LOOPEXIT:%.*]], label [[SCALAR_PH]]
-; UNROLL:       scalar.ph:
-; UNROLL-NEXT:    [[SCALAR_RECUR_INIT:%.*]] = phi i32 [ [[DOTPRE]], [[FOR_PREHEADER]] ], [ [[VECTOR_RECUR_EXTRACT]], [[MIDDLE_BLOCK]] ]
-; UNROLL-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 0, [[FOR_PREHEADER]] ], [ [[N_VEC]], [[MIDDLE_BLOCK]] ]
-; UNROLL-NEXT:    [[BC_MERGE_RDX:%.*]] = phi i32 [ poison, [[FOR_PREHEADER]] ], [ [[TMP17]], [[MIDDLE_BLOCK]] ]
-; UNROLL-NEXT:    br label [[SCALAR_BODY:%.*]]
-; UNROLL:       for.cond.cleanup.loopexit:
-; UNROLL-NEXT:    [[MINMAX_0_COND_LCSSA:%.*]] = phi i32 [ [[TMP20:%.*]], [[SCALAR_BODY]] ], [ [[TMP17]], [[MIDDLE_BLOCK]] ]
-; UNROLL-NEXT:    br label [[FOR_COND_CLEANUP]]
-; UNROLL:       for.cond.cleanup:
-; UNROLL-NEXT:    [[MINMAX_0_LCSSA:%.*]] = phi i32 [ poison, [[ENTRY:%.*]] ], [ [[MINMAX_0_COND_LCSSA]], [[FOR_COND_CLEANUP_LOOPEXIT]] ]
-; UNROLL-NEXT:    ret i32 [[MINMAX_0_LCSSA]]
-; UNROLL:       scalar.body:
-; UNROLL-NEXT:    [[SCALAR_RECUR:%.*]] = phi i32 [ [[SCALAR_RECUR_INIT]], [[SCALAR_PH]] ], [ [[TMP18:%.*]], [[SCALAR_BODY]] ]
-; UNROLL-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[SCALAR_BODY]] ]
-; UNROLL-NEXT:    [[MINMAX_028:%.*]] = phi i32 [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ], [ [[TMP20]], [[SCALAR_BODY]] ]
-; UNROLL-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[A]], i64 [[INDVARS_IV]]
-; UNROLL-NEXT:    [[TMP18]] = load i32, i32* [[ARRAYIDX]], align 4
-; UNROLL-NEXT:    [[SUB3:%.*]] = sub nsw i32 [[TMP18]], [[SCALAR_RECUR]]
-; UNROLL-NEXT:    [[TMP19:%.*]] = call i32 @llvm.smax.i32(i32 [[SUB3]], i32 0)
-; UNROLL-NEXT:    [[TMP20]] = call i32 @llvm.smin.i32(i32 [[MINMAX_028]], i32 [[TMP19]])
-; UNROLL-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; UNROLL-NEXT:    [[LFTR_WIDEIV:%.*]] = trunc i64 [[INDVARS_IV_NEXT]] to i32
-; UNROLL-NEXT:    [[EXITCOND:%.*]] = icmp eq i32 [[LFTR_WIDEIV]], [[N]]
-; UNROLL-NEXT:    br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP_LOOPEXIT]], label [[SCALAR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
-;
 ; UNROLL-NO-IC-LABEL: @recurrence_2(
 ; UNROLL-NO-IC-NEXT:  entry:
 ; UNROLL-NO-IC-NEXT:    [[CMP27:%.*]] = icmp sgt i32 [[N:%.*]], 0
@@ -742,159 +499,6 @@ scalar.body:
 ;
 ;
 define void @recurrence_3(i16* readonly noalias %a, double* noalias %b, i32 %n, float %f, i16 %p) {
-; CHECK-LABEL: @recurrence_3(
-; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[TMP0:%.*]] = load i16, i16* [[A:%.*]], align 2
-; CHECK-NEXT:    [[CONV:%.*]] = sitofp i16 [[TMP0]] to double
-; CHECK-NEXT:    [[CONV1:%.*]] = fpext float [[F:%.*]] to double
-; CHECK-NEXT:    [[CONV2:%.*]] = sitofp i16 [[P:%.*]] to double
-; CHECK-NEXT:    [[MUL:%.*]] = fmul fast double [[CONV2]], [[CONV1]]
-; CHECK-NEXT:    [[SUB:%.*]] = fsub fast double [[CONV]], [[MUL]]
-; CHECK-NEXT:    store double [[SUB]], double* [[B:%.*]], align 8
-; CHECK-NEXT:    [[CMP25:%.*]] = icmp sgt i32 [[N:%.*]], 1
-; CHECK-NEXT:    br i1 [[CMP25]], label [[FOR_PREHEADER:%.*]], label [[FOR_END:%.*]]
-; CHECK:       for.preheader:
-; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[N]], -2
-; CHECK-NEXT:    [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
-; CHECK-NEXT:    [[TMP3:%.*]] = add nuw nsw i64 [[TMP2]], 1
-; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[TMP1]], 3
-; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; CHECK:       vector.ph:
-; CHECK-NEXT:    [[N_VEC:%.*]] = and i64 [[TMP3]], 8589934588
-; CHECK-NEXT:    [[IND_END:%.*]] = or i64 [[N_VEC]], 1
-; CHECK-NEXT:    [[VECTOR_RECUR_INIT:%.*]] = insertelement <4 x i16> poison, i16 [[TMP0]], i64 3
-; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x double> poison, double [[CONV1]], i64 0
-; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x double> [[BROADCAST_SPLATINSERT]], <4 x double> poison, <4 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:    [[VECTOR_RECUR:%.*]] = phi <4 x i16> [ [[VECTOR_RECUR_INIT]], [[VECTOR_PH]] ], [ [[WIDE_LOAD:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = or i64 [[INDEX]], 1
-; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i16, i16* [[A]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT:    [[TMP5:%.*]] = bitcast i16* [[TMP4]] to <4 x i16>*
-; CHECK-NEXT:    [[WIDE_LOAD]] = load <4 x i16>, <4 x i16>* [[TMP5]], align 2
-; CHECK-NEXT:    [[TMP6:%.*]] = shufflevector <4 x i16> [[VECTOR_RECUR]], <4 x i16> [[WIDE_LOAD]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; CHECK-NEXT:    [[TMP7:%.*]] = sitofp <4 x i16> [[WIDE_LOAD]] to <4 x double>
-; CHECK-NEXT:    [[TMP8:%.*]] = sitofp <4 x i16> [[TMP6]] to <4 x double>
-; CHECK-NEXT:    [[TMP9:%.*]] = fmul fast <4 x double> [[BROADCAST_SPLAT]], [[TMP8]]
-; CHECK-NEXT:    [[TMP10:%.*]] = fsub fast <4 x double> [[TMP7]], [[TMP9]]
-; CHECK-NEXT:    [[TMP11:%.*]] = getelementptr inbounds double, double* [[B]], i64 [[OFFSET_IDX]]
-; CHECK-NEXT:    [[TMP12:%.*]] = bitcast double* [[TMP11]] to <4 x double>*
-; CHECK-NEXT:    store <4 x double> [[TMP10]], <4 x double>* [[TMP12]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-NEXT:    [[TMP13:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT:    br i1 [[TMP13]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP3]], [[N_VEC]]
-; CHECK-NEXT:    [[VECTOR_RECUR_EXTRACT:%.*]] = extractelement <4 x i16> [[WIDE_LOAD]], i64 3
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[FOR_END_LOOPEXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[SCALAR_RECUR_INIT:%.*]] = phi i16 [ [[TMP0]], [[FOR_PREHEADER]] ], [ [[VECTOR_RECUR_EXTRACT]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1, [[FOR_PREHEADER]] ], [ [[IND_END]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT:    br label [[SCALAR_BODY:%.*]]
-; CHECK:       scalar.body:
-; CHECK-NEXT:    [[SCALAR_RECUR:%.*]] = phi i16 [ [[SCALAR_RECUR_INIT]], [[SCALAR_PH]] ], [ [[TMP14:%.*]], [[SCALAR_BODY]] ]
-; CHECK-NEXT:    [[ADVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[ADVARS_IV_NEXT:%.*]], [[SCALAR_BODY]] ]
-; CHECK-NEXT:    [[ARRAYIDX5:%.*]] = getelementptr inbounds i16, i16* [[A]], i64 [[ADVARS_IV]]
-; CHECK-NEXT:    [[TMP14]] = load i16, i16* [[ARRAYIDX5]], align 2
-; CHECK-NEXT:    [[CONV6:%.*]] = sitofp i16 [[TMP14]] to double
-; CHECK-NEXT:    [[CONV11:%.*]] = sitofp i16 [[SCALAR_RECUR]] to double
-; CHECK-NEXT:    [[MUL12:%.*]] = fmul fast double [[CONV11]], [[CONV1]]
-; CHECK-NEXT:    [[SUB13:%.*]] = fsub fast double [[CONV6]], [[MUL12]]
-; CHECK-NEXT:    [[ARRAYIDX15:%.*]] = getelementptr inbounds double, double* [[B]], i64 [[ADVARS_IV]]
-; CHECK-NEXT:    store double [[SUB13]], double* [[ARRAYIDX15]], align 8
-; CHECK-NEXT:    [[ADVARS_IV_NEXT]] = add nuw nsw i64 [[ADVARS_IV]], 1
-; CHECK-NEXT:    [[LFTR_WIDEIV:%.*]] = trunc i64 [[ADVARS_IV_NEXT]] to i32
-; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp eq i32 [[LFTR_WIDEIV]], [[N]]
-; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_END_LOOPEXIT]], label [[SCALAR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
-; CHECK:       for.end.loopexit:
-; CHECK-NEXT:    br label [[FOR_END]]
-; CHECK:       for.end:
-; CHECK-NEXT:    ret void
-;
-; UNROLL-LABEL: @recurrence_3(
-; UNROLL-NEXT:  entry:
-; UNROLL-NEXT:    [[TMP0:%.*]] = load i16, i16* [[A:%.*]], align 2
-; UNROLL-NEXT:    [[CONV:%.*]] = sitofp i16 [[TMP0]] to double
-; UNROLL-NEXT:    [[CONV1:%.*]] = fpext float [[F:%.*]] to double
-; UNROLL-NEXT:    [[CONV2:%.*]] = sitofp i16 [[P:%.*]] to double
-; UNROLL-NEXT:    [[MUL:%.*]] = fmul fast double [[CONV2]], [[CONV1]]
-; UNROLL-NEXT:    [[SUB:%.*]] = fsub fast double [[CONV]], [[MUL]]
-; UNROLL-NEXT:    store double [[SUB]], double* [[B:%.*]], align 8
-; UNROLL-NEXT:    [[CMP25:%.*]] = icmp sgt i32 [[N:%.*]], 1
-; UNROLL-NEXT:    br i1 [[CMP25]], label [[FOR_PREHEADER:%.*]], label [[FOR_END:%.*]]
-; UNROLL:       for.preheader:
-; UNROLL-NEXT:    [[TMP1:%.*]] = add i32 [[N]], -2
-; UNROLL-NEXT:    [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
-; UNROLL-NEXT:    [[TMP3:%.*]] = add nuw nsw i64 [[TMP2]], 1
-; UNROLL-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[TMP1]], 7
-; UNROLL-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; UNROLL:       vector.ph:
-; UNROLL-NEXT:    [[N_VEC:%.*]] = and i64 [[TMP3]], 8589934584
-; UNROLL-NEXT:    [[IND_END:%.*]] = or i64 [[N_VEC]], 1
-; UNROLL-NEXT:    [[VECTOR_RECUR_INIT:%.*]] = insertelement <4 x i16> poison, i16 [[TMP0]], i64 3
-; UNROLL-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x double> poison, double [[CONV1]], i64 0
-; UNROLL-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x double> [[BROADCAST_SPLATINSERT]], <4 x double> poison, <4 x i32> zeroinitializer
-; UNROLL-NEXT:    [[BROADCAST_SPLATINSERT2:%.*]] = insertelement <4 x double> poison, double [[CONV1]], i64 0
-; UNROLL-NEXT:    [[BROADCAST_SPLAT3:%.*]] = shufflevector <4 x double> [[BROADCAST_SPLATINSERT2]], <4 x double> poison, <4 x i32> zeroinitializer
-; UNROLL-NEXT:    br label [[VECTOR_BODY:%.*]]
-; UNROLL:       vector.body:
-; UNROLL-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[VECTOR_RECUR:%.*]] = phi <4 x i16> [ [[VECTOR_RECUR_INIT]], [[VECTOR_PH]] ], [ [[WIDE_LOAD1:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[OFFSET_IDX:%.*]] = or i64 [[INDEX]], 1
-; UNROLL-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i16, i16* [[A]], i64 [[OFFSET_IDX]]
-; UNROLL-NEXT:    [[TMP5:%.*]] = bitcast i16* [[TMP4]] to <4 x i16>*
-; UNROLL-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i16>, <4 x i16>* [[TMP5]], align 2
-; UNROLL-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i16, i16* [[TMP4]], i64 4
-; UNROLL-NEXT:    [[TMP7:%.*]] = bitcast i16* [[TMP6]] to <4 x i16>*
-; UNROLL-NEXT:    [[WIDE_LOAD1]] = load <4 x i16>, <4 x i16>* [[TMP7]], align 2
-; UNROLL-NEXT:    [[TMP8:%.*]] = shufflevector <4 x i16> [[VECTOR_RECUR]], <4 x i16> [[WIDE_LOAD]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; UNROLL-NEXT:    [[TMP9:%.*]] = shufflevector <4 x i16> [[WIDE_LOAD]], <4 x i16> [[WIDE_LOAD1]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; UNROLL-NEXT:    [[TMP10:%.*]] = sitofp <4 x i16> [[WIDE_LOAD]] to <4 x double>
-; UNROLL-NEXT:    [[TMP11:%.*]] = sitofp <4 x i16> [[WIDE_LOAD1]] to <4 x double>
-; UNROLL-NEXT:    [[TMP12:%.*]] = sitofp <4 x i16> [[TMP8]] to <4 x double>
-; UNROLL-NEXT:    [[TMP13:%.*]] = sitofp <4 x i16> [[TMP9]] to <4 x double>
-; UNROLL-NEXT:    [[TMP14:%.*]] = fmul fast <4 x double> [[BROADCAST_SPLAT]], [[TMP12]]
-; UNROLL-NEXT:    [[TMP15:%.*]] = fmul fast <4 x double> [[BROADCAST_SPLAT3]], [[TMP13]]
-; UNROLL-NEXT:    [[TMP16:%.*]] = fsub fast <4 x double> [[TMP10]], [[TMP14]]
-; UNROLL-NEXT:    [[TMP17:%.*]] = fsub fast <4 x double> [[TMP11]], [[TMP15]]
-; UNROLL-NEXT:    [[TMP18:%.*]] = getelementptr inbounds double, double* [[B]], i64 [[OFFSET_IDX]]
-; UNROLL-NEXT:    [[TMP19:%.*]] = bitcast double* [[TMP18]] to <4 x double>*
-; UNROLL-NEXT:    store <4 x double> [[TMP16]], <4 x double>* [[TMP19]], align 8
-; UNROLL-NEXT:    [[TMP20:%.*]] = getelementptr inbounds double, double* [[TMP18]], i64 4
-; UNROLL-NEXT:    [[TMP21:%.*]] = bitcast double* [[TMP20]] to <4 x double>*
-; UNROLL-NEXT:    store <4 x double> [[TMP17]], <4 x double>* [[TMP21]], align 8
-; UNROLL-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
-; UNROLL-NEXT:    [[TMP22:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; UNROLL-NEXT:    br i1 [[TMP22]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
-; UNROLL:       middle.block:
-; UNROLL-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP3]], [[N_VEC]]
-; UNROLL-NEXT:    [[VECTOR_RECUR_EXTRACT:%.*]] = extractelement <4 x i16> [[WIDE_LOAD1]], i64 3
-; UNROLL-NEXT:    br i1 [[CMP_N]], label [[FOR_END_LOOPEXIT:%.*]], label [[SCALAR_PH]]
-; UNROLL:       scalar.ph:
-; UNROLL-NEXT:    [[SCALAR_RECUR_INIT:%.*]] = phi i16 [ [[TMP0]], [[FOR_PREHEADER]] ], [ [[VECTOR_RECUR_EXTRACT]], [[MIDDLE_BLOCK]] ]
-; UNROLL-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1, [[FOR_PREHEADER]] ], [ [[IND_END]], [[MIDDLE_BLOCK]] ]
-; UNROLL-NEXT:    br label [[SCALAR_BODY:%.*]]
-; UNROLL:       scalar.body:
-; UNROLL-NEXT:    [[SCALAR_RECUR:%.*]] = phi i16 [ [[SCALAR_RECUR_INIT]], [[SCALAR_PH]] ], [ [[TMP23:%.*]], [[SCALAR_BODY]] ]
-; UNROLL-NEXT:    [[ADVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[ADVARS_IV_NEXT:%.*]], [[SCALAR_BODY]] ]
-; UNROLL-NEXT:    [[ARRAYIDX5:%.*]] = getelementptr inbounds i16, i16* [[A]], i64 [[ADVARS_IV]]
-; UNROLL-NEXT:    [[TMP23]] = load i16, i16* [[ARRAYIDX5]], align 2
-; UNROLL-NEXT:    [[CONV6:%.*]] = sitofp i16 [[TMP23]] to double
-; UNROLL-NEXT:    [[CONV11:%.*]] = sitofp i16 [[SCALAR_RECUR]] to double
-; UNROLL-NEXT:    [[MUL12:%.*]] = fmul fast double [[CONV11]], [[CONV1]]
-; UNROLL-NEXT:    [[SUB13:%.*]] = fsub fast double [[CONV6]], [[MUL12]]
-; UNROLL-NEXT:    [[ARRAYIDX15:%.*]] = getelementptr inbounds double, double* [[B]], i64 [[ADVARS_IV]]
-; UNROLL-NEXT:    store double [[SUB13]], double* [[ARRAYIDX15]], align 8
-; UNROLL-NEXT:    [[ADVARS_IV_NEXT]] = add nuw nsw i64 [[ADVARS_IV]], 1
-; UNROLL-NEXT:    [[LFTR_WIDEIV:%.*]] = trunc i64 [[ADVARS_IV_NEXT]] to i32
-; UNROLL-NEXT:    [[EXITCOND:%.*]] = icmp eq i32 [[LFTR_WIDEIV]], [[N]]
-; UNROLL-NEXT:    br i1 [[EXITCOND]], label [[FOR_END_LOOPEXIT]], label [[SCALAR_BODY]], !llvm.loop [[LOOP7:![0-9]+]]
-; UNROLL:       for.end.loopexit:
-; UNROLL-NEXT:    br label [[FOR_END]]
-; UNROLL:       for.end:
-; UNROLL-NEXT:    ret void
-;
 ; UNROLL-NO-IC-LABEL: @recurrence_3(
 ; UNROLL-NO-IC-NEXT:  entry:
 ; UNROLL-NO-IC-NEXT:    [[TMP0:%.*]] = load i16, i16* [[A:%.*]], align 2
@@ -1182,64 +786,6 @@ for.end:
 ;
 ;
 define void @PR26734(i16* %a, i32* %b, i32* %c, i32 %d, i16* %e) {
-; CHECK-LABEL: @PR26734(
-; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[CMP4:%.*]] = icmp eq i32 [[D:%.*]], 21
-; CHECK-NEXT:    br i1 [[CMP4]], label [[ENTRY_FOR_END_CRIT_EDGE:%.*]], label [[FOR_BODY_LR_PH:%.*]]
-; CHECK:       entry.for.end_crit_edge:
-; CHECK-NEXT:    br label [[FOR_END:%.*]]
-; CHECK:       for.body.lr.ph:
-; CHECK-NEXT:    [[TMP0:%.*]] = load i16, i16* [[A:%.*]], align 2
-; CHECK-NEXT:    [[SUB:%.*]] = add i16 [[TMP0]], -6
-; CHECK-NEXT:    [[CONV2:%.*]] = sext i16 [[SUB]] to i32
-; CHECK-NEXT:    [[C_PROMOTED:%.*]] = load i32, i32* [[C:%.*]], align 4
-; CHECK-NEXT:    [[B_PROMOTED:%.*]] = load i32, i32* [[B:%.*]], align 4
-; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
-; CHECK:       for.body:
-; CHECK-NEXT:    [[INC7:%.*]] = phi i32 [ [[D]], [[FOR_BODY_LR_PH]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT:    [[AND6:%.*]] = phi i32 [ [[B_PROMOTED]], [[FOR_BODY_LR_PH]] ], [ [[AND:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT:    [[CONV25:%.*]] = phi i32 [ [[C_PROMOTED]], [[FOR_BODY_LR_PH]] ], [ [[CONV2]], [[FOR_BODY]] ]
-; CHECK-NEXT:    [[AND]] = and i32 [[AND6]], [[CONV25]]
-; CHECK-NEXT:    [[INC]] = add nsw i32 [[INC7]], 1
-; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[INC]], 21
-; CHECK-NEXT:    br i1 [[CMP]], label [[FOR_COND_FOR_END_CRIT_EDGE:%.*]], label [[FOR_BODY]]
-; CHECK:       for.cond.for.end_crit_edge:
-; CHECK-NEXT:    store i32 [[CONV2]], i32* [[C]], align 4
-; CHECK-NEXT:    store i32 [[AND]], i32* [[B]], align 4
-; CHECK-NEXT:    store i16 [[SUB]], i16* [[E:%.*]], align 2
-; CHECK-NEXT:    br label [[FOR_END]]
-; CHECK:       for.end:
-; CHECK-NEXT:    ret void
-;
-; UNROLL-LABEL: @PR26734(
-; UNROLL-NEXT:  entry:
-; UNROLL-NEXT:    [[CMP4:%.*]] = icmp eq i32 [[D:%.*]], 21
-; UNROLL-NEXT:    br i1 [[CMP4]], label [[ENTRY_FOR_END_CRIT_EDGE:%.*]], label [[FOR_BODY_LR_PH:%.*]]
-; UNROLL:       entry.for.end_crit_edge:
-; UNROLL-NEXT:    br label [[FOR_END:%.*]]
-; UNROLL:       for.body.lr.ph:
-; UNROLL-NEXT:    [[TMP0:%.*]] = load i16, i16* [[A:%.*]], align 2
-; UNROLL-NEXT:    [[SUB:%.*]] = add i16 [[TMP0]], -6
-; UNROLL-NEXT:    [[CONV2:%.*]] = sext i16 [[SUB]] to i32
-; UNROLL-NEXT:    [[C_PROMOTED:%.*]] = load i32, i32* [[C:%.*]], align 4
-; UNROLL-NEXT:    [[B_PROMOTED:%.*]] = load i32, i32* [[B:%.*]], align 4
-; UNROLL-NEXT:    br label [[FOR_BODY:%.*]]
-; UNROLL:       for.body:
-; UNROLL-NEXT:    [[INC7:%.*]] = phi i32 [ [[D]], [[FOR_BODY_LR_PH]] ], [ [[INC:%.*]], [[FOR_BODY]] ]
-; UNROLL-NEXT:    [[AND6:%.*]] = phi i32 [ [[B_PROMOTED]], [[FOR_BODY_LR_PH]] ], [ [[AND:%.*]], [[FOR_BODY]] ]
-; UNROLL-NEXT:    [[CONV25:%.*]] = phi i32 [ [[C_PROMOTED]], [[FOR_BODY_LR_PH]] ], [ [[CONV2]], [[FOR_BODY]] ]
-; UNROLL-NEXT:    [[AND]] = and i32 [[AND6]], [[CONV25]]
-; UNROLL-NEXT:    [[INC]] = add nsw i32 [[INC7]], 1
-; UNROLL-NEXT:    [[CMP:%.*]] = icmp eq i32 [[INC]], 21
-; UNROLL-NEXT:    br i1 [[CMP]], label [[FOR_COND_FOR_END_CRIT_EDGE:%.*]], label [[FOR_BODY]]
-; UNROLL:       for.cond.for.end_crit_edge:
-; UNROLL-NEXT:    store i32 [[CONV2]], i32* [[C]], align 4
-; UNROLL-NEXT:    store i32 [[AND]], i32* [[B]], align 4
-; UNROLL-NEXT:    store i16 [[SUB]], i16* [[E:%.*]], align 2
-; UNROLL-NEXT:    br label [[FOR_END]]
-; UNROLL:       for.end:
-; UNROLL-NEXT:    ret void
-;
 ; UNROLL-NO-IC-LABEL: @PR26734(
 ; UNROLL-NO-IC-NEXT:  entry:
 ; UNROLL-NO-IC-NEXT:    [[CMP4:%.*]] = icmp eq i32 [[D:%.*]], 21
@@ -1381,46 +927,6 @@ for.end:
 ;
 ;
 define i32 @PR27246() {
-; CHECK-LABEL: @PR27246(
-; CHECK-NEXT:  entry:
-; CHECK-NEXT:    br label [[FOR_COND1_PREHEADER:%.*]]
-; CHECK:       for.cond1.preheader:
-; CHECK-NEXT:    [[I_016:%.*]] = phi i32 [ 1, [[ENTRY:%.*]] ], [ [[INC:%.*]], [[FOR_COND_CLEANUP3:%.*]] ]
-; CHECK-NEXT:    [[E_015:%.*]] = phi i32 [ poison, [[ENTRY]] ], [ [[E_1:%.*]], [[FOR_COND_CLEANUP3]] ]
-; CHECK-NEXT:    br label [[FOR_COND1:%.*]]
-; CHECK:       for.cond.cleanup:
-; CHECK-NEXT:    ret i32 [[E_1]]
-; CHECK:       for.cond1:
-; CHECK-NEXT:    [[E_1]] = phi i32 [ [[K_0:%.*]], [[FOR_COND1]] ], [ [[E_015]], [[FOR_COND1_PREHEADER]] ]
-; CHECK-NEXT:    [[K_0]] = phi i32 [ [[DEC:%.*]], [[FOR_COND1]] ], [ [[I_016]], [[FOR_COND1_PREHEADER]] ]
-; CHECK-NEXT:    [[CMP2:%.*]] = icmp sgt i32 [[K_0]], 1
-; CHECK-NEXT:    [[DEC]] = add nsw i32 [[K_0]], -1
-; CHECK-NEXT:    br i1 [[CMP2]], label [[FOR_COND1]], label [[FOR_COND_CLEANUP3]]
-; CHECK:       for.cond.cleanup3:
-; CHECK-NEXT:    [[INC]] = add nuw nsw i32 [[I_016]], 1
-; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 49
-; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_COND1_PREHEADER]]
-;
-; UNROLL-LABEL: @PR27246(
-; UNROLL-NEXT:  entry:
-; UNROLL-NEXT:    br label [[FOR_COND1_PREHEADER:%.*]]
-; UNROLL:       for.cond1.preheader:
-; UNROLL-NEXT:    [[I_016:%.*]] = phi i32 [ 1, [[ENTRY:%.*]] ], [ [[INC:%.*]], [[FOR_COND_CLEANUP3:%.*]] ]
-; UNROLL-NEXT:    [[E_015:%.*]] = phi i32 [ poison, [[ENTRY]] ], [ [[E_1:%.*]], [[FOR_COND_CLEANUP3]] ]
-; UNROLL-NEXT:    br label [[FOR_COND1:%.*]]
-; UNROLL:       for.cond.cleanup:
-; UNROLL-NEXT:    ret i32 [[E_1]]
-; UNROLL:       for.cond1:
-; UNROLL-NEXT:    [[E_1]] = phi i32 [ [[K_0:%.*]], [[FOR_COND1]] ], [ [[E_015]], [[FOR_COND1_PREHEADER]] ]
-; UNROLL-NEXT:    [[K_0]] = phi i32 [ [[DEC:%.*]], [[FOR_COND1]] ], [ [[I_016]], [[FOR_COND1_PREHEADER]] ]
-; UNROLL-NEXT:    [[CMP2:%.*]] = icmp sgt i32 [[K_0]], 1
-; UNROLL-NEXT:    [[DEC]] = add nsw i32 [[K_0]], -1
-; UNROLL-NEXT:    br i1 [[CMP2]], label [[FOR_COND1]], label [[FOR_COND_CLEANUP3]]
-; UNROLL:       for.cond.cleanup3:
-; UNROLL-NEXT:    [[INC]] = add nuw nsw i32 [[I_016]], 1
-; UNROLL-NEXT:    [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 49
-; UNROLL-NEXT:    br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_COND1_PREHEADER]]
-;
 ; UNROLL-NO-IC-LABEL: @PR27246(
 ; UNROLL-NO-IC-NEXT:  entry:
 ; UNROLL-NO-IC-NEXT:    br label [[FOR_COND1_PREHEADER:%.*]]
@@ -1515,66 +1021,6 @@ for.cond.cleanup3:
 
 ;
 define void @PR30183(i32 %pre_load, i32* %a, i32* %b, i64 %n) {
-; CHECK-LABEL: @PR30183(
-; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[N:%.*]], -2
-; CHECK-NEXT:    [[TMP1:%.*]] = lshr i64 [[TMP0]], 1
-; CHECK-NEXT:    [[TMP2:%.*]] = add nuw i64 [[TMP1]], 1
-; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP0]], 6
-; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; CHECK:       vector.ph:
-; CHECK-NEXT:    [[N_VEC:%.*]] = and i64 [[TMP2]], -4
-; CHECK-NEXT:    [[IND_END:%.*]] = shl i64 [[N_VEC]], 1
-; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
-; CHECK:       vector.body:
-; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-NEXT:    [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT:    br i1 [[TMP3]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP2]], [[N_VEC]]
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    br label [[SCALAR_BODY:%.*]]
-; CHECK:       scalar.body:
-; CHECK-NEXT:    [[I:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[I_NEXT:%.*]], [[SCALAR_BODY]] ]
-; CHECK-NEXT:    [[I_NEXT]] = add nuw nsw i64 [[I]], 2
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[I_NEXT]], [[N]]
-; CHECK-NEXT:    br i1 [[COND]], label [[FOR_END]], label [[SCALAR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
-; CHECK:       for.end:
-; CHECK-NEXT:    ret void
-;
-; UNROLL-LABEL: @PR30183(
-; UNROLL-NEXT:  entry:
-; UNROLL-NEXT:    [[TMP0:%.*]] = add i64 [[N:%.*]], -2
-; UNROLL-NEXT:    [[TMP1:%.*]] = lshr i64 [[TMP0]], 1
-; UNROLL-NEXT:    [[TMP2:%.*]] = add nuw i64 [[TMP1]], 1
-; UNROLL-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP0]], 14
-; UNROLL-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; UNROLL:       vector.ph:
-; UNROLL-NEXT:    [[N_VEC:%.*]] = and i64 [[TMP2]], -8
-; UNROLL-NEXT:    [[IND_END:%.*]] = shl i64 [[N_VEC]], 1
-; UNROLL-NEXT:    br label [[VECTOR_BODY:%.*]]
-; UNROLL:       vector.body:
-; UNROLL-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
-; UNROLL-NEXT:    [[TMP3:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; UNROLL-NEXT:    br i1 [[TMP3]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
-; UNROLL:       middle.block:
-; UNROLL-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[TMP2]], [[N_VEC]]
-; UNROLL-NEXT:    br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
-; UNROLL:       scalar.ph:
-; UNROLL-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[IND_END]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
-; UNROLL-NEXT:    br label [[SCALAR_BODY:%.*]]
-; UNROLL:       scalar.body:
-; UNROLL-NEXT:    [[I:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[I_NEXT:%.*]], [[SCALAR_BODY]] ]
-; UNROLL-NEXT:    [[I_NEXT]] = add nuw nsw i64 [[I]], 2
-; UNROLL-NEXT:    [[COND:%.*]] = icmp eq i64 [[I_NEXT]], [[N]]
-; UNROLL-NEXT:    br i1 [[COND]], label [[FOR_END]], label [[SCALAR_BODY]], !llvm.loop [[LOOP9:![0-9]+]]
-; UNROLL:       for.end:
-; UNROLL-NEXT:    ret void
-;
 ; UNROLL-NO-IC-LABEL: @PR30183(
 ; UNROLL-NO-IC-NEXT:  entry:
 ; UNROLL-NO-IC-NEXT:    [[TMP0:%.*]] = add i64 [[N:%.*]], -2
@@ -1781,38 +1227,6 @@ for.end:
 
 ;
 define void @constant_folded_previous_value() {
-; CHECK-LABEL: @constant_folded_previous_value(
-; CHECK-NEXT:  entry:
-; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; CHECK:       vector.ph:
-; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
-; CHECK:       vector.body:
-; CHECK-NEXT:    br i1 undef, label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    br i1 undef, label [[FOR_END:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    br label [[SCALAR_BODY:%.*]]
-; CHECK:       scalar.body:
-; CHECK-NEXT:    br i1 undef, label [[FOR_END]], label [[SCALAR_BODY]], !llvm.loop [[LOOP11:![0-9]+]]
-; CHECK:       for.end:
-; CHECK-NEXT:    ret void
-;
-; UNROLL-LABEL: @constant_folded_previous_value(
-; UNROLL-NEXT:  entry:
-; UNROLL-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; UNROLL:       vector.ph:
-; UNROLL-NEXT:    br label [[VECTOR_BODY:%.*]]
-; UNROLL:       vector.body:
-; UNROLL-NEXT:    br i1 undef, label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
-; UNROLL:       middle.block:
-; UNROLL-NEXT:    br i1 undef, label [[FOR_END:%.*]], label [[SCALAR_PH]]
-; UNROLL:       scalar.ph:
-; UNROLL-NEXT:    br label [[SCALAR_BODY:%.*]]
-; UNROLL:       scalar.body:
-; UNROLL-NEXT:    br i1 undef, label [[FOR_END]], label [[SCALAR_BODY]], !llvm.loop [[LOOP11:![0-9]+]]
-; UNROLL:       for.end:
-; UNROLL-NEXT:    ret void
-;
 ; UNROLL-NO-IC-LABEL: @constant_folded_previous_value(
 ; UNROLL-NO-IC-NEXT:  entry:
 ; UNROLL-NO-IC-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
@@ -1923,53 +1337,6 @@ for.end:
 ; itself and not its update (addx).
 ; Check the case when unrolled but not vectorized.
 define i32 @extract_second_last_iteration(i32* %cval, i32 %x)  {
-; CHECK-LABEL: @extract_second_last_iteration(
-; CHECK-NEXT:  entry:
-; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; CHECK:       vector.ph:
-; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
-; CHECK:       vector.body:
-; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = phi i32 [ 2, [[VECTOR_PH]] ], [ [[TMP1:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
-; CHECK-NEXT:    [[TMP1]] = add i32 [[TMP0]], 4
-; CHECK-NEXT:    [[TMP2:%.*]] = icmp eq i32 [[INDEX_NEXT]], 96
-; CHECK-NEXT:    br i1 [[TMP2]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[VECTOR_RECUR_EXTRACT_FOR_PHI:%.*]] = add i32 [[TMP0]], [[X:%.*]]
-; CHECK-NEXT:    br i1 true, label [[FOR_END:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
-; CHECK:       for.body:
-; CHECK-NEXT:    br i1 undef, label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP13:![0-9]+]]
-; CHECK:       for.end:
-; CHECK-NEXT:    [[VAL_PHI_LCSSA:%.*]] = phi i32 [ undef, [[FOR_BODY]] ], [ [[VECTOR_RECUR_EXTRACT_FOR_PHI]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT:    ret i32 [[VAL_PHI_LCSSA]]
-;
-; UNROLL-LABEL: @extract_second_last_iteration(
-; UNROLL-NEXT:  entry:
-; UNROLL-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; UNROLL:       vector.ph:
-; UNROLL-NEXT:    br label [[VECTOR_BODY:%.*]]
-; UNROLL:       vector.body:
-; UNROLL-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[TMP0:%.*]] = phi i32 [ 2, [[VECTOR_PH]] ], [ [[TMP1:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 8
-; UNROLL-NEXT:    [[TMP1]] = add i32 [[TMP0]], 8
-; UNROLL-NEXT:    [[TMP2:%.*]] = icmp eq i32 [[INDEX_NEXT]], 96
-; UNROLL-NEXT:    br i1 [[TMP2]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
-; UNROLL:       middle.block:
-; UNROLL-NEXT:    [[TMP3:%.*]] = add i32 [[TMP0]], 4
-; UNROLL-NEXT:    [[VECTOR_RECUR_EXTRACT_FOR_PHI:%.*]] = add i32 [[TMP3]], [[X:%.*]]
-; UNROLL-NEXT:    br i1 true, label [[FOR_END:%.*]], label [[SCALAR_PH]]
-; UNROLL:       scalar.ph:
-; UNROLL-NEXT:    br label [[FOR_BODY:%.*]]
-; UNROLL:       for.body:
-; UNROLL-NEXT:    br i1 undef, label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP13:![0-9]+]]
-; UNROLL:       for.end:
-; UNROLL-NEXT:    [[VAL_PHI_LCSSA:%.*]] = phi i32 [ undef, [[FOR_BODY]] ], [ [[VECTOR_RECUR_EXTRACT_FOR_PHI]], [[MIDDLE_BLOCK]] ]
-; UNROLL-NEXT:    ret i32 [[VAL_PHI_LCSSA]]
-;
 ; UNROLL-NO-IC-LABEL: @extract_second_last_iteration(
 ; UNROLL-NO-IC-NEXT:  entry:
 ; UNROLL-NO-IC-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
@@ -2120,143 +1487,6 @@ for.end:
 ;
 ;
 define i32 @PR33613(double* %b, double %j, i32 %d) {
-; CHECK-LABEL: @PR33613(
-; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[IDXPROM:%.*]] = sext i32 [[D:%.*]] to i64
-; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; CHECK:       vector.ph:
-; CHECK-NEXT:    [[VECTOR_RECUR_INIT:%.*]] = insertelement <4 x double> poison, double [[J:%.*]], i64 3
-; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
-; CHECK:       vector.body:
-; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP23:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VECTOR_RECUR:%.*]] = phi <4 x double> [ [[VECTOR_RECUR_INIT]], [[VECTOR_PH]] ], [ [[TMP18:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 25
-; CHECK-NEXT:    [[NEXT_GEP5:%.*]] = getelementptr double, double* [[B:%.*]], i64 [[IDXPROM]]
-; CHECK-NEXT:    [[TMP1:%.*]] = or i64 [[INDEX]], 1
-; CHECK-NEXT:    [[TMP2:%.*]] = mul i64 [[TMP1]], 25
-; CHECK-NEXT:    [[NEXT_GEP26:%.*]] = getelementptr double, double* [[B]], i64 [[IDXPROM]]
-; CHECK-NEXT:    [[TMP3:%.*]] = or i64 [[INDEX]], 2
-; CHECK-NEXT:    [[TMP4:%.*]] = mul i64 [[TMP3]], 25
-; CHECK-NEXT:    [[NEXT_GEP37:%.*]] = getelementptr double, double* [[B]], i64 [[IDXPROM]]
-; CHECK-NEXT:    [[TMP5:%.*]] = or i64 [[INDEX]], 3
-; CHECK-NEXT:    [[TMP6:%.*]] = mul i64 [[TMP5]], 25
-; CHECK-NEXT:    [[NEXT_GEP48:%.*]] = getelementptr double, double* [[B]], i64 [[IDXPROM]]
-; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr double, double* [[NEXT_GEP5]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr double, double* [[NEXT_GEP26]], i64 [[TMP2]]
-; CHECK-NEXT:    [[TMP9:%.*]] = getelementptr double, double* [[NEXT_GEP37]], i64 [[TMP4]]
-; CHECK-NEXT:    [[TMP10:%.*]] = getelementptr double, double* [[NEXT_GEP48]], i64 [[TMP6]]
-; CHECK-NEXT:    [[TMP11:%.*]] = load double, double* [[TMP7]], align 8
-; CHECK-NEXT:    [[TMP12:%.*]] = load double, double* [[TMP8]], align 8
-; CHECK-NEXT:    [[TMP13:%.*]] = load double, double* [[TMP9]], align 8
-; CHECK-NEXT:    [[TMP14:%.*]] = load double, double* [[TMP10]], align 8
-; CHECK-NEXT:    [[TMP15:%.*]] = insertelement <4 x double> poison, double [[TMP11]], i64 0
-; CHECK-NEXT:    [[TMP16:%.*]] = insertelement <4 x double> [[TMP15]], double [[TMP12]], i64 1
-; CHECK-NEXT:    [[TMP17:%.*]] = insertelement <4 x double> [[TMP16]], double [[TMP13]], i64 2
-; CHECK-NEXT:    [[TMP18]] = insertelement <4 x double> [[TMP17]], double [[TMP14]], i64 3
-; CHECK-NEXT:    [[TMP19:%.*]] = shufflevector <4 x double> [[VECTOR_RECUR]], <4 x double> [[TMP17]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; CHECK-NEXT:    [[TMP20:%.*]] = fmul <4 x double> [[TMP19]], [[TMP18]]
-; CHECK-NEXT:    [[TMP21:%.*]] = fcmp une <4 x double> [[TMP20]], zeroinitializer
-; CHECK-NEXT:    [[TMP22:%.*]] = zext <4 x i1> [[TMP21]] to <4 x i32>
-; CHECK-NEXT:    [[TMP23]] = add <4 x i32> [[VEC_PHI]], [[TMP22]]
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-NEXT:    [[TMP24:%.*]] = icmp eq i64 [[INDEX_NEXT]], 10240
-; CHECK-NEXT:    br i1 [[TMP24]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[TMP25:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP23]])
-; CHECK-NEXT:    br i1 true, label [[FOR_COND_CLEANUP:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
-; CHECK:       for.cond.cleanup:
-; CHECK-NEXT:    [[A_1_LCSSA:%.*]] = phi i32 [ undef, [[FOR_BODY]] ], [ [[TMP25]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT:    ret i32 [[A_1_LCSSA]]
-; CHECK:       for.body:
-; CHECK-NEXT:    br i1 undef, label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop [[LOOP15:![0-9]+]]
-;
-; UNROLL-LABEL: @PR33613(
-; UNROLL-NEXT:  entry:
-; UNROLL-NEXT:    [[IDXPROM:%.*]] = sext i32 [[D:%.*]] to i64
-; UNROLL-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; UNROLL:       vector.ph:
-; UNROLL-NEXT:    [[VECTOR_RECUR_INIT:%.*]] = insertelement <4 x double> poison, double [[J:%.*]], i64 3
-; UNROLL-NEXT:    br label [[VECTOR_BODY:%.*]]
-; UNROLL:       vector.body:
-; UNROLL-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP47:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[VEC_PHI9:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP48:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[VECTOR_RECUR:%.*]] = phi <4 x double> [ [[VECTOR_RECUR_INIT]], [[VECTOR_PH]] ], [ [[TMP38:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 25
-; UNROLL-NEXT:    [[NEXT_GEP10:%.*]] = getelementptr double, double* [[B:%.*]], i64 [[IDXPROM]]
-; UNROLL-NEXT:    [[TMP1:%.*]] = or i64 [[INDEX]], 1
-; UNROLL-NEXT:    [[TMP2:%.*]] = mul i64 [[TMP1]], 25
-; UNROLL-NEXT:    [[NEXT_GEP211:%.*]] = getelementptr double, double* [[B]], i64 [[IDXPROM]]
-; UNROLL-NEXT:    [[TMP3:%.*]] = or i64 [[INDEX]], 2
-; UNROLL-NEXT:    [[TMP4:%.*]] = mul i64 [[TMP3]], 25
-; UNROLL-NEXT:    [[NEXT_GEP312:%.*]] = getelementptr double, double* [[B]], i64 [[IDXPROM]]
-; UNROLL-NEXT:    [[TMP5:%.*]] = or i64 [[INDEX]], 3
-; UNROLL-NEXT:    [[TMP6:%.*]] = mul i64 [[TMP5]], 25
-; UNROLL-NEXT:    [[NEXT_GEP413:%.*]] = getelementptr double, double* [[B]], i64 [[IDXPROM]]
-; UNROLL-NEXT:    [[TMP7:%.*]] = or i64 [[INDEX]], 4
-; UNROLL-NEXT:    [[TMP8:%.*]] = mul i64 [[TMP7]], 25
-; UNROLL-NEXT:    [[NEXT_GEP514:%.*]] = getelementptr double, double* [[B]], i64 [[IDXPROM]]
-; UNROLL-NEXT:    [[TMP9:%.*]] = or i64 [[INDEX]], 5
-; UNROLL-NEXT:    [[TMP10:%.*]] = mul i64 [[TMP9]], 25
-; UNROLL-NEXT:    [[NEXT_GEP615:%.*]] = getelementptr double, double* [[B]], i64 [[IDXPROM]]
-; UNROLL-NEXT:    [[TMP11:%.*]] = or i64 [[INDEX]], 6
-; UNROLL-NEXT:    [[TMP12:%.*]] = mul i64 [[TMP11]], 25
-; UNROLL-NEXT:    [[NEXT_GEP716:%.*]] = getelementptr double, double* [[B]], i64 [[IDXPROM]]
-; UNROLL-NEXT:    [[TMP13:%.*]] = or i64 [[INDEX]], 7
-; UNROLL-NEXT:    [[TMP14:%.*]] = mul i64 [[TMP13]], 25
-; UNROLL-NEXT:    [[NEXT_GEP817:%.*]] = getelementptr double, double* [[B]], i64 [[IDXPROM]]
-; UNROLL-NEXT:    [[TMP15:%.*]] = getelementptr double, double* [[NEXT_GEP10]], i64 [[TMP0]]
-; UNROLL-NEXT:    [[TMP16:%.*]] = getelementptr double, double* [[NEXT_GEP211]], i64 [[TMP2]]
-; UNROLL-NEXT:    [[TMP17:%.*]] = getelementptr double, double* [[NEXT_GEP312]], i64 [[TMP4]]
-; UNROLL-NEXT:    [[TMP18:%.*]] = getelementptr double, double* [[NEXT_GEP413]], i64 [[TMP6]]
-; UNROLL-NEXT:    [[TMP19:%.*]] = getelementptr double, double* [[NEXT_GEP514]], i64 [[TMP8]]
-; UNROLL-NEXT:    [[TMP20:%.*]] = getelementptr double, double* [[NEXT_GEP615]], i64 [[TMP10]]
-; UNROLL-NEXT:    [[TMP21:%.*]] = getelementptr double, double* [[NEXT_GEP716]], i64 [[TMP12]]
-; UNROLL-NEXT:    [[TMP22:%.*]] = getelementptr double, double* [[NEXT_GEP817]], i64 [[TMP14]]
-; UNROLL-NEXT:    [[TMP23:%.*]] = load double, double* [[TMP15]], align 8
-; UNROLL-NEXT:    [[TMP24:%.*]] = load double, double* [[TMP16]], align 8
-; UNROLL-NEXT:    [[TMP25:%.*]] = load double, double* [[TMP17]], align 8
-; UNROLL-NEXT:    [[TMP26:%.*]] = load double, double* [[TMP18]], align 8
-; UNROLL-NEXT:    [[TMP27:%.*]] = insertelement <4 x double> poison, double [[TMP23]], i64 0
-; UNROLL-NEXT:    [[TMP28:%.*]] = insertelement <4 x double> [[TMP27]], double [[TMP24]], i64 1
-; UNROLL-NEXT:    [[TMP29:%.*]] = insertelement <4 x double> [[TMP28]], double [[TMP25]], i64 2
-; UNROLL-NEXT:    [[TMP30:%.*]] = insertelement <4 x double> [[TMP29]], double [[TMP26]], i64 3
-; UNROLL-NEXT:    [[TMP31:%.*]] = load double, double* [[TMP19]], align 8
-; UNROLL-NEXT:    [[TMP32:%.*]] = load double, double* [[TMP20]], align 8
-; UNROLL-NEXT:    [[TMP33:%.*]] = load double, double* [[TMP21]], align 8
-; UNROLL-NEXT:    [[TMP34:%.*]] = load double, double* [[TMP22]], align 8
-; UNROLL-NEXT:    [[TMP35:%.*]] = insertelement <4 x double> poison, double [[TMP31]], i64 0
-; UNROLL-NEXT:    [[TMP36:%.*]] = insertelement <4 x double> [[TMP35]], double [[TMP32]], i64 1
-; UNROLL-NEXT:    [[TMP37:%.*]] = insertelement <4 x double> [[TMP36]], double [[TMP33]], i64 2
-; UNROLL-NEXT:    [[TMP38]] = insertelement <4 x double> [[TMP37]], double [[TMP34]], i64 3
-; UNROLL-NEXT:    [[TMP39:%.*]] = shufflevector <4 x double> [[VECTOR_RECUR]], <4 x double> [[TMP29]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; UNROLL-NEXT:    [[TMP40:%.*]] = shufflevector <4 x double> [[TMP30]], <4 x double> [[TMP37]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; UNROLL-NEXT:    [[TMP41:%.*]] = fmul <4 x double> [[TMP39]], [[TMP30]]
-; UNROLL-NEXT:    [[TMP42:%.*]] = fmul <4 x double> [[TMP40]], [[TMP38]]
-; UNROLL-NEXT:    [[TMP43:%.*]] = fcmp une <4 x double> [[TMP41]], zeroinitializer
-; UNROLL-NEXT:    [[TMP44:%.*]] = fcmp une <4 x double> [[TMP42]], zeroinitializer
-; UNROLL-NEXT:    [[TMP45:%.*]] = zext <4 x i1> [[TMP43]] to <4 x i32>
-; UNROLL-NEXT:    [[TMP46:%.*]] = zext <4 x i1> [[TMP44]] to <4 x i32>
-; UNROLL-NEXT:    [[TMP47]] = add <4 x i32> [[VEC_PHI]], [[TMP45]]
-; UNROLL-NEXT:    [[TMP48]] = add <4 x i32> [[VEC_PHI9]], [[TMP46]]
-; UNROLL-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
-; UNROLL-NEXT:    [[TMP49:%.*]] = icmp eq i64 [[INDEX_NEXT]], 10240
-; UNROLL-NEXT:    br i1 [[TMP49]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
-; UNROLL:       middle.block:
-; UNROLL-NEXT:    [[BIN_RDX:%.*]] = add <4 x i32> [[TMP48]], [[TMP47]]
-; UNROLL-NEXT:    [[TMP50:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[BIN_RDX]])
-; UNROLL-NEXT:    br i1 true, label [[FOR_COND_CLEANUP:%.*]], label [[SCALAR_PH]]
-; UNROLL:       scalar.ph:
-; UNROLL-NEXT:    br label [[FOR_BODY:%.*]]
-; UNROLL:       for.cond.cleanup:
-; UNROLL-NEXT:    [[A_1_LCSSA:%.*]] = phi i32 [ undef, [[FOR_BODY]] ], [ [[TMP50]], [[MIDDLE_BLOCK]] ]
-; UNROLL-NEXT:    ret i32 [[A_1_LCSSA]]
-; UNROLL:       for.body:
-; UNROLL-NEXT:    br i1 undef, label [[FOR_COND_CLEANUP]], label [[FOR_BODY]], !llvm.loop [[LOOP15:![0-9]+]]
-;
 ; UNROLL-NO-IC-LABEL: @PR33613(
 ; UNROLL-NO-IC-NEXT:  entry:
 ; UNROLL-NO-IC-NEXT:    [[IDXPROM:%.*]] = sext i32 [[D:%.*]] to i64
@@ -2533,116 +1763,6 @@ for.body:
 ; Check that the sext sank after the load in the vector loop.
 ;
 define void @sink_after(i16* noalias %a, i32* noalias %b, i64 %n) {
-; CHECK-LABEL: @sink_after(
-; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[DOTPRE:%.*]] = load i16, i16* [[A:%.*]], align 2
-; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N:%.*]], 4
-; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; CHECK:       vector.ph:
-; CHECK-NEXT:    [[N_VEC:%.*]] = and i64 [[N]], -4
-; CHECK-NEXT:    [[VECTOR_RECUR_INIT:%.*]] = insertelement <4 x i16> poison, i16 [[DOTPRE]], i64 3
-; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
-; CHECK:       vector.body:
-; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VECTOR_RECUR:%.*]] = phi <4 x i16> [ [[VECTOR_RECUR_INIT]], [[VECTOR_PH]] ], [ [[WIDE_LOAD:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = or i64 [[INDEX]], 1
-; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i16, i16* [[A]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP2:%.*]] = bitcast i16* [[TMP1]] to <4 x i16>*
-; CHECK-NEXT:    [[WIDE_LOAD]] = load <4 x i16>, <4 x i16>* [[TMP2]], align 2
-; CHECK-NEXT:    [[TMP3:%.*]] = shufflevector <4 x i16> [[VECTOR_RECUR]], <4 x i16> [[WIDE_LOAD]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; CHECK-NEXT:    [[TMP4:%.*]] = sext <4 x i16> [[TMP3]] to <4 x i32>
-; CHECK-NEXT:    [[TMP5:%.*]] = sext <4 x i16> [[WIDE_LOAD]] to <4 x i32>
-; CHECK-NEXT:    [[TMP6:%.*]] = mul nsw <4 x i32> [[TMP5]], [[TMP4]]
-; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i32, i32* [[B:%.*]], i64 [[INDEX]]
-; CHECK-NEXT:    [[TMP8:%.*]] = bitcast i32* [[TMP7]] to <4 x i32>*
-; CHECK-NEXT:    store <4 x i32> [[TMP6]], <4 x i32>* [[TMP8]], align 4
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-NEXT:    [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT:    br i1 [[TMP9]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[N_VEC]], [[N]]
-; CHECK-NEXT:    [[VECTOR_RECUR_EXTRACT:%.*]] = extractelement <4 x i16> [[WIDE_LOAD]], i64 3
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[SCALAR_RECUR_INIT:%.*]] = phi i16 [ [[DOTPRE]], [[ENTRY:%.*]] ], [ [[VECTOR_RECUR_EXTRACT]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 0, [[ENTRY]] ], [ [[N_VEC]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
-; CHECK:       for.body:
-; CHECK-NEXT:    [[SCALAR_RECUR:%.*]] = phi i16 [ [[SCALAR_RECUR_INIT]], [[SCALAR_PH]] ], [ [[TMP10:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT:    [[CONV:%.*]] = sext i16 [[SCALAR_RECUR]] to i32
-; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; CHECK-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[A]], i64 [[INDVARS_IV_NEXT]]
-; CHECK-NEXT:    [[TMP10]] = load i16, i16* [[ARRAYIDX2]], align 2
-; CHECK-NEXT:    [[CONV3:%.*]] = sext i16 [[TMP10]] to i32
-; CHECK-NEXT:    [[MUL:%.*]] = mul nsw i32 [[CONV3]], [[CONV]]
-; CHECK-NEXT:    [[ARRAYIDX5:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    store i32 [[MUL]], i32* [[ARRAYIDX5]], align 4
-; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[N]]
-; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP17:![0-9]+]]
-; CHECK:       for.end:
-; CHECK-NEXT:    ret void
-;
-; UNROLL-LABEL: @sink_after(
-; UNROLL-NEXT:  entry:
-; UNROLL-NEXT:    [[DOTPRE:%.*]] = load i16, i16* [[A:%.*]], align 2
-; UNROLL-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N:%.*]], 8
-; UNROLL-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; UNROLL:       vector.ph:
-; UNROLL-NEXT:    [[N_VEC:%.*]] = and i64 [[N]], -8
-; UNROLL-NEXT:    [[VECTOR_RECUR_INIT:%.*]] = insertelement <4 x i16> poison, i16 [[DOTPRE]], i64 3
-; UNROLL-NEXT:    br label [[VECTOR_BODY:%.*]]
-; UNROLL:       vector.body:
-; UNROLL-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[VECTOR_RECUR:%.*]] = phi <4 x i16> [ [[VECTOR_RECUR_INIT]], [[VECTOR_PH]] ], [ [[WIDE_LOAD1:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[TMP0:%.*]] = or i64 [[INDEX]], 1
-; UNROLL-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i16, i16* [[A]], i64 [[TMP0]]
-; UNROLL-NEXT:    [[TMP2:%.*]] = bitcast i16* [[TMP1]] to <4 x i16>*
-; UNROLL-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i16>, <4 x i16>* [[TMP2]], align 2
-; UNROLL-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i16, i16* [[TMP1]], i64 4
-; UNROLL-NEXT:    [[TMP4:%.*]] = bitcast i16* [[TMP3]] to <4 x i16>*
-; UNROLL-NEXT:    [[WIDE_LOAD1]] = load <4 x i16>, <4 x i16>* [[TMP4]], align 2
-; UNROLL-NEXT:    [[TMP5:%.*]] = shufflevector <4 x i16> [[VECTOR_RECUR]], <4 x i16> [[WIDE_LOAD]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; UNROLL-NEXT:    [[TMP6:%.*]] = shufflevector <4 x i16> [[WIDE_LOAD]], <4 x i16> [[WIDE_LOAD1]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; UNROLL-NEXT:    [[TMP7:%.*]] = sext <4 x i16> [[TMP5]] to <4 x i32>
-; UNROLL-NEXT:    [[TMP8:%.*]] = sext <4 x i16> [[TMP6]] to <4 x i32>
-; UNROLL-NEXT:    [[TMP9:%.*]] = sext <4 x i16> [[WIDE_LOAD]] to <4 x i32>
-; UNROLL-NEXT:    [[TMP10:%.*]] = sext <4 x i16> [[WIDE_LOAD1]] to <4 x i32>
-; UNROLL-NEXT:    [[TMP11:%.*]] = mul nsw <4 x i32> [[TMP9]], [[TMP7]]
-; UNROLL-NEXT:    [[TMP12:%.*]] = mul nsw <4 x i32> [[TMP10]], [[TMP8]]
-; UNROLL-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i32, i32* [[B:%.*]], i64 [[INDEX]]
-; UNROLL-NEXT:    [[TMP14:%.*]] = bitcast i32* [[TMP13]] to <4 x i32>*
-; UNROLL-NEXT:    store <4 x i32> [[TMP11]], <4 x i32>* [[TMP14]], align 4
-; UNROLL-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i32, i32* [[TMP13]], i64 4
-; UNROLL-NEXT:    [[TMP16:%.*]] = bitcast i32* [[TMP15]] to <4 x i32>*
-; UNROLL-NEXT:    store <4 x i32> [[TMP12]], <4 x i32>* [[TMP16]], align 4
-; UNROLL-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
-; UNROLL-NEXT:    [[TMP17:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; UNROLL-NEXT:    br i1 [[TMP17]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
-; UNROLL:       middle.block:
-; UNROLL-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[N_VEC]], [[N]]
-; UNROLL-NEXT:    [[VECTOR_RECUR_EXTRACT:%.*]] = extractelement <4 x i16> [[WIDE_LOAD1]], i64 3
-; UNROLL-NEXT:    br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
-; UNROLL:       scalar.ph:
-; UNROLL-NEXT:    [[SCALAR_RECUR_INIT:%.*]] = phi i16 [ [[DOTPRE]], [[ENTRY:%.*]] ], [ [[VECTOR_RECUR_EXTRACT]], [[MIDDLE_BLOCK]] ]
-; UNROLL-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 0, [[ENTRY]] ], [ [[N_VEC]], [[MIDDLE_BLOCK]] ]
-; UNROLL-NEXT:    br label [[FOR_BODY:%.*]]
-; UNROLL:       for.body:
-; UNROLL-NEXT:    [[SCALAR_RECUR:%.*]] = phi i16 [ [[SCALAR_RECUR_INIT]], [[SCALAR_PH]] ], [ [[TMP18:%.*]], [[FOR_BODY]] ]
-; UNROLL-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; UNROLL-NEXT:    [[CONV:%.*]] = sext i16 [[SCALAR_RECUR]] to i32
-; UNROLL-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; UNROLL-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[A]], i64 [[INDVARS_IV_NEXT]]
-; UNROLL-NEXT:    [[TMP18]] = load i16, i16* [[ARRAYIDX2]], align 2
-; UNROLL-NEXT:    [[CONV3:%.*]] = sext i16 [[TMP18]] to i32
-; UNROLL-NEXT:    [[MUL:%.*]] = mul nsw i32 [[CONV3]], [[CONV]]
-; UNROLL-NEXT:    [[ARRAYIDX5:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 [[INDVARS_IV]]
-; UNROLL-NEXT:    store i32 [[MUL]], i32* [[ARRAYIDX5]], align 4
-; UNROLL-NEXT:    [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[N]]
-; UNROLL-NEXT:    br i1 [[EXITCOND]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP17:![0-9]+]]
-; UNROLL:       for.end:
-; UNROLL-NEXT:    ret void
-;
 ; UNROLL-NO-IC-LABEL: @sink_after(
 ; UNROLL-NO-IC-NEXT:  entry:
 ; UNROLL-NO-IC-NEXT:    [[DOTPRE:%.*]] = load i16, i16* [[A:%.*]], align 2
@@ -2862,164 +1982,6 @@ for.end:
 ; Check that the sext sank after the load in the vector loop.
 ;
 define void @PR34711([2 x i16]* noalias %a, i32* noalias %b, i32* noalias %c, i64 %n) {
-; CHECK-LABEL: @PR34711(
-; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[PRE_INDEX:%.*]] = getelementptr inbounds [2 x i16], [2 x i16]* [[A:%.*]], i64 0, i64 0
-; CHECK-NEXT:    [[DOTPRE:%.*]] = load i16, i16* [[PRE_INDEX]], align 2
-; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N:%.*]], 4
-; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; CHECK:       vector.ph:
-; CHECK-NEXT:    [[N_VEC:%.*]] = and i64 [[N]], -4
-; CHECK-NEXT:    [[VECTOR_RECUR_INIT:%.*]] = insertelement <4 x i16> poison, i16 [[DOTPRE]], i64 3
-; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
-; CHECK:       vector.body:
-; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VECTOR_RECUR:%.*]] = phi <4 x i16> [ [[VECTOR_RECUR_INIT]], [[VECTOR_PH]] ], [ [[TMP16:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = or i64 [[INDEX]], 1
-; CHECK-NEXT:    [[TMP1:%.*]] = or i64 [[INDEX]], 2
-; CHECK-NEXT:    [[TMP2:%.*]] = or i64 [[INDEX]], 3
-; CHECK-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i32, i32* [[C:%.*]], i64 [[INDEX]]
-; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [2 x i16], [2 x i16]* [[A]], i64 [[INDEX]], i64 1
-; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds [2 x i16], [2 x i16]* [[A]], i64 [[TMP0]], i64 1
-; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [2 x i16], [2 x i16]* [[A]], i64 [[TMP1]], i64 1
-; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds [2 x i16], [2 x i16]* [[A]], i64 [[TMP2]], i64 1
-; CHECK-NEXT:    [[TMP8:%.*]] = bitcast i32* [[TMP3]] to <4 x i32>*
-; CHECK-NEXT:    store <4 x i32> <i32 7, i32 7, i32 7, i32 7>, <4 x i32>* [[TMP8]], align 4
-; CHECK-NEXT:    [[TMP9:%.*]] = load i16, i16* [[TMP4]], align 2
-; CHECK-NEXT:    [[TMP10:%.*]] = load i16, i16* [[TMP5]], align 2
-; CHECK-NEXT:    [[TMP11:%.*]] = load i16, i16* [[TMP6]], align 2
-; CHECK-NEXT:    [[TMP12:%.*]] = load i16, i16* [[TMP7]], align 2
-; CHECK-NEXT:    [[TMP13:%.*]] = insertelement <4 x i16> poison, i16 [[TMP9]], i64 0
-; CHECK-NEXT:    [[TMP14:%.*]] = insertelement <4 x i16> [[TMP13]], i16 [[TMP10]], i64 1
-; CHECK-NEXT:    [[TMP15:%.*]] = insertelement <4 x i16> [[TMP14]], i16 [[TMP11]], i64 2
-; CHECK-NEXT:    [[TMP16]] = insertelement <4 x i16> [[TMP15]], i16 [[TMP12]], i64 3
-; CHECK-NEXT:    [[TMP17:%.*]] = shufflevector <4 x i16> [[VECTOR_RECUR]], <4 x i16> [[TMP15]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; CHECK-NEXT:    [[TMP18:%.*]] = sext <4 x i16> [[TMP17]] to <4 x i32>
-; CHECK-NEXT:    [[TMP19:%.*]] = sext <4 x i16> [[TMP16]] to <4 x i32>
-; CHECK-NEXT:    [[TMP20:%.*]] = mul nsw <4 x i32> [[TMP19]], [[TMP18]]
-; CHECK-NEXT:    [[TMP21:%.*]] = getelementptr inbounds i32, i32* [[B:%.*]], i64 [[INDEX]]
-; CHECK-NEXT:    [[TMP22:%.*]] = bitcast i32* [[TMP21]] to <4 x i32>*
-; CHECK-NEXT:    store <4 x i32> [[TMP20]], <4 x i32>* [[TMP22]], align 4
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-NEXT:    [[TMP23:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT:    br i1 [[TMP23]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[N_VEC]], [[N]]
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[SCALAR_RECUR_INIT:%.*]] = phi i16 [ [[DOTPRE]], [[ENTRY:%.*]] ], [ [[TMP12]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 0, [[ENTRY]] ], [ [[N_VEC]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
-; CHECK:       for.body:
-; CHECK-NEXT:    [[SCALAR_RECUR:%.*]] = phi i16 [ [[SCALAR_RECUR_INIT]], [[SCALAR_PH]] ], [ [[TMP24:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT:    [[ARRAYCIDX:%.*]] = getelementptr inbounds i32, i32* [[C]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    [[CUR_INDEX:%.*]] = getelementptr inbounds [2 x i16], [2 x i16]* [[A]], i64 [[INDVARS_IV]], i64 1
-; CHECK-NEXT:    store i32 7, i32* [[ARRAYCIDX]], align 4
-; CHECK-NEXT:    [[CONV:%.*]] = sext i16 [[SCALAR_RECUR]] to i32
-; CHECK-NEXT:    [[TMP24]] = load i16, i16* [[CUR_INDEX]], align 2
-; CHECK-NEXT:    [[CONV3:%.*]] = sext i16 [[TMP24]] to i32
-; CHECK-NEXT:    [[MUL:%.*]] = mul nsw i32 [[CONV3]], [[CONV]]
-; CHECK-NEXT:    [[ARRAYIDX5:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    store i32 [[MUL]], i32* [[ARRAYIDX5]], align 4
-; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[N]]
-; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP19:![0-9]+]]
-; CHECK:       for.end:
-; CHECK-NEXT:    ret void
-;
-; UNROLL-LABEL: @PR34711(
-; UNROLL-NEXT:  entry:
-; UNROLL-NEXT:    [[PRE_INDEX:%.*]] = getelementptr inbounds [2 x i16], [2 x i16]* [[A:%.*]], i64 0, i64 0
-; UNROLL-NEXT:    [[DOTPRE:%.*]] = load i16, i16* [[PRE_INDEX]], align 2
-; UNROLL-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N:%.*]], 8
-; UNROLL-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; UNROLL:       vector.ph:
-; UNROLL-NEXT:    [[N_VEC:%.*]] = and i64 [[N]], -8
-; UNROLL-NEXT:    [[VECTOR_RECUR_INIT:%.*]] = insertelement <4 x i16> poison, i16 [[DOTPRE]], i64 3
-; UNROLL-NEXT:    br label [[VECTOR_BODY:%.*]]
-; UNROLL:       vector.body:
-; UNROLL-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[VECTOR_RECUR:%.*]] = phi <4 x i16> [ [[VECTOR_RECUR_INIT]], [[VECTOR_PH]] ], [ [[TMP34:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[TMP0:%.*]] = or i64 [[INDEX]], 1
-; UNROLL-NEXT:    [[TMP1:%.*]] = or i64 [[INDEX]], 2
-; UNROLL-NEXT:    [[TMP2:%.*]] = or i64 [[INDEX]], 3
-; UNROLL-NEXT:    [[TMP3:%.*]] = or i64 [[INDEX]], 4
-; UNROLL-NEXT:    [[TMP4:%.*]] = or i64 [[INDEX]], 5
-; UNROLL-NEXT:    [[TMP5:%.*]] = or i64 [[INDEX]], 6
-; UNROLL-NEXT:    [[TMP6:%.*]] = or i64 [[INDEX]], 7
-; UNROLL-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i32, i32* [[C:%.*]], i64 [[INDEX]]
-; UNROLL-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [2 x i16], [2 x i16]* [[A]], i64 [[INDEX]], i64 1
-; UNROLL-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [2 x i16], [2 x i16]* [[A]], i64 [[TMP0]], i64 1
-; UNROLL-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [2 x i16], [2 x i16]* [[A]], i64 [[TMP1]], i64 1
-; UNROLL-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [2 x i16], [2 x i16]* [[A]], i64 [[TMP2]], i64 1
-; UNROLL-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [2 x i16], [2 x i16]* [[A]], i64 [[TMP3]], i64 1
-; UNROLL-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [2 x i16], [2 x i16]* [[A]], i64 [[TMP4]], i64 1
-; UNROLL-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [2 x i16], [2 x i16]* [[A]], i64 [[TMP5]], i64 1
-; UNROLL-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [2 x i16], [2 x i16]* [[A]], i64 [[TMP6]], i64 1
-; UNROLL-NEXT:    [[TMP16:%.*]] = bitcast i32* [[TMP7]] to <4 x i32>*
-; UNROLL-NEXT:    store <4 x i32> <i32 7, i32 7, i32 7, i32 7>, <4 x i32>* [[TMP16]], align 4
-; UNROLL-NEXT:    [[TMP17:%.*]] = getelementptr inbounds i32, i32* [[TMP7]], i64 4
-; UNROLL-NEXT:    [[TMP18:%.*]] = bitcast i32* [[TMP17]] to <4 x i32>*
-; UNROLL-NEXT:    store <4 x i32> <i32 7, i32 7, i32 7, i32 7>, <4 x i32>* [[TMP18]], align 4
-; UNROLL-NEXT:    [[TMP19:%.*]] = load i16, i16* [[TMP8]], align 2
-; UNROLL-NEXT:    [[TMP20:%.*]] = load i16, i16* [[TMP9]], align 2
-; UNROLL-NEXT:    [[TMP21:%.*]] = load i16, i16* [[TMP10]], align 2
-; UNROLL-NEXT:    [[TMP22:%.*]] = load i16, i16* [[TMP11]], align 2
-; UNROLL-NEXT:    [[TMP23:%.*]] = insertelement <4 x i16> poison, i16 [[TMP19]], i64 0
-; UNROLL-NEXT:    [[TMP24:%.*]] = insertelement <4 x i16> [[TMP23]], i16 [[TMP20]], i64 1
-; UNROLL-NEXT:    [[TMP25:%.*]] = insertelement <4 x i16> [[TMP24]], i16 [[TMP21]], i64 2
-; UNROLL-NEXT:    [[TMP26:%.*]] = insertelement <4 x i16> [[TMP25]], i16 [[TMP22]], i64 3
-; UNROLL-NEXT:    [[TMP27:%.*]] = load i16, i16* [[TMP12]], align 2
-; UNROLL-NEXT:    [[TMP28:%.*]] = load i16, i16* [[TMP13]], align 2
-; UNROLL-NEXT:    [[TMP29:%.*]] = load i16, i16* [[TMP14]], align 2
-; UNROLL-NEXT:    [[TMP30:%.*]] = load i16, i16* [[TMP15]], align 2
-; UNROLL-NEXT:    [[TMP31:%.*]] = insertelement <4 x i16> poison, i16 [[TMP27]], i64 0
-; UNROLL-NEXT:    [[TMP32:%.*]] = insertelement <4 x i16> [[TMP31]], i16 [[TMP28]], i64 1
-; UNROLL-NEXT:    [[TMP33:%.*]] = insertelement <4 x i16> [[TMP32]], i16 [[TMP29]], i64 2
-; UNROLL-NEXT:    [[TMP34]] = insertelement <4 x i16> [[TMP33]], i16 [[TMP30]], i64 3
-; UNROLL-NEXT:    [[TMP35:%.*]] = shufflevector <4 x i16> [[VECTOR_RECUR]], <4 x i16> [[TMP25]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; UNROLL-NEXT:    [[TMP36:%.*]] = shufflevector <4 x i16> [[TMP26]], <4 x i16> [[TMP33]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; UNROLL-NEXT:    [[TMP37:%.*]] = sext <4 x i16> [[TMP35]] to <4 x i32>
-; UNROLL-NEXT:    [[TMP38:%.*]] = sext <4 x i16> [[TMP36]] to <4 x i32>
-; UNROLL-NEXT:    [[TMP39:%.*]] = sext <4 x i16> [[TMP26]] to <4 x i32>
-; UNROLL-NEXT:    [[TMP40:%.*]] = sext <4 x i16> [[TMP34]] to <4 x i32>
-; UNROLL-NEXT:    [[TMP41:%.*]] = mul nsw <4 x i32> [[TMP39]], [[TMP37]]
-; UNROLL-NEXT:    [[TMP42:%.*]] = mul nsw <4 x i32> [[TMP40]], [[TMP38]]
-; UNROLL-NEXT:    [[TMP43:%.*]] = getelementptr inbounds i32, i32* [[B:%.*]], i64 [[INDEX]]
-; UNROLL-NEXT:    [[TMP44:%.*]] = bitcast i32* [[TMP43]] to <4 x i32>*
-; UNROLL-NEXT:    store <4 x i32> [[TMP41]], <4 x i32>* [[TMP44]], align 4
-; UNROLL-NEXT:    [[TMP45:%.*]] = getelementptr inbounds i32, i32* [[TMP43]], i64 4
-; UNROLL-NEXT:    [[TMP46:%.*]] = bitcast i32* [[TMP45]] to <4 x i32>*
-; UNROLL-NEXT:    store <4 x i32> [[TMP42]], <4 x i32>* [[TMP46]], align 4
-; UNROLL-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
-; UNROLL-NEXT:    [[TMP47:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; UNROLL-NEXT:    br i1 [[TMP47]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
-; UNROLL:       middle.block:
-; UNROLL-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[N_VEC]], [[N]]
-; UNROLL-NEXT:    br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
-; UNROLL:       scalar.ph:
-; UNROLL-NEXT:    [[SCALAR_RECUR_INIT:%.*]] = phi i16 [ [[DOTPRE]], [[ENTRY:%.*]] ], [ [[TMP30]], [[MIDDLE_BLOCK]] ]
-; UNROLL-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 0, [[ENTRY]] ], [ [[N_VEC]], [[MIDDLE_BLOCK]] ]
-; UNROLL-NEXT:    br label [[FOR_BODY:%.*]]
-; UNROLL:       for.body:
-; UNROLL-NEXT:    [[SCALAR_RECUR:%.*]] = phi i16 [ [[SCALAR_RECUR_INIT]], [[SCALAR_PH]] ], [ [[TMP48:%.*]], [[FOR_BODY]] ]
-; UNROLL-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; UNROLL-NEXT:    [[ARRAYCIDX:%.*]] = getelementptr inbounds i32, i32* [[C]], i64 [[INDVARS_IV]]
-; UNROLL-NEXT:    [[CUR_INDEX:%.*]] = getelementptr inbounds [2 x i16], [2 x i16]* [[A]], i64 [[INDVARS_IV]], i64 1
-; UNROLL-NEXT:    store i32 7, i32* [[ARRAYCIDX]], align 4
-; UNROLL-NEXT:    [[CONV:%.*]] = sext i16 [[SCALAR_RECUR]] to i32
-; UNROLL-NEXT:    [[TMP48]] = load i16, i16* [[CUR_INDEX]], align 2
-; UNROLL-NEXT:    [[CONV3:%.*]] = sext i16 [[TMP48]] to i32
-; UNROLL-NEXT:    [[MUL:%.*]] = mul nsw i32 [[CONV3]], [[CONV]]
-; UNROLL-NEXT:    [[ARRAYIDX5:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 [[INDVARS_IV]]
-; UNROLL-NEXT:    store i32 [[MUL]], i32* [[ARRAYIDX5]], align 4
-; UNROLL-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; UNROLL-NEXT:    [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[N]]
-; UNROLL-NEXT:    br i1 [[EXITCOND]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP19:![0-9]+]]
-; UNROLL:       for.end:
-; UNROLL-NEXT:    ret void
-;
 ; UNROLL-NO-IC-LABEL: @PR34711(
 ; UNROLL-NO-IC-NEXT:  entry:
 ; UNROLL-NO-IC-NEXT:    [[PRE_INDEX:%.*]] = getelementptr inbounds [2 x i16], [2 x i16]* [[A:%.*]], i64 0, i64 0
@@ -3287,121 +2249,6 @@ for.end:
 
 ;
 define void @sink_after_with_multiple_users(i16* noalias %a, i32* noalias %b, i64 %n) {
-; CHECK-LABEL: @sink_after_with_multiple_users(
-; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[DOTPRE:%.*]] = load i16, i16* [[A:%.*]], align 2
-; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N:%.*]], 4
-; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; CHECK:       vector.ph:
-; CHECK-NEXT:    [[N_VEC:%.*]] = and i64 [[N]], -4
-; CHECK-NEXT:    [[VECTOR_RECUR_INIT:%.*]] = insertelement <4 x i16> poison, i16 [[DOTPRE]], i64 3
-; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
-; CHECK:       vector.body:
-; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VECTOR_RECUR:%.*]] = phi <4 x i16> [ [[VECTOR_RECUR_INIT]], [[VECTOR_PH]] ], [ [[WIDE_LOAD:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = or i64 [[INDEX]], 1
-; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i16, i16* [[A]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP2:%.*]] = bitcast i16* [[TMP1]] to <4 x i16>*
-; CHECK-NEXT:    [[WIDE_LOAD]] = load <4 x i16>, <4 x i16>* [[TMP2]], align 2
-; CHECK-NEXT:    [[TMP3:%.*]] = shufflevector <4 x i16> [[VECTOR_RECUR]], <4 x i16> [[WIDE_LOAD]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; CHECK-NEXT:    [[TMP4:%.*]] = sext <4 x i16> [[TMP3]] to <4 x i32>
-; CHECK-NEXT:    [[TMP5:%.*]] = add nsw <4 x i32> [[TMP4]], <i32 2, i32 2, i32 2, i32 2>
-; CHECK-NEXT:    [[TMP6:%.*]] = sext <4 x i16> [[WIDE_LOAD]] to <4 x i32>
-; CHECK-NEXT:    [[TMP7:%.*]] = mul nsw <4 x i32> [[TMP5]], [[TMP6]]
-; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i32, i32* [[B:%.*]], i64 [[INDEX]]
-; CHECK-NEXT:    [[TMP9:%.*]] = bitcast i32* [[TMP8]] to <4 x i32>*
-; CHECK-NEXT:    store <4 x i32> [[TMP7]], <4 x i32>* [[TMP9]], align 4
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT:    br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP20:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[N_VEC]], [[N]]
-; CHECK-NEXT:    [[VECTOR_RECUR_EXTRACT:%.*]] = extractelement <4 x i16> [[WIDE_LOAD]], i64 3
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[SCALAR_RECUR_INIT:%.*]] = phi i16 [ [[DOTPRE]], [[ENTRY:%.*]] ], [ [[VECTOR_RECUR_EXTRACT]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 0, [[ENTRY]] ], [ [[N_VEC]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT:    br label [[FOR_BODY:%.*]]
-; CHECK:       for.body:
-; CHECK-NEXT:    [[SCALAR_RECUR:%.*]] = phi i16 [ [[SCALAR_RECUR_INIT]], [[SCALAR_PH]] ], [ [[TMP11:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; CHECK-NEXT:    [[CONV:%.*]] = sext i16 [[SCALAR_RECUR]] to i32
-; CHECK-NEXT:    [[ADD:%.*]] = add nsw i32 [[CONV]], 2
-; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; CHECK-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[A]], i64 [[INDVARS_IV_NEXT]]
-; CHECK-NEXT:    [[TMP11]] = load i16, i16* [[ARRAYIDX2]], align 2
-; CHECK-NEXT:    [[CONV3:%.*]] = sext i16 [[TMP11]] to i32
-; CHECK-NEXT:    [[MUL:%.*]] = mul nsw i32 [[ADD]], [[CONV3]]
-; CHECK-NEXT:    [[ARRAYIDX5:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 [[INDVARS_IV]]
-; CHECK-NEXT:    store i32 [[MUL]], i32* [[ARRAYIDX5]], align 4
-; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[N]]
-; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP21:![0-9]+]]
-; CHECK:       for.end:
-; CHECK-NEXT:    ret void
-;
-; UNROLL-LABEL: @sink_after_with_multiple_users(
-; UNROLL-NEXT:  entry:
-; UNROLL-NEXT:    [[DOTPRE:%.*]] = load i16, i16* [[A:%.*]], align 2
-; UNROLL-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[N:%.*]], 8
-; UNROLL-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; UNROLL:       vector.ph:
-; UNROLL-NEXT:    [[N_VEC:%.*]] = and i64 [[N]], -8
-; UNROLL-NEXT:    [[VECTOR_RECUR_INIT:%.*]] = insertelement <4 x i16> poison, i16 [[DOTPRE]], i64 3
-; UNROLL-NEXT:    br label [[VECTOR_BODY:%.*]]
-; UNROLL:       vector.body:
-; UNROLL-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[VECTOR_RECUR:%.*]] = phi <4 x i16> [ [[VECTOR_RECUR_INIT]], [[VECTOR_PH]] ], [ [[WIDE_LOAD1:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[TMP0:%.*]] = or i64 [[INDEX]], 1
-; UNROLL-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i16, i16* [[A]], i64 [[TMP0]]
-; UNROLL-NEXT:    [[TMP2:%.*]] = bitcast i16* [[TMP1]] to <4 x i16>*
-; UNROLL-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i16>, <4 x i16>* [[TMP2]], align 2
-; UNROLL-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i16, i16* [[TMP1]], i64 4
-; UNROLL-NEXT:    [[TMP4:%.*]] = bitcast i16* [[TMP3]] to <4 x i16>*
-; UNROLL-NEXT:    [[WIDE_LOAD1]] = load <4 x i16>, <4 x i16>* [[TMP4]], align 2
-; UNROLL-NEXT:    [[TMP5:%.*]] = shufflevector <4 x i16> [[VECTOR_RECUR]], <4 x i16> [[WIDE_LOAD]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; UNROLL-NEXT:    [[TMP6:%.*]] = shufflevector <4 x i16> [[WIDE_LOAD]], <4 x i16> [[WIDE_LOAD1]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; UNROLL-NEXT:    [[TMP7:%.*]] = sext <4 x i16> [[TMP5]] to <4 x i32>
-; UNROLL-NEXT:    [[TMP8:%.*]] = sext <4 x i16> [[TMP6]] to <4 x i32>
-; UNROLL-NEXT:    [[TMP9:%.*]] = add nsw <4 x i32> [[TMP7]], <i32 2, i32 2, i32 2, i32 2>
-; UNROLL-NEXT:    [[TMP10:%.*]] = add nsw <4 x i32> [[TMP8]], <i32 2, i32 2, i32 2, i32 2>
-; UNROLL-NEXT:    [[TMP11:%.*]] = sext <4 x i16> [[WIDE_LOAD]] to <4 x i32>
-; UNROLL-NEXT:    [[TMP12:%.*]] = sext <4 x i16> [[WIDE_LOAD1]] to <4 x i32>
-; UNROLL-NEXT:    [[TMP13:%.*]] = mul nsw <4 x i32> [[TMP9]], [[TMP11]]
-; UNROLL-NEXT:    [[TMP14:%.*]] = mul nsw <4 x i32> [[TMP10]], [[TMP12]]
-; UNROLL-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i32, i32* [[B:%.*]], i64 [[INDEX]]
-; UNROLL-NEXT:    [[TMP16:%.*]] = bitcast i32* [[TMP15]] to <4 x i32>*
-; UNROLL-NEXT:    store <4 x i32> [[TMP13]], <4 x i32>* [[TMP16]], align 4
-; UNROLL-NEXT:    [[TMP17:%.*]] = getelementptr inbounds i32, i32* [[TMP15]], i64 4
-; UNROLL-NEXT:    [[TMP18:%.*]] = bitcast i32* [[TMP17]] to <4 x i32>*
-; UNROLL-NEXT:    store <4 x i32> [[TMP14]], <4 x i32>* [[TMP18]], align 4
-; UNROLL-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
-; UNROLL-NEXT:    [[TMP19:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; UNROLL-NEXT:    br i1 [[TMP19]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP20:![0-9]+]]
-; UNROLL:       middle.block:
-; UNROLL-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[N_VEC]], [[N]]
-; UNROLL-NEXT:    [[VECTOR_RECUR_EXTRACT:%.*]] = extractelement <4 x i16> [[WIDE_LOAD1]], i64 3
-; UNROLL-NEXT:    br i1 [[CMP_N]], label [[FOR_END:%.*]], label [[SCALAR_PH]]
-; UNROLL:       scalar.ph:
-; UNROLL-NEXT:    [[SCALAR_RECUR_INIT:%.*]] = phi i16 [ [[DOTPRE]], [[ENTRY:%.*]] ], [ [[VECTOR_RECUR_EXTRACT]], [[MIDDLE_BLOCK]] ]
-; UNROLL-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 0, [[ENTRY]] ], [ [[N_VEC]], [[MIDDLE_BLOCK]] ]
-; UNROLL-NEXT:    br label [[FOR_BODY:%.*]]
-; UNROLL:       for.body:
-; UNROLL-NEXT:    [[SCALAR_RECUR:%.*]] = phi i16 [ [[SCALAR_RECUR_INIT]], [[SCALAR_PH]] ], [ [[TMP20:%.*]], [[FOR_BODY]] ]
-; UNROLL-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
-; UNROLL-NEXT:    [[CONV:%.*]] = sext i16 [[SCALAR_RECUR]] to i32
-; UNROLL-NEXT:    [[ADD:%.*]] = add nsw i32 [[CONV]], 2
-; UNROLL-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
-; UNROLL-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds i16, i16* [[A]], i64 [[INDVARS_IV_NEXT]]
-; UNROLL-NEXT:    [[TMP20]] = load i16, i16* [[ARRAYIDX2]], align 2
-; UNROLL-NEXT:    [[CONV3:%.*]] = sext i16 [[TMP20]] to i32
-; UNROLL-NEXT:    [[MUL:%.*]] = mul nsw i32 [[ADD]], [[CONV3]]
-; UNROLL-NEXT:    [[ARRAYIDX5:%.*]] = getelementptr inbounds i32, i32* [[B]], i64 [[INDVARS_IV]]
-; UNROLL-NEXT:    store i32 [[MUL]], i32* [[ARRAYIDX5]], align 4
-; UNROLL-NEXT:    [[EXITCOND:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[N]]
-; UNROLL-NEXT:    br i1 [[EXITCOND]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP21:![0-9]+]]
-; UNROLL:       for.end:
-; UNROLL-NEXT:    ret void
-;
 ; UNROLL-NO-IC-LABEL: @sink_after_with_multiple_users(
 ; UNROLL-NO-IC-NEXT:  entry:
 ; UNROLL-NO-IC-NEXT:    [[DOTPRE:%.*]] = load i16, i16* [[A:%.*]], align 2
@@ -3618,56 +2465,6 @@ for.end:
 ; branches will become users of the condition bit after vectorization and would
 ; need to be sunk if the loop is vectorized.
 define void @do_not_sink_branch(i32 %x, i32* %in, i32* %out, i32 %tc) local_unnamed_addr #0 {
-; CHECK-LABEL: @do_not_sink_branch(
-; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[CMP530:%.*]] = icmp sgt i32 [[TC:%.*]], 0
-; CHECK-NEXT:    br label [[FOR_BODY4:%.*]]
-; CHECK:       for.body4:
-; CHECK-NEXT:    [[INDVARS_IV:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[COND_END:%.*]] ]
-; CHECK-NEXT:    [[CMP534:%.*]] = phi i1 [ [[CMP530]], [[ENTRY]] ], [ [[CMP5:%.*]], [[COND_END]] ]
-; CHECK-NEXT:    br i1 [[CMP534]], label [[COND_TRUE:%.*]], label [[COND_END]]
-; CHECK:       cond.true:
-; CHECK-NEXT:    [[TMP0:%.*]] = zext i32 [[INDVARS_IV]] to i64
-; CHECK-NEXT:    [[ARRAYIDX7:%.*]] = getelementptr inbounds i32, i32* [[IN:%.*]], i64 [[TMP0]]
-; CHECK-NEXT:    [[IN_VAL:%.*]] = load i32, i32* [[ARRAYIDX7]], align 4
-; CHECK-NEXT:    br label [[COND_END]]
-; CHECK:       cond.end:
-; CHECK-NEXT:    [[COND:%.*]] = phi i32 [ [[IN_VAL]], [[COND_TRUE]] ], [ 0, [[FOR_BODY4]] ]
-; CHECK-NEXT:    [[TMP1:%.*]] = zext i32 [[INDVARS_IV]] to i64
-; CHECK-NEXT:    [[ARRAYIDX8:%.*]] = getelementptr inbounds i32, i32* [[OUT:%.*]], i64 [[TMP1]]
-; CHECK-NEXT:    store i32 [[COND]], i32* [[ARRAYIDX8]], align 4
-; CHECK-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i32 [[INDVARS_IV]], 1
-; CHECK-NEXT:    [[CMP5]] = icmp slt i32 [[INDVARS_IV_NEXT]], [[TC]]
-; CHECK-NEXT:    [[EXITCOND:%.*]] = icmp eq i32 [[INDVARS_IV_NEXT]], [[X:%.*]]
-; CHECK-NEXT:    br i1 [[EXITCOND]], label [[FOR_END12_LOOPEXIT:%.*]], label [[FOR_BODY4]]
-; CHECK:       for.end12.loopexit:
-; CHECK-NEXT:    ret void
-;
-; UNROLL-LABEL: @do_not_sink_branch(
-; UNROLL-NEXT:  entry:
-; UNROLL-NEXT:    [[CMP530:%.*]] = icmp sgt i32 [[TC:%.*]], 0
-; UNROLL-NEXT:    br label [[FOR_BODY4:%.*]]
-; UNROLL:       for.body4:
-; UNROLL-NEXT:    [[INDVARS_IV:%.*]] = phi i32 [ 0, [[ENTRY:%.*]] ], [ [[INDVARS_IV_NEXT:%.*]], [[COND_END:%.*]] ]
-; UNROLL-NEXT:    [[CMP534:%.*]] = phi i1 [ [[CMP530]], [[ENTRY]] ], [ [[CMP5:%.*]], [[COND_END]] ]
-; UNROLL-NEXT:    br i1 [[CMP534]], label [[COND_TRUE:%.*]], label [[COND_END]]
-; UNROLL:       cond.true:
-; UNROLL-NEXT:    [[TMP0:%.*]] = zext i32 [[INDVARS_IV]] to i64
-; UNROLL-NEXT:    [[ARRAYIDX7:%.*]] = getelementptr inbounds i32, i32* [[IN:%.*]], i64 [[TMP0]]
-; UNROLL-NEXT:    [[IN_VAL:%.*]] = load i32, i32* [[ARRAYIDX7]], align 4
-; UNROLL-NEXT:    br label [[COND_END]]
-; UNROLL:       cond.end:
-; UNROLL-NEXT:    [[COND:%.*]] = phi i32 [ [[IN_VAL]], [[COND_TRUE]] ], [ 0, [[FOR_BODY4]] ]
-; UNROLL-NEXT:    [[TMP1:%.*]] = zext i32 [[INDVARS_IV]] to i64
-; UNROLL-NEXT:    [[ARRAYIDX8:%.*]] = getelementptr inbounds i32, i32* [[OUT:%.*]], i64 [[TMP1]]
-; UNROLL-NEXT:    store i32 [[COND]], i32* [[ARRAYIDX8]], align 4
-; UNROLL-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i32 [[INDVARS_IV]], 1
-; UNROLL-NEXT:    [[CMP5]] = icmp slt i32 [[INDVARS_IV_NEXT]], [[TC]]
-; UNROLL-NEXT:    [[EXITCOND:%.*]] = icmp eq i32 [[INDVARS_IV_NEXT]], [[X:%.*]]
-; UNROLL-NEXT:    br i1 [[EXITCOND]], label [[FOR_END12_LOOPEXIT:%.*]], label [[FOR_BODY4]]
-; UNROLL:       for.end12.loopexit:
-; UNROLL-NEXT:    ret void
-;
 ; UNROLL-NO-IC-LABEL: @do_not_sink_branch(
 ; UNROLL-NO-IC-NEXT:  entry:
 ; UNROLL-NO-IC-NEXT:    [[CMP530:%.*]] = icmp slt i32 0, [[TC:%.*]]
@@ -3767,67 +2564,6 @@ for.end12.loopexit:                               ; preds = %cond.end
 ; Dead instructions, like the exit condition are not part of the actual VPlan
 ; and do not need to be sunk. PR44634.
 define void @sink_dead_inst() {
-;
-; CHECK-LABEL: @sink_dead_inst(
-; CHECK-NEXT:  entry:
-; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; CHECK:       vector.ph:
-; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
-; CHECK:       vector.body:
-; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = phi i16 [ -24, [[VECTOR_PH]] ], [ [[TMP1:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
-; CHECK-NEXT:    [[TMP1]] = add i16 [[TMP0]], 4
-; CHECK-NEXT:    [[TMP2:%.*]] = icmp eq i32 [[INDEX_NEXT]], 40
-; CHECK-NEXT:    br i1 [[TMP2]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP22:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[TMP3:%.*]] = or i16 [[TMP0]], 1
-; CHECK-NEXT:    [[VECTOR_RECUR_EXTRACT2:%.*]] = zext i16 [[TMP3]] to i32
-; CHECK-NEXT:    br i1 false, label [[FOR_END:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[SCALAR_RECUR_INIT4:%.*]] = phi i32 [ -27, [[ENTRY:%.*]] ], [ [[VECTOR_RECUR_EXTRACT2]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i16 [ -27, [[ENTRY]] ], [ 13, [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT:    br label [[FOR_COND:%.*]]
-; CHECK:       for.cond:
-; CHECK-NEXT:    [[IV:%.*]] = phi i16 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_COND]] ]
-; CHECK-NEXT:    [[SCALAR_RECUR5:%.*]] = phi i32 [ [[SCALAR_RECUR_INIT4]], [[SCALAR_PH]] ], [ [[REC_2_PREV:%.*]], [[FOR_COND]] ]
-; CHECK-NEXT:    [[CMP:%.*]] = icmp eq i32 [[SCALAR_RECUR5]], 15
-; CHECK-NEXT:    [[IV_NEXT]] = add i16 [[IV]], 1
-; CHECK-NEXT:    [[REC_2_PREV]] = zext i16 [[IV_NEXT]] to i32
-; CHECK-NEXT:    br i1 [[CMP]], label [[FOR_END]], label [[FOR_COND]], !llvm.loop [[LOOP23:![0-9]+]]
-; CHECK:       for.end:
-; CHECK-NEXT:    ret void
-;
-; UNROLL-LABEL: @sink_dead_inst(
-; UNROLL-NEXT:  entry:
-; UNROLL-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; UNROLL:       vector.ph:
-; UNROLL-NEXT:    br label [[VECTOR_BODY:%.*]]
-; UNROLL:       vector.body:
-; UNROLL-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[TMP0:%.*]] = phi i16 [ -24, [[VECTOR_PH]] ], [ [[TMP1:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 8
-; UNROLL-NEXT:    [[TMP1]] = add i16 [[TMP0]], 8
-; UNROLL-NEXT:    [[TMP2:%.*]] = icmp eq i32 [[INDEX_NEXT]], 40
-; UNROLL-NEXT:    br i1 [[TMP2]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP22:![0-9]+]]
-; UNROLL:       middle.block:
-; UNROLL-NEXT:    [[TMP3:%.*]] = or i16 [[TMP0]], 5
-; UNROLL-NEXT:    [[VECTOR_RECUR_EXTRACT3:%.*]] = zext i16 [[TMP3]] to i32
-; UNROLL-NEXT:    br i1 false, label [[FOR_END:%.*]], label [[SCALAR_PH]]
-; UNROLL:       scalar.ph:
-; UNROLL-NEXT:    [[SCALAR_RECUR_INIT5:%.*]] = phi i32 [ -27, [[ENTRY:%.*]] ], [ [[VECTOR_RECUR_EXTRACT3]], [[MIDDLE_BLOCK]] ]
-; UNROLL-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i16 [ -27, [[ENTRY]] ], [ 13, [[MIDDLE_BLOCK]] ]
-; UNROLL-NEXT:    br label [[FOR_COND:%.*]]
-; UNROLL:       for.cond:
-; UNROLL-NEXT:    [[IV:%.*]] = phi i16 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_COND]] ]
-; UNROLL-NEXT:    [[SCALAR_RECUR6:%.*]] = phi i32 [ [[SCALAR_RECUR_INIT5]], [[SCALAR_PH]] ], [ [[REC_2_PREV:%.*]], [[FOR_COND]] ]
-; UNROLL-NEXT:    [[CMP:%.*]] = icmp eq i32 [[SCALAR_RECUR6]], 15
-; UNROLL-NEXT:    [[IV_NEXT]] = add i16 [[IV]], 1
-; UNROLL-NEXT:    [[REC_2_PREV]] = zext i16 [[IV_NEXT]] to i32
-; UNROLL-NEXT:    br i1 [[CMP]], label [[FOR_END]], label [[FOR_COND]], !llvm.loop [[LOOP23:![0-9]+]]
-; UNROLL:       for.end:
-; UNROLL-NEXT:    ret void
-;
 ; UNROLL-NO-IC-LABEL: @sink_dead_inst(
 ; UNROLL-NO-IC-NEXT:  entry:
 ; UNROLL-NO-IC-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
@@ -3984,194 +2720,6 @@ for.end:
 }
 
 define i32 @sink_into_replication_region(i32 %y) {
-;
-; CHECK-LABEL: @sink_into_replication_region(
-; CHECK-NEXT:  bb:
-; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; CHECK:       vector.ph:
-; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @llvm.smax.i32(i32 [[Y:%.*]], i32 1)
-; CHECK-NEXT:    [[N_RND_UP:%.*]] = add nuw i32 [[TMP0]], 3
-; CHECK-NEXT:    [[N_VEC:%.*]] = and i32 [[N_RND_UP]], -4
-; CHECK-NEXT:    [[TRIP_COUNT_MINUS_1:%.*]] = add nsw i32 [[TMP0]], -1
-; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[TRIP_COUNT_MINUS_1]], i64 0
-; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
-; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
-; CHECK:       vector.body:
-; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_UDIV_CONTINUE8:%.*]] ]
-; CHECK-NEXT:    [[VECTOR_RECUR:%.*]] = phi <4 x i32> [ <i32 poison, i32 poison, i32 poison, i32 0>, [[VECTOR_PH]] ], [ [[TMP20:%.*]], [[PRED_UDIV_CONTINUE8]] ]
-; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP22:%.*]], [[PRED_UDIV_CONTINUE8]] ]
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = sub i32 [[Y]], [[INDEX]]
-; CHECK-NEXT:    [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <4 x i32> poison, i32 [[INDEX]], i64 0
-; CHECK-NEXT:    [[BROADCAST_SPLAT2:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT1]], <4 x i32> poison, <4 x i32> zeroinitializer
-; CHECK-NEXT:    [[VEC_IV:%.*]] = or <4 x i32> [[BROADCAST_SPLAT2]], <i32 0, i32 1, i32 2, i32 3>
-; CHECK-NEXT:    [[TMP1:%.*]] = icmp ule <4 x i32> [[VEC_IV]], [[BROADCAST_SPLAT]]
-; CHECK-NEXT:    [[TMP2:%.*]] = extractelement <4 x i1> [[TMP1]], i64 0
-; CHECK-NEXT:    br i1 [[TMP2]], label [[PRED_UDIV_IF:%.*]], label [[PRED_UDIV_CONTINUE:%.*]]
-; CHECK:       pred.udiv.if:
-; CHECK-NEXT:    [[TMP3:%.*]] = udiv i32 219220132, [[OFFSET_IDX]]
-; CHECK-NEXT:    [[TMP4:%.*]] = insertelement <4 x i32> poison, i32 [[TMP3]], i64 0
-; CHECK-NEXT:    br label [[PRED_UDIV_CONTINUE]]
-; CHECK:       pred.udiv.continue:
-; CHECK-NEXT:    [[TMP5:%.*]] = phi <4 x i32> [ poison, [[VECTOR_BODY]] ], [ [[TMP4]], [[PRED_UDIV_IF]] ]
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <4 x i1> [[TMP1]], i64 1
-; CHECK-NEXT:    br i1 [[TMP6]], label [[PRED_UDIV_IF3:%.*]], label [[PRED_UDIV_CONTINUE4:%.*]]
-; CHECK:       pred.udiv.if3:
-; CHECK-NEXT:    [[TMP7:%.*]] = add i32 [[OFFSET_IDX]], -1
-; CHECK-NEXT:    [[TMP8:%.*]] = udiv i32 219220132, [[TMP7]]
-; CHECK-NEXT:    [[TMP9:%.*]] = insertelement <4 x i32> [[TMP5]], i32 [[TMP8]], i64 1
-; CHECK-NEXT:    br label [[PRED_UDIV_CONTINUE4]]
-; CHECK:       pred.udiv.continue4:
-; CHECK-NEXT:    [[TMP10:%.*]] = phi <4 x i32> [ [[TMP5]], [[PRED_UDIV_CONTINUE]] ], [ [[TMP9]], [[PRED_UDIV_IF3]] ]
-; CHECK-NEXT:    [[TMP11:%.*]] = extractelement <4 x i1> [[TMP1]], i64 2
-; CHECK-NEXT:    br i1 [[TMP11]], label [[PRED_UDIV_IF5:%.*]], label [[PRED_UDIV_CONTINUE6:%.*]]
-; CHECK:       pred.udiv.if5:
-; CHECK-NEXT:    [[TMP12:%.*]] = add i32 [[OFFSET_IDX]], -2
-; CHECK-NEXT:    [[TMP13:%.*]] = udiv i32 219220132, [[TMP12]]
-; CHECK-NEXT:    [[TMP14:%.*]] = insertelement <4 x i32> [[TMP10]], i32 [[TMP13]], i64 2
-; CHECK-NEXT:    br label [[PRED_UDIV_CONTINUE6]]
-; CHECK:       pred.udiv.continue6:
-; CHECK-NEXT:    [[TMP15:%.*]] = phi <4 x i32> [ [[TMP10]], [[PRED_UDIV_CONTINUE4]] ], [ [[TMP14]], [[PRED_UDIV_IF5]] ]
-; CHECK-NEXT:    [[TMP16:%.*]] = extractelement <4 x i1> [[TMP1]], i64 3
-; CHECK-NEXT:    br i1 [[TMP16]], label [[PRED_UDIV_IF7:%.*]], label [[PRED_UDIV_CONTINUE8]]
-; CHECK:       pred.udiv.if7:
-; CHECK-NEXT:    [[TMP17:%.*]] = add i32 [[OFFSET_IDX]], -3
-; CHECK-NEXT:    [[TMP18:%.*]] = udiv i32 219220132, [[TMP17]]
-; CHECK-NEXT:    [[TMP19:%.*]] = insertelement <4 x i32> [[TMP15]], i32 [[TMP18]], i64 3
-; CHECK-NEXT:    br label [[PRED_UDIV_CONTINUE8]]
-; CHECK:       pred.udiv.continue8:
-; CHECK-NEXT:    [[TMP20]] = phi <4 x i32> [ [[TMP15]], [[PRED_UDIV_CONTINUE6]] ], [ [[TMP19]], [[PRED_UDIV_IF7]] ]
-; CHECK-NEXT:    [[TMP21:%.*]] = shufflevector <4 x i32> [[VECTOR_RECUR]], <4 x i32> [[TMP20]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; CHECK-NEXT:    [[TMP22]] = add <4 x i32> [[VEC_PHI]], [[TMP21]]
-; CHECK-NEXT:    [[INDEX_NEXT]] = add i32 [[INDEX]], 4
-; CHECK-NEXT:    [[TMP23:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT:    br i1 [[TMP23]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !prof [[PROF24:![0-9]+]], !llvm.loop [[LOOP25:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[TMP24:%.*]] = select <4 x i1> [[TMP1]], <4 x i32> [[TMP22]], <4 x i32> [[VEC_PHI]]
-; CHECK-NEXT:    [[TMP25:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP24]])
-; CHECK-NEXT:    br i1 true, label [[BB1:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    br label [[BB2:%.*]]
-; CHECK:       bb1:
-; CHECK-NEXT:    [[VAR:%.*]] = phi i32 [ undef, [[BB2]] ], [ [[TMP25]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT:    ret i32 [[VAR]]
-; CHECK:       bb2:
-; CHECK-NEXT:    br i1 undef, label [[BB1]], label [[BB2]], !prof [[PROF26:![0-9]+]], !llvm.loop [[LOOP27:![0-9]+]]
-;
-; UNROLL-LABEL: @sink_into_replication_region(
-; UNROLL-NEXT:  bb:
-; UNROLL-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; UNROLL:       vector.ph:
-; UNROLL-NEXT:    [[TMP0:%.*]] = call i32 @llvm.smax.i32(i32 [[Y:%.*]], i32 1)
-; UNROLL-NEXT:    [[N_RND_UP:%.*]] = add nuw i32 [[TMP0]], 7
-; UNROLL-NEXT:    [[N_VEC:%.*]] = and i32 [[N_RND_UP]], -8
-; UNROLL-NEXT:    [[TRIP_COUNT_MINUS_1:%.*]] = add nsw i32 [[TMP0]], -1
-; UNROLL-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[TRIP_COUNT_MINUS_1]], i64 0
-; UNROLL-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
-; UNROLL-NEXT:    br label [[VECTOR_BODY:%.*]]
-; UNROLL:       vector.body:
-; UNROLL-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_UDIV_CONTINUE18:%.*]] ]
-; UNROLL-NEXT:    [[VECTOR_RECUR:%.*]] = phi <4 x i32> [ <i32 poison, i32 poison, i32 poison, i32 0>, [[VECTOR_PH]] ], [ [[TMP41:%.*]], [[PRED_UDIV_CONTINUE18]] ]
-; UNROLL-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP44:%.*]], [[PRED_UDIV_CONTINUE18]] ]
-; UNROLL-NEXT:    [[VEC_PHI1:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP45:%.*]], [[PRED_UDIV_CONTINUE18]] ]
-; UNROLL-NEXT:    [[OFFSET_IDX:%.*]] = sub i32 [[Y]], [[INDEX]]
-; UNROLL-NEXT:    [[BROADCAST_SPLATINSERT2:%.*]] = insertelement <4 x i32> poison, i32 [[INDEX]], i64 0
-; UNROLL-NEXT:    [[BROADCAST_SPLAT3:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT2]], <4 x i32> poison, <4 x i32> zeroinitializer
-; UNROLL-NEXT:    [[VEC_IV:%.*]] = or <4 x i32> [[BROADCAST_SPLAT3]], <i32 0, i32 1, i32 2, i32 3>
-; UNROLL-NEXT:    [[VEC_IV4:%.*]] = or <4 x i32> [[BROADCAST_SPLAT3]], <i32 4, i32 5, i32 6, i32 7>
-; UNROLL-NEXT:    [[TMP1:%.*]] = icmp ule <4 x i32> [[VEC_IV]], [[BROADCAST_SPLAT]]
-; UNROLL-NEXT:    [[TMP2:%.*]] = icmp ule <4 x i32> [[VEC_IV4]], [[BROADCAST_SPLAT]]
-; UNROLL-NEXT:    [[TMP3:%.*]] = extractelement <4 x i1> [[TMP1]], i64 0
-; UNROLL-NEXT:    br i1 [[TMP3]], label [[PRED_UDIV_IF:%.*]], label [[PRED_UDIV_CONTINUE:%.*]]
-; UNROLL:       pred.udiv.if:
-; UNROLL-NEXT:    [[TMP4:%.*]] = udiv i32 219220132, [[OFFSET_IDX]]
-; UNROLL-NEXT:    [[TMP5:%.*]] = insertelement <4 x i32> poison, i32 [[TMP4]], i64 0
-; UNROLL-NEXT:    br label [[PRED_UDIV_CONTINUE]]
-; UNROLL:       pred.udiv.continue:
-; UNROLL-NEXT:    [[TMP6:%.*]] = phi <4 x i32> [ poison, [[VECTOR_BODY]] ], [ [[TMP5]], [[PRED_UDIV_IF]] ]
-; UNROLL-NEXT:    [[TMP7:%.*]] = extractelement <4 x i1> [[TMP1]], i64 1
-; UNROLL-NEXT:    br i1 [[TMP7]], label [[PRED_UDIV_IF5:%.*]], label [[PRED_UDIV_CONTINUE6:%.*]]
-; UNROLL:       pred.udiv.if5:
-; UNROLL-NEXT:    [[TMP8:%.*]] = add i32 [[OFFSET_IDX]], -1
-; UNROLL-NEXT:    [[TMP9:%.*]] = udiv i32 219220132, [[TMP8]]
-; UNROLL-NEXT:    [[TMP10:%.*]] = insertelement <4 x i32> [[TMP6]], i32 [[TMP9]], i64 1
-; UNROLL-NEXT:    br label [[PRED_UDIV_CONTINUE6]]
-; UNROLL:       pred.udiv.continue6:
-; UNROLL-NEXT:    [[TMP11:%.*]] = phi <4 x i32> [ [[TMP6]], [[PRED_UDIV_CONTINUE]] ], [ [[TMP10]], [[PRED_UDIV_IF5]] ]
-; UNROLL-NEXT:    [[TMP12:%.*]] = extractelement <4 x i1> [[TMP1]], i64 2
-; UNROLL-NEXT:    br i1 [[TMP12]], label [[PRED_UDIV_IF7:%.*]], label [[PRED_UDIV_CONTINUE8:%.*]]
-; UNROLL:       pred.udiv.if7:
-; UNROLL-NEXT:    [[TMP13:%.*]] = add i32 [[OFFSET_IDX]], -2
-; UNROLL-NEXT:    [[TMP14:%.*]] = udiv i32 219220132, [[TMP13]]
-; UNROLL-NEXT:    [[TMP15:%.*]] = insertelement <4 x i32> [[TMP11]], i32 [[TMP14]], i64 2
-; UNROLL-NEXT:    br label [[PRED_UDIV_CONTINUE8]]
-; UNROLL:       pred.udiv.continue8:
-; UNROLL-NEXT:    [[TMP16:%.*]] = phi <4 x i32> [ [[TMP11]], [[PRED_UDIV_CONTINUE6]] ], [ [[TMP15]], [[PRED_UDIV_IF7]] ]
-; UNROLL-NEXT:    [[TMP17:%.*]] = extractelement <4 x i1> [[TMP1]], i64 3
-; UNROLL-NEXT:    br i1 [[TMP17]], label [[PRED_UDIV_IF9:%.*]], label [[PRED_UDIV_CONTINUE10:%.*]]
-; UNROLL:       pred.udiv.if9:
-; UNROLL-NEXT:    [[TMP18:%.*]] = add i32 [[OFFSET_IDX]], -3
-; UNROLL-NEXT:    [[TMP19:%.*]] = udiv i32 219220132, [[TMP18]]
-; UNROLL-NEXT:    [[TMP20:%.*]] = insertelement <4 x i32> [[TMP16]], i32 [[TMP19]], i64 3
-; UNROLL-NEXT:    br label [[PRED_UDIV_CONTINUE10]]
-; UNROLL:       pred.udiv.continue10:
-; UNROLL-NEXT:    [[TMP21:%.*]] = phi <4 x i32> [ [[TMP16]], [[PRED_UDIV_CONTINUE8]] ], [ [[TMP20]], [[PRED_UDIV_IF9]] ]
-; UNROLL-NEXT:    [[TMP22:%.*]] = extractelement <4 x i1> [[TMP2]], i64 0
-; UNROLL-NEXT:    br i1 [[TMP22]], label [[PRED_UDIV_IF11:%.*]], label [[PRED_UDIV_CONTINUE12:%.*]]
-; UNROLL:       pred.udiv.if11:
-; UNROLL-NEXT:    [[TMP23:%.*]] = add i32 [[OFFSET_IDX]], -4
-; UNROLL-NEXT:    [[TMP24:%.*]] = udiv i32 219220132, [[TMP23]]
-; UNROLL-NEXT:    [[TMP25:%.*]] = insertelement <4 x i32> poison, i32 [[TMP24]], i64 0
-; UNROLL-NEXT:    br label [[PRED_UDIV_CONTINUE12]]
-; UNROLL:       pred.udiv.continue12:
-; UNROLL-NEXT:    [[TMP26:%.*]] = phi <4 x i32> [ poison, [[PRED_UDIV_CONTINUE10]] ], [ [[TMP25]], [[PRED_UDIV_IF11]] ]
-; UNROLL-NEXT:    [[TMP27:%.*]] = extractelement <4 x i1> [[TMP2]], i64 1
-; UNROLL-NEXT:    br i1 [[TMP27]], label [[PRED_UDIV_IF13:%.*]], label [[PRED_UDIV_CONTINUE14:%.*]]
-; UNROLL:       pred.udiv.if13:
-; UNROLL-NEXT:    [[TMP28:%.*]] = add i32 [[OFFSET_IDX]], -5
-; UNROLL-NEXT:    [[TMP29:%.*]] = udiv i32 219220132, [[TMP28]]
-; UNROLL-NEXT:    [[TMP30:%.*]] = insertelement <4 x i32> [[TMP26]], i32 [[TMP29]], i64 1
-; UNROLL-NEXT:    br label [[PRED_UDIV_CONTINUE14]]
-; UNROLL:       pred.udiv.continue14:
-; UNROLL-NEXT:    [[TMP31:%.*]] = phi <4 x i32> [ [[TMP26]], [[PRED_UDIV_CONTINUE12]] ], [ [[TMP30]], [[PRED_UDIV_IF13]] ]
-; UNROLL-NEXT:    [[TMP32:%.*]] = extractelement <4 x i1> [[TMP2]], i64 2
-; UNROLL-NEXT:    br i1 [[TMP32]], label [[PRED_UDIV_IF15:%.*]], label [[PRED_UDIV_CONTINUE16:%.*]]
-; UNROLL:       pred.udiv.if15:
-; UNROLL-NEXT:    [[TMP33:%.*]] = add i32 [[OFFSET_IDX]], -6
-; UNROLL-NEXT:    [[TMP34:%.*]] = udiv i32 219220132, [[TMP33]]
-; UNROLL-NEXT:    [[TMP35:%.*]] = insertelement <4 x i32> [[TMP31]], i32 [[TMP34]], i64 2
-; UNROLL-NEXT:    br label [[PRED_UDIV_CONTINUE16]]
-; UNROLL:       pred.udiv.continue16:
-; UNROLL-NEXT:    [[TMP36:%.*]] = phi <4 x i32> [ [[TMP31]], [[PRED_UDIV_CONTINUE14]] ], [ [[TMP35]], [[PRED_UDIV_IF15]] ]
-; UNROLL-NEXT:    [[TMP37:%.*]] = extractelement <4 x i1> [[TMP2]], i64 3
-; UNROLL-NEXT:    br i1 [[TMP37]], label [[PRED_UDIV_IF17:%.*]], label [[PRED_UDIV_CONTINUE18]]
-; UNROLL:       pred.udiv.if17:
-; UNROLL-NEXT:    [[TMP38:%.*]] = add i32 [[OFFSET_IDX]], -7
-; UNROLL-NEXT:    [[TMP39:%.*]] = udiv i32 219220132, [[TMP38]]
-; UNROLL-NEXT:    [[TMP40:%.*]] = insertelement <4 x i32> [[TMP36]], i32 [[TMP39]], i64 3
-; UNROLL-NEXT:    br label [[PRED_UDIV_CONTINUE18]]
-; UNROLL:       pred.udiv.continue18:
-; UNROLL-NEXT:    [[TMP41]] = phi <4 x i32> [ [[TMP36]], [[PRED_UDIV_CONTINUE16]] ], [ [[TMP40]], [[PRED_UDIV_IF17]] ]
-; UNROLL-NEXT:    [[TMP42:%.*]] = shufflevector <4 x i32> [[VECTOR_RECUR]], <4 x i32> [[TMP21]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; UNROLL-NEXT:    [[TMP43:%.*]] = shufflevector <4 x i32> [[TMP21]], <4 x i32> [[TMP41]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; UNROLL-NEXT:    [[TMP44]] = add <4 x i32> [[VEC_PHI]], [[TMP42]]
-; UNROLL-NEXT:    [[TMP45]] = add <4 x i32> [[VEC_PHI1]], [[TMP43]]
-; UNROLL-NEXT:    [[INDEX_NEXT]] = add i32 [[INDEX]], 8
-; UNROLL-NEXT:    [[TMP46:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]]
-; UNROLL-NEXT:    br i1 [[TMP46]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !prof [[PROF24:![0-9]+]], !llvm.loop [[LOOP25:![0-9]+]]
-; UNROLL:       middle.block:
-; UNROLL-NEXT:    [[TMP47:%.*]] = select <4 x i1> [[TMP2]], <4 x i32> [[TMP45]], <4 x i32> [[VEC_PHI1]]
-; UNROLL-NEXT:    [[TMP48:%.*]] = select <4 x i1> [[TMP1]], <4 x i32> [[TMP44]], <4 x i32> [[VEC_PHI]]
-; UNROLL-NEXT:    [[BIN_RDX:%.*]] = add <4 x i32> [[TMP47]], [[TMP48]]
-; UNROLL-NEXT:    [[TMP49:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[BIN_RDX]])
-; UNROLL-NEXT:    br i1 true, label [[BB1:%.*]], label [[SCALAR_PH]]
-; UNROLL:       scalar.ph:
-; UNROLL-NEXT:    br label [[BB2:%.*]]
-; UNROLL:       bb1:
-; UNROLL-NEXT:    [[VAR:%.*]] = phi i32 [ undef, [[BB2]] ], [ [[TMP49]], [[MIDDLE_BLOCK]] ]
-; UNROLL-NEXT:    ret i32 [[VAR]]
-; UNROLL:       bb2:
-; UNROLL-NEXT:    br i1 undef, label [[BB1]], label [[BB2]], !prof [[PROF26:![0-9]+]], !llvm.loop [[LOOP27:![0-9]+]]
-;
 ; UNROLL-NO-IC-LABEL: @sink_into_replication_region(
 ; UNROLL-NO-IC-NEXT:  bb:
 ; UNROLL-NO-IC-NEXT:    [[TMP0:%.*]] = add i32 [[Y:%.*]], 1
@@ -4479,296 +3027,6 @@ bb:
 define i32 @sink_into_replication_region_multiple(i32 *%x, i32 %y) {
 ;
 ; CHECK-LABEL: @sink_into_replication_region_multiple(
-; CHECK-NEXT:  bb:
-; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; CHECK:       vector.ph:
-; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @llvm.smax.i32(i32 [[Y:%.*]], i32 1)
-; CHECK-NEXT:    [[N_RND_UP:%.*]] = add nuw i32 [[TMP0]], 3
-; CHECK-NEXT:    [[N_VEC:%.*]] = and i32 [[N_RND_UP]], -4
-; CHECK-NEXT:    [[TRIP_COUNT_MINUS_1:%.*]] = add nsw i32 [[TMP0]], -1
-; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[TRIP_COUNT_MINUS_1]], i64 0
-; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
-; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
-; CHECK:       vector.body:
-; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE13:%.*]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 0, i32 1, i32 2, i32 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[PRED_STORE_CONTINUE13]] ]
-; CHECK-NEXT:    [[VECTOR_RECUR:%.*]] = phi <4 x i32> [ <i32 poison, i32 poison, i32 poison, i32 0>, [[VECTOR_PH]] ], [ [[TMP20:%.*]], [[PRED_STORE_CONTINUE13]] ]
-; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP22:%.*]], [[PRED_STORE_CONTINUE13]] ]
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = sub i32 [[Y]], [[INDEX]]
-; CHECK-NEXT:    [[TMP1:%.*]] = add i32 [[OFFSET_IDX]], -1
-; CHECK-NEXT:    [[TMP2:%.*]] = add i32 [[OFFSET_IDX]], -2
-; CHECK-NEXT:    [[TMP3:%.*]] = add i32 [[OFFSET_IDX]], -3
-; CHECK-NEXT:    [[TMP4:%.*]] = icmp ule <4 x i32> [[VEC_IND]], [[BROADCAST_SPLAT]]
-; CHECK-NEXT:    [[TMP5:%.*]] = extractelement <4 x i1> [[TMP4]], i64 0
-; CHECK-NEXT:    br i1 [[TMP5]], label [[PRED_UDIV_IF:%.*]], label [[PRED_UDIV_CONTINUE:%.*]]
-; CHECK:       pred.udiv.if:
-; CHECK-NEXT:    [[TMP6:%.*]] = udiv i32 219220132, [[OFFSET_IDX]]
-; CHECK-NEXT:    [[TMP7:%.*]] = insertelement <4 x i32> poison, i32 [[TMP6]], i64 0
-; CHECK-NEXT:    br label [[PRED_UDIV_CONTINUE]]
-; CHECK:       pred.udiv.continue:
-; CHECK-NEXT:    [[TMP8:%.*]] = phi <4 x i32> [ poison, [[VECTOR_BODY]] ], [ [[TMP7]], [[PRED_UDIV_IF]] ]
-; CHECK-NEXT:    [[TMP9:%.*]] = extractelement <4 x i1> [[TMP4]], i64 1
-; CHECK-NEXT:    br i1 [[TMP9]], label [[PRED_UDIV_IF2:%.*]], label [[PRED_UDIV_CONTINUE3:%.*]]
-; CHECK:       pred.udiv.if2:
-; CHECK-NEXT:    [[TMP10:%.*]] = udiv i32 219220132, [[TMP1]]
-; CHECK-NEXT:    [[TMP11:%.*]] = insertelement <4 x i32> [[TMP8]], i32 [[TMP10]], i64 1
-; CHECK-NEXT:    br label [[PRED_UDIV_CONTINUE3]]
-; CHECK:       pred.udiv.continue3:
-; CHECK-NEXT:    [[TMP12:%.*]] = phi <4 x i32> [ [[TMP8]], [[PRED_UDIV_CONTINUE]] ], [ [[TMP11]], [[PRED_UDIV_IF2]] ]
-; CHECK-NEXT:    [[TMP13:%.*]] = extractelement <4 x i1> [[TMP4]], i64 2
-; CHECK-NEXT:    br i1 [[TMP13]], label [[PRED_UDIV_IF4:%.*]], label [[PRED_UDIV_CONTINUE5:%.*]]
-; CHECK:       pred.udiv.if4:
-; CHECK-NEXT:    [[TMP14:%.*]] = udiv i32 219220132, [[TMP2]]
-; CHECK-NEXT:    [[TMP15:%.*]] = insertelement <4 x i32> [[TMP12]], i32 [[TMP14]], i64 2
-; CHECK-NEXT:    br label [[PRED_UDIV_CONTINUE5]]
-; CHECK:       pred.udiv.continue5:
-; CHECK-NEXT:    [[TMP16:%.*]] = phi <4 x i32> [ [[TMP12]], [[PRED_UDIV_CONTINUE3]] ], [ [[TMP15]], [[PRED_UDIV_IF4]] ]
-; CHECK-NEXT:    [[TMP17:%.*]] = extractelement <4 x i1> [[TMP4]], i64 3
-; CHECK-NEXT:    br i1 [[TMP17]], label [[PRED_UDIV_IF6:%.*]], label [[PRED_UDIV_CONTINUE7:%.*]]
-; CHECK:       pred.udiv.if6:
-; CHECK-NEXT:    [[TMP18:%.*]] = udiv i32 219220132, [[TMP3]]
-; CHECK-NEXT:    [[TMP19:%.*]] = insertelement <4 x i32> [[TMP16]], i32 [[TMP18]], i64 3
-; CHECK-NEXT:    br label [[PRED_UDIV_CONTINUE7]]
-; CHECK:       pred.udiv.continue7:
-; CHECK-NEXT:    [[TMP20]] = phi <4 x i32> [ [[TMP16]], [[PRED_UDIV_CONTINUE5]] ], [ [[TMP19]], [[PRED_UDIV_IF6]] ]
-; CHECK-NEXT:    [[TMP21:%.*]] = shufflevector <4 x i32> [[VECTOR_RECUR]], <4 x i32> [[TMP20]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; CHECK-NEXT:    [[TMP22]] = add <4 x i32> [[VEC_PHI]], [[TMP21]]
-; CHECK-NEXT:    [[TMP23:%.*]] = extractelement <4 x i1> [[TMP4]], i64 0
-; CHECK-NEXT:    br i1 [[TMP23]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
-; CHECK:       pred.store.if:
-; CHECK-NEXT:    [[TMP24:%.*]] = sext i32 [[INDEX]] to i64
-; CHECK-NEXT:    [[TMP25:%.*]] = getelementptr inbounds i32, i32* [[X:%.*]], i64 [[TMP24]]
-; CHECK-NEXT:    store i32 [[OFFSET_IDX]], i32* [[TMP25]], align 4
-; CHECK-NEXT:    br label [[PRED_STORE_CONTINUE]]
-; CHECK:       pred.store.continue:
-; CHECK-NEXT:    [[TMP26:%.*]] = extractelement <4 x i1> [[TMP4]], i64 1
-; CHECK-NEXT:    br i1 [[TMP26]], label [[PRED_STORE_IF8:%.*]], label [[PRED_STORE_CONTINUE9:%.*]]
-; CHECK:       pred.store.if8:
-; CHECK-NEXT:    [[TMP27:%.*]] = or i32 [[INDEX]], 1
-; CHECK-NEXT:    [[TMP28:%.*]] = sext i32 [[TMP27]] to i64
-; CHECK-NEXT:    [[TMP29:%.*]] = getelementptr inbounds i32, i32* [[X]], i64 [[TMP28]]
-; CHECK-NEXT:    store i32 [[TMP1]], i32* [[TMP29]], align 4
-; CHECK-NEXT:    br label [[PRED_STORE_CONTINUE9]]
-; CHECK:       pred.store.continue9:
-; CHECK-NEXT:    [[TMP30:%.*]] = extractelement <4 x i1> [[TMP4]], i64 2
-; CHECK-NEXT:    br i1 [[TMP30]], label [[PRED_STORE_IF10:%.*]], label [[PRED_STORE_CONTINUE11:%.*]]
-; CHECK:       pred.store.if10:
-; CHECK-NEXT:    [[TMP31:%.*]] = or i32 [[INDEX]], 2
-; CHECK-NEXT:    [[TMP32:%.*]] = sext i32 [[TMP31]] to i64
-; CHECK-NEXT:    [[TMP33:%.*]] = getelementptr inbounds i32, i32* [[X]], i64 [[TMP32]]
-; CHECK-NEXT:    store i32 [[TMP2]], i32* [[TMP33]], align 4
-; CHECK-NEXT:    br label [[PRED_STORE_CONTINUE11]]
-; CHECK:       pred.store.continue11:
-; CHECK-NEXT:    [[TMP34:%.*]] = extractelement <4 x i1> [[TMP4]], i64 3
-; CHECK-NEXT:    br i1 [[TMP34]], label [[PRED_STORE_IF12:%.*]], label [[PRED_STORE_CONTINUE13]]
-; CHECK:       pred.store.if12:
-; CHECK-NEXT:    [[TMP35:%.*]] = or i32 [[INDEX]], 3
-; CHECK-NEXT:    [[TMP36:%.*]] = sext i32 [[TMP35]] to i64
-; CHECK-NEXT:    [[TMP37:%.*]] = getelementptr inbounds i32, i32* [[X]], i64 [[TMP36]]
-; CHECK-NEXT:    store i32 [[TMP3]], i32* [[TMP37]], align 4
-; CHECK-NEXT:    br label [[PRED_STORE_CONTINUE13]]
-; CHECK:       pred.store.continue13:
-; CHECK-NEXT:    [[INDEX_NEXT]] = add i32 [[INDEX]], 4
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <4 x i32> [[VEC_IND]], <i32 4, i32 4, i32 4, i32 4>
-; CHECK-NEXT:    [[TMP38:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-NEXT:    br i1 [[TMP38]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !prof [[PROF24]], !llvm.loop [[LOOP28:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[TMP39:%.*]] = select <4 x i1> [[TMP4]], <4 x i32> [[TMP22]], <4 x i32> [[VEC_PHI]]
-; CHECK-NEXT:    [[TMP40:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP39]])
-; CHECK-NEXT:    br i1 true, label [[BB1:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    br label [[BB2:%.*]]
-; CHECK:       bb1:
-; CHECK-NEXT:    [[VAR:%.*]] = phi i32 [ undef, [[BB2]] ], [ [[TMP40]], [[MIDDLE_BLOCK]] ]
-; CHECK-NEXT:    ret i32 [[VAR]]
-; CHECK:       bb2:
-; CHECK-NEXT:    br i1 undef, label [[BB1]], label [[BB2]], !prof [[PROF26]], !llvm.loop [[LOOP29:![0-9]+]]
-;
-; UNROLL-LABEL: @sink_into_replication_region_multiple(
-; UNROLL-NEXT:  bb:
-; UNROLL-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; UNROLL:       vector.ph:
-; UNROLL-NEXT:    [[TMP0:%.*]] = call i32 @llvm.smax.i32(i32 [[Y:%.*]], i32 1)
-; UNROLL-NEXT:    [[N_RND_UP:%.*]] = add nuw i32 [[TMP0]], 7
-; UNROLL-NEXT:    [[N_VEC:%.*]] = and i32 [[N_RND_UP]], -8
-; UNROLL-NEXT:    [[TRIP_COUNT_MINUS_1:%.*]] = add nsw i32 [[TMP0]], -1
-; UNROLL-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i32> poison, i32 [[TRIP_COUNT_MINUS_1]], i64 0
-; UNROLL-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i32> [[BROADCAST_SPLATINSERT]], <4 x i32> poison, <4 x i32> zeroinitializer
-; UNROLL-NEXT:    br label [[VECTOR_BODY:%.*]]
-; UNROLL:       vector.body:
-; UNROLL-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[PRED_STORE_CONTINUE31:%.*]] ]
-; UNROLL-NEXT:    [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 0, i32 1, i32 2, i32 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[PRED_STORE_CONTINUE31]] ]
-; UNROLL-NEXT:    [[VECTOR_RECUR:%.*]] = phi <4 x i32> [ <i32 poison, i32 poison, i32 poison, i32 0>, [[VECTOR_PH]] ], [ [[TMP41:%.*]], [[PRED_STORE_CONTINUE31]] ]
-; UNROLL-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP44:%.*]], [[PRED_STORE_CONTINUE31]] ]
-; UNROLL-NEXT:    [[VEC_PHI3:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[TMP45:%.*]], [[PRED_STORE_CONTINUE31]] ]
-; UNROLL-NEXT:    [[STEP_ADD:%.*]] = add <4 x i32> [[VEC_IND]], <i32 4, i32 4, i32 4, i32 4>
-; UNROLL-NEXT:    [[OFFSET_IDX:%.*]] = sub i32 [[Y]], [[INDEX]]
-; UNROLL-NEXT:    [[TMP1:%.*]] = add i32 [[OFFSET_IDX]], -1
-; UNROLL-NEXT:    [[TMP2:%.*]] = add i32 [[OFFSET_IDX]], -2
-; UNROLL-NEXT:    [[TMP3:%.*]] = add i32 [[OFFSET_IDX]], -3
-; UNROLL-NEXT:    [[TMP4:%.*]] = add i32 [[OFFSET_IDX]], -4
-; UNROLL-NEXT:    [[TMP5:%.*]] = add i32 [[OFFSET_IDX]], -5
-; UNROLL-NEXT:    [[TMP6:%.*]] = add i32 [[OFFSET_IDX]], -6
-; UNROLL-NEXT:    [[TMP7:%.*]] = add i32 [[OFFSET_IDX]], -7
-; UNROLL-NEXT:    [[TMP8:%.*]] = icmp ule <4 x i32> [[VEC_IND]], [[BROADCAST_SPLAT]]
-; UNROLL-NEXT:    [[TMP9:%.*]] = icmp ule <4 x i32> [[STEP_ADD]], [[BROADCAST_SPLAT]]
-; UNROLL-NEXT:    [[TMP10:%.*]] = extractelement <4 x i1> [[TMP8]], i64 0
-; UNROLL-NEXT:    br i1 [[TMP10]], label [[PRED_UDIV_IF:%.*]], label [[PRED_UDIV_CONTINUE:%.*]]
-; UNROLL:       pred.udiv.if:
-; UNROLL-NEXT:    [[TMP11:%.*]] = udiv i32 219220132, [[OFFSET_IDX]]
-; UNROLL-NEXT:    [[TMP12:%.*]] = insertelement <4 x i32> poison, i32 [[TMP11]], i64 0
-; UNROLL-NEXT:    br label [[PRED_UDIV_CONTINUE]]
-; UNROLL:       pred.udiv.continue:
-; UNROLL-NEXT:    [[TMP13:%.*]] = phi <4 x i32> [ poison, [[VECTOR_BODY]] ], [ [[TMP12]], [[PRED_UDIV_IF]] ]
-; UNROLL-NEXT:    [[TMP14:%.*]] = extractelement <4 x i1> [[TMP8]], i64 1
-; UNROLL-NEXT:    br i1 [[TMP14]], label [[PRED_UDIV_IF4:%.*]], label [[PRED_UDIV_CONTINUE5:%.*]]
-; UNROLL:       pred.udiv.if4:
-; UNROLL-NEXT:    [[TMP15:%.*]] = udiv i32 219220132, [[TMP1]]
-; UNROLL-NEXT:    [[TMP16:%.*]] = insertelement <4 x i32> [[TMP13]], i32 [[TMP15]], i64 1
-; UNROLL-NEXT:    br label [[PRED_UDIV_CONTINUE5]]
-; UNROLL:       pred.udiv.continue5:
-; UNROLL-NEXT:    [[TMP17:%.*]] = phi <4 x i32> [ [[TMP13]], [[PRED_UDIV_CONTINUE]] ], [ [[TMP16]], [[PRED_UDIV_IF4]] ]
-; UNROLL-NEXT:    [[TMP18:%.*]] = extractelement <4 x i1> [[TMP8]], i64 2
-; UNROLL-NEXT:    br i1 [[TMP18]], label [[PRED_UDIV_IF6:%.*]], label [[PRED_UDIV_CONTINUE7:%.*]]
-; UNROLL:       pred.udiv.if6:
-; UNROLL-NEXT:    [[TMP19:%.*]] = udiv i32 219220132, [[TMP2]]
-; UNROLL-NEXT:    [[TMP20:%.*]] = insertelement <4 x i32> [[TMP17]], i32 [[TMP19]], i64 2
-; UNROLL-NEXT:    br label [[PRED_UDIV_CONTINUE7]]
-; UNROLL:       pred.udiv.continue7:
-; UNROLL-NEXT:    [[TMP21:%.*]] = phi <4 x i32> [ [[TMP17]], [[PRED_UDIV_CONTINUE5]] ], [ [[TMP20]], [[PRED_UDIV_IF6]] ]
-; UNROLL-NEXT:    [[TMP22:%.*]] = extractelement <4 x i1> [[TMP8]], i64 3
-; UNROLL-NEXT:    br i1 [[TMP22]], label [[PRED_UDIV_IF8:%.*]], label [[PRED_UDIV_CONTINUE9:%.*]]
-; UNROLL:       pred.udiv.if8:
-; UNROLL-NEXT:    [[TMP23:%.*]] = udiv i32 219220132, [[TMP3]]
-; UNROLL-NEXT:    [[TMP24:%.*]] = insertelement <4 x i32> [[TMP21]], i32 [[TMP23]], i64 3
-; UNROLL-NEXT:    br label [[PRED_UDIV_CONTINUE9]]
-; UNROLL:       pred.udiv.continue9:
-; UNROLL-NEXT:    [[TMP25:%.*]] = phi <4 x i32> [ [[TMP21]], [[PRED_UDIV_CONTINUE7]] ], [ [[TMP24]], [[PRED_UDIV_IF8]] ]
-; UNROLL-NEXT:    [[TMP26:%.*]] = extractelement <4 x i1> [[TMP9]], i64 0
-; UNROLL-NEXT:    br i1 [[TMP26]], label [[PRED_UDIV_IF10:%.*]], label [[PRED_UDIV_CONTINUE11:%.*]]
-; UNROLL:       pred.udiv.if10:
-; UNROLL-NEXT:    [[TMP27:%.*]] = udiv i32 219220132, [[TMP4]]
-; UNROLL-NEXT:    [[TMP28:%.*]] = insertelement <4 x i32> poison, i32 [[TMP27]], i64 0
-; UNROLL-NEXT:    br label [[PRED_UDIV_CONTINUE11]]
-; UNROLL:       pred.udiv.continue11:
-; UNROLL-NEXT:    [[TMP29:%.*]] = phi <4 x i32> [ poison, [[PRED_UDIV_CONTINUE9]] ], [ [[TMP28]], [[PRED_UDIV_IF10]] ]
-; UNROLL-NEXT:    [[TMP30:%.*]] = extractelement <4 x i1> [[TMP9]], i64 1
-; UNROLL-NEXT:    br i1 [[TMP30]], label [[PRED_UDIV_IF12:%.*]], label [[PRED_UDIV_CONTINUE13:%.*]]
-; UNROLL:       pred.udiv.if12:
-; UNROLL-NEXT:    [[TMP31:%.*]] = udiv i32 219220132, [[TMP5]]
-; UNROLL-NEXT:    [[TMP32:%.*]] = insertelement <4 x i32> [[TMP29]], i32 [[TMP31]], i64 1
-; UNROLL-NEXT:    br label [[PRED_UDIV_CONTINUE13]]
-; UNROLL:       pred.udiv.continue13:
-; UNROLL-NEXT:    [[TMP33:%.*]] = phi <4 x i32> [ [[TMP29]], [[PRED_UDIV_CONTINUE11]] ], [ [[TMP32]], [[PRED_UDIV_IF12]] ]
-; UNROLL-NEXT:    [[TMP34:%.*]] = extractelement <4 x i1> [[TMP9]], i64 2
-; UNROLL-NEXT:    br i1 [[TMP34]], label [[PRED_UDIV_IF14:%.*]], label [[PRED_UDIV_CONTINUE15:%.*]]
-; UNROLL:       pred.udiv.if14:
-; UNROLL-NEXT:    [[TMP35:%.*]] = udiv i32 219220132, [[TMP6]]
-; UNROLL-NEXT:    [[TMP36:%.*]] = insertelement <4 x i32> [[TMP33]], i32 [[TMP35]], i64 2
-; UNROLL-NEXT:    br label [[PRED_UDIV_CONTINUE15]]
-; UNROLL:       pred.udiv.continue15:
-; UNROLL-NEXT:    [[TMP37:%.*]] = phi <4 x i32> [ [[TMP33]], [[PRED_UDIV_CONTINUE13]] ], [ [[TMP36]], [[PRED_UDIV_IF14]] ]
-; UNROLL-NEXT:    [[TMP38:%.*]] = extractelement <4 x i1> [[TMP9]], i64 3
-; UNROLL-NEXT:    br i1 [[TMP38]], label [[PRED_UDIV_IF16:%.*]], label [[PRED_UDIV_CONTINUE17:%.*]]
-; UNROLL:       pred.udiv.if16:
-; UNROLL-NEXT:    [[TMP39:%.*]] = udiv i32 219220132, [[TMP7]]
-; UNROLL-NEXT:    [[TMP40:%.*]] = insertelement <4 x i32> [[TMP37]], i32 [[TMP39]], i64 3
-; UNROLL-NEXT:    br label [[PRED_UDIV_CONTINUE17]]
-; UNROLL:       pred.udiv.continue17:
-; UNROLL-NEXT:    [[TMP41]] = phi <4 x i32> [ [[TMP37]], [[PRED_UDIV_CONTINUE15]] ], [ [[TMP40]], [[PRED_UDIV_IF16]] ]
-; UNROLL-NEXT:    [[TMP42:%.*]] = shufflevector <4 x i32> [[VECTOR_RECUR]], <4 x i32> [[TMP25]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; UNROLL-NEXT:    [[TMP43:%.*]] = shufflevector <4 x i32> [[TMP25]], <4 x i32> [[TMP41]], <4 x i32> <i32 3, i32 4, i32 5, i32 6>
-; UNROLL-NEXT:    [[TMP44]] = add <4 x i32> [[VEC_PHI]], [[TMP42]]
-; UNROLL-NEXT:    [[TMP45]] = add <4 x i32> [[VEC_PHI3]], [[TMP43]]
-; UNROLL-NEXT:    [[TMP46:%.*]] = extractelement <4 x i1> [[TMP8]], i64 0
-; UNROLL-NEXT:    br i1 [[TMP46]], label [[PRED_STORE_IF:%.*]], label [[PRED_STORE_CONTINUE:%.*]]
-; UNROLL:       pred.store.if:
-; UNROLL-NEXT:    [[TMP47:%.*]] = sext i32 [[INDEX]] to i64
-; UNROLL-NEXT:    [[TMP48:%.*]] = getelementptr inbounds i32, i32* [[X:%.*]], i64 [[TMP47]]
-; UNROLL-NEXT:    store i32 [[OFFSET_IDX]], i32* [[TMP48]], align 4
-; UNROLL-NEXT:    br label [[PRED_STORE_CONTINUE]]
-; UNROLL:       pred.store.continue:
-; UNROLL-NEXT:    [[TMP49:%.*]] = extractelement <4 x i1> [[TMP8]], i64 1
-; UNROLL-NEXT:    br i1 [[TMP49]], label [[PRED_STORE_IF18:%.*]], label [[PRED_STORE_CONTINUE19:%.*]]
-; UNROLL:       pred.store.if18:
-; UNROLL-NEXT:    [[TMP50:%.*]] = or i32 [[INDEX]], 1
-; UNROLL-NEXT:    [[TMP51:%.*]] = sext i32 [[TMP50]] to i64
-; UNROLL-NEXT:    [[TMP52:%.*]] = getelementptr inbounds i32, i32* [[X]], i64 [[TMP51]]
-; UNROLL-NEXT:    store i32 [[TMP1]], i32* [[TMP52]], align 4
-; UNROLL-NEXT:    br label [[PRED_STORE_CONTINUE19]]
-; UNROLL:       pred.store.continue19:
-; UNROLL-NEXT:    [[TMP53:%.*]] = extractelement <4 x i1> [[TMP8]], i64 2
-; UNROLL-NEXT:    br i1 [[TMP53]], label [[PRED_STORE_IF20:%.*]], label [[PRED_STORE_CONTINUE21:%.*]]
-; UNROLL:       pred.store.if20:
-; UNROLL-NEXT:    [[TMP54:%.*]] = or i32 [[INDEX]], 2
-; UNROLL-NEXT:    [[TMP55:%.*]] = sext i32 [[TMP54]] to i64
-; UNROLL-NEXT:    [[TMP56:%.*]] = getelementptr inbounds i32, i32* [[X]], i64 [[TMP55]]
-; UNROLL-NEXT:    store i32 [[TMP2]], i32* [[TMP56]], align 4
-; UNROLL-NEXT:    br label [[PRED_STORE_CONTINUE21]]
-; UNROLL:       pred.store.continue21:
-; UNROLL-NEXT:    [[TMP57:%.*]] = extractelement <4 x i1> [[TMP8]], i64 3
-; UNROLL-NEXT:    br i1 [[TMP57]], label [[PRED_STORE_IF22:%.*]], label [[PRED_STORE_CONTINUE23:%.*]]
-; UNROLL:       pred.store.if22:
-; UNROLL-NEXT:    [[TMP58:%.*]] = or i32 [[INDEX]], 3
-; UNROLL-NEXT:    [[TMP59:%.*]] = sext i32 [[TMP58]] to i64
-; UNROLL-NEXT:    [[TMP60:%.*]] = getelementptr inbounds i32, i32* [[X]], i64 [[TMP59]]
-; UNROLL-NEXT:    store i32 [[TMP3]], i32* [[TMP60]], align 4
-; UNROLL-NEXT:    br label [[PRED_STORE_CONTINUE23]]
-; UNROLL:       pred.store.continue23:
-; UNROLL-NEXT:    [[TMP61:%.*]] = extractelement <4 x i1> [[TMP9]], i64 0
-; UNROLL-NEXT:    br i1 [[TMP61]], label [[PRED_STORE_IF24:%.*]], label [[PRED_STORE_CONTINUE25:%.*]]
-; UNROLL:       pred.store.if24:
-; UNROLL-NEXT:    [[TMP62:%.*]] = or i32 [[INDEX]], 4
-; UNROLL-NEXT:    [[TMP63:%.*]] = sext i32 [[TMP62]] to i64
-; UNROLL-NEXT:    [[TMP64:%.*]] = getelementptr inbounds i32, i32* [[X]], i64 [[TMP63]]
-; UNROLL-NEXT:    store i32 [[TMP4]], i32* [[TMP64]], align 4
-; UNROLL-NEXT:    br label [[PRED_STORE_CONTINUE25]]
-; UNROLL:       pred.store.continue25:
-; UNROLL-NEXT:    [[TMP65:%.*]] = extractelement <4 x i1> [[TMP9]], i64 1
-; UNROLL-NEXT:    br i1 [[TMP65]], label [[PRED_STORE_IF26:%.*]], label [[PRED_STORE_CONTINUE27:%.*]]
-; UNROLL:       pred.store.if26:
-; UNROLL-NEXT:    [[TMP66:%.*]] = or i32 [[INDEX]], 5
-; UNROLL-NEXT:    [[TMP67:%.*]] = sext i32 [[TMP66]] to i64
-; UNROLL-NEXT:    [[TMP68:%.*]] = getelementptr inbounds i32, i32* [[X]], i64 [[TMP67]]
-; UNROLL-NEXT:    store i32 [[TMP5]], i32* [[TMP68]], align 4
-; UNROLL-NEXT:    br label [[PRED_STORE_CONTINUE27]]
-; UNROLL:       pred.store.continue27:
-; UNROLL-NEXT:    [[TMP69:%.*]] = extractelement <4 x i1> [[TMP9]], i64 2
-; UNROLL-NEXT:    br i1 [[TMP69]], label [[PRED_STORE_IF28:%.*]], label [[PRED_STORE_CONTINUE29:%.*]]
-; UNROLL:       pred.store.if28:
-; UNROLL-NEXT:    [[TMP70:%.*]] = or i32 [[INDEX]], 6
-; UNROLL-NEXT:    [[TMP71:%.*]] = sext i32 [[TMP70]] to i64
-; UNROLL-NEXT:    [[TMP72:%.*]] = getelementptr inbounds i32, i32* [[X]], i64 [[TMP71]]
-; UNROLL-NEXT:    store i32 [[TMP6]], i32* [[TMP72]], align 4
-; UNROLL-NEXT:    br label [[PRED_STORE_CONTINUE29]]
-; UNROLL:       pred.store.continue29:
-; UNROLL-NEXT:    [[TMP73:%.*]] = extractelement <4 x i1> [[TMP9]], i64 3
-; UNROLL-NEXT:    br i1 [[TMP73]], label [[PRED_STORE_IF30:%.*]], label [[PRED_STORE_CONTINUE31]]
-; UNROLL:       pred.store.if30:
-; UNROLL-NEXT:    [[TMP74:%.*]] = or i32 [[INDEX]], 7
-; UNROLL-NEXT:    [[TMP75:%.*]] = sext i32 [[TMP74]] to i64
-; UNROLL-NEXT:    [[TMP76:%.*]] = getelementptr inbounds i32, i32* [[X]], i64 [[TMP75]]
-; UNROLL-NEXT:    store i32 [[TMP7]], i32* [[TMP76]], align 4
-; UNROLL-NEXT:    br label [[PRED_STORE_CONTINUE31]]
-; UNROLL:       pred.store.continue31:
-; UNROLL-NEXT:    [[INDEX_NEXT]] = add i32 [[INDEX]], 8
-; UNROLL-NEXT:    [[VEC_IND_NEXT]] = add <4 x i32> [[VEC_IND]], <i32 8, i32 8, i32 8, i32 8>
-; UNROLL-NEXT:    [[TMP77:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]]
-; UNROLL-NEXT:    br i1 [[TMP77]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !prof [[PROF24]], !llvm.loop [[LOOP28:![0-9]+]]
-; UNROLL:       middle.block:
-; UNROLL-NEXT:    [[TMP78:%.*]] = select <4 x i1> [[TMP9]], <4 x i32> [[TMP45]], <4 x i32> [[VEC_PHI3]]
-; UNROLL-NEXT:    [[TMP79:%.*]] = select <4 x i1> [[TMP8]], <4 x i32> [[TMP44]], <4 x i32> [[VEC_PHI]]
-; UNROLL-NEXT:    [[BIN_RDX:%.*]] = add <4 x i32> [[TMP78]], [[TMP79]]
-; UNROLL-NEXT:    [[TMP80:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[BIN_RDX]])
-; UNROLL-NEXT:    br i1 true, label [[BB1:%.*]], label [[SCALAR_PH]]
-; UNROLL:       scalar.ph:
-; UNROLL-NEXT:    br label [[BB2:%.*]]
-; UNROLL:       bb1:
-; UNROLL-NEXT:    [[VAR:%.*]] = phi i32 [ undef, [[BB2]] ], [ [[TMP80]], [[MIDDLE_BLOCK]] ]
-; UNROLL-NEXT:    ret i32 [[VAR]]
-; UNROLL:       bb2:
-; UNROLL-NEXT:    br i1 undef, label [[BB1]], label [[BB2]], !prof [[PROF26]], !llvm.loop [[LOOP29:![0-9]+]]
-;
 ; UNROLL-NO-IC-LABEL: @sink_into_replication_region_multiple(
 ; UNROLL-NO-IC-NEXT:  bb:
 ; UNROLL-NO-IC-NEXT:    [[TMP0:%.*]] = add i32 [[Y:%.*]], 1
@@ -5203,59 +3461,6 @@ bb:
 ; %vec.dead will be marked as dead instruction in the vector loop and no recipe
 ; will be created for it. Make sure a valid sink target is used.
 define void @sink_after_dead_inst(i32* %A.ptr) {
-; CHECK-LABEL: @sink_after_dead_inst(
-; CHECK-NEXT:  entry:
-; CHECK-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; CHECK:       vector.ph:
-; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
-; CHECK:       vector.body:
-; CHECK-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = zext i32 [[INDEX]] to i64
-; CHECK-NEXT:    [[SEXT:%.*]] = shl i64 [[OFFSET_IDX]], 48
-; CHECK-NEXT:    [[TMP0:%.*]] = ashr exact i64 [[SEXT]], 48
-; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr i32, i32* [[A_PTR:%.*]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP2:%.*]] = bitcast i32* [[TMP1]] to <4 x i32>*
-; CHECK-NEXT:    store <4 x i32> zeroinitializer, <4 x i32>* [[TMP2]], align 4
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
-; CHECK-NEXT:    [[TMP3:%.*]] = icmp eq i32 [[INDEX_NEXT]], 16
-; CHECK-NEXT:    br i1 [[TMP3]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP30:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    br i1 true, label [[FOR_END:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    br i1 undef, label [[FOR_END]], label [[LOOP]], !llvm.loop [[LOOP31:![0-9]+]]
-; CHECK:       for.end:
-; CHECK-NEXT:    ret void
-;
-; UNROLL-LABEL: @sink_after_dead_inst(
-; UNROLL-NEXT:  entry:
-; UNROLL-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
-; UNROLL:       vector.ph:
-; UNROLL-NEXT:    br label [[VECTOR_BODY:%.*]]
-; UNROLL:       vector.body:
-; UNROLL-NEXT:    [[INDEX:%.*]] = phi i32 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; UNROLL-NEXT:    [[OFFSET_IDX:%.*]] = zext i32 [[INDEX]] to i64
-; UNROLL-NEXT:    [[SEXT:%.*]] = shl i64 [[OFFSET_IDX]], 48
-; UNROLL-NEXT:    [[TMP0:%.*]] = ashr exact i64 [[SEXT]], 48
-; UNROLL-NEXT:    [[TMP1:%.*]] = getelementptr i32, i32* [[A_PTR:%.*]], i64 [[TMP0]]
-; UNROLL-NEXT:    [[TMP2:%.*]] = bitcast i32* [[TMP1]] to <4 x i32>*
-; UNROLL-NEXT:    store <4 x i32> zeroinitializer, <4 x i32>* [[TMP2]], align 4
-; UNROLL-NEXT:    [[TMP3:%.*]] = getelementptr i32, i32* [[TMP1]], i64 4
-; UNROLL-NEXT:    [[TMP4:%.*]] = bitcast i32* [[TMP3]] to <4 x i32>*
-; UNROLL-NEXT:    store <4 x i32> zeroinitializer, <4 x i32>* [[TMP4]], align 4
-; UNROLL-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 8
-; UNROLL-NEXT:    [[TMP5:%.*]] = icmp eq i32 [[INDEX_NEXT]], 16
-; UNROLL-NEXT:    br i1 [[TMP5]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP30:![0-9]+]]
-; UNROLL:       middle.block:
-; UNROLL-NEXT:    br i1 true, label [[FOR_END:%.*]], label [[SCALAR_PH]]
-; UNROLL:       scalar.ph:
-; UNROLL-NEXT:    br label [[LOOP:%.*]]
-; UNROLL:       loop:
-; UNROLL-NEXT:    br i1 undef, label [[FOR_END]], label [[LOOP]], !llvm.loop [[LOOP31:![0-9]+]]
-; UNROLL:       for.end:
-; UNROLL-NEXT:    ret void
-;
 ; UNROLL-NO-IC-LABEL: @sink_after_dead_inst(
 ; UNROLL-NO-IC-NEXT:  entry:
 ; UNROLL-NO-IC-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]


        
