[LV] Add check line with VF=4 to uniformity test.

Extend test coverage for D148841.

diff --git a/llvm/test/Transforms/LoopVectorize/uniform_across_vf_induction1_lshr.ll b/llvm/test/Transforms/LoopVectorize/uniform_across_vf_induction1_lshr.ll
index 7373363..8c27b0b 100644 (file)
--- a/llvm/test/Transforms/LoopVectorize/uniform_across_vf_induction1_lshr.ll
+++ b/llvm/test/Transforms/LoopVectorize/uniform_across_vf_induction1_lshr.ll
@@ -1,49 +1,90 @@
 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 2
-; RUN: opt -passes=loop-vectorize -force-vector-interleave=1 -force-vector-width=2 %s -S | FileCheck %s
+; RUN: opt -passes=loop-vectorize -force-vector-interleave=1 -force-vector-width=2 %s -S | FileCheck --check-prefixes=VF2 %s
+; RUN: opt -passes=loop-vectorize -force-vector-interleave=1 -force-vector-width=4 %s -S | FileCheck --check-prefixes=VF4 %s
 
 ; Tests for checking uniformity within a VF.
 
 ; for (iv = 0 ; ; iv += 1) B[iv] = A[iv>>0] + 42;
 define void @ld_lshr0_step1_start0_ind1(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_lshr0_step1_start0_ind1
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
-; CHECK-NEXT:    [[TMP1:%.*]] = lshr i64 [[TMP0]], 0
-; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP1]]
-; CHECK-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 0
-; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP3]], align 8
-; CHECK-NEXT:    [[TMP4:%.*]] = add nsw <2 x i64> [[WIDE_LOAD]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[TMP5]], i32 0
-; CHECK-NEXT:    store <2 x i64> [[TMP4]], ptr [[TMP6]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[TMP7:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[TMP7]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1000, 1000
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 0
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP3:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_lshr0_step1_start0_ind1
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; VF2-NEXT:    [[TMP1:%.*]] = lshr i64 [[TMP0]], 0
+; VF2-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP1]]
+; VF2-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 0
+; VF2-NEXT:    [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP3]], align 8
+; VF2-NEXT:    [[TMP4:%.*]] = add nsw <2 x i64> [[WIDE_LOAD]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[TMP5]], i32 0
+; VF2-NEXT:    store <2 x i64> [[TMP4]], ptr [[TMP6]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[TMP7:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
+; VF2-NEXT:    br i1 [[TMP7]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1000, 1000
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 0
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP3:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_lshr0_step1_start0_ind1
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; VF4-NEXT:    [[TMP1:%.*]] = lshr i64 [[TMP0]], 0
+; VF4-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP1]]
+; VF4-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[TMP2]], i32 0
+; VF4-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i64>, ptr [[TMP3]], align 8
+; VF4-NEXT:    [[TMP4:%.*]] = add nsw <4 x i64> [[WIDE_LOAD]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF4-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[TMP5]], i32 0
+; VF4-NEXT:    store <4 x i64> [[TMP4]], ptr [[TMP6]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[TMP7:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
+; VF4-NEXT:    br i1 [[TMP7]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1000, 1000
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 0
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP3:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -64,52 +105,107 @@ exit:
 
 ; for (iv = 0 ; ; iv += 1) B[iv] = A[iv>>1] + 42;
 define void @ld_lshr1_step1_start0_ind1(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_lshr1_step1_start0_ind1
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
-; CHECK-NEXT:    [[TMP1:%.*]] = lshr <2 x i64> [[VEC_IND]], <i64 1, i64 1>
-; CHECK-NEXT:    [[TMP2:%.*]] = extractelement <2 x i64> [[TMP1]], i32 0
-; CHECK-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP2]]
-; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP1]], i32 1
-; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
-; CHECK-NEXT:    [[TMP6:%.*]] = load i64, ptr [[TMP3]], align 8
-; CHECK-NEXT:    [[TMP7:%.*]] = load i64, ptr [[TMP5]], align 8
-; CHECK-NEXT:    [[TMP8:%.*]] = insertelement <2 x i64> poison, i64 [[TMP6]], i32 0
-; CHECK-NEXT:    [[TMP9:%.*]] = insertelement <2 x i64> [[TMP8]], i64 [[TMP7]], i32 1
-; CHECK-NEXT:    [[TMP10:%.*]] = add nsw <2 x i64> [[TMP9]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[TMP11]], i32 0
-; CHECK-NEXT:    store <2 x i64> [[TMP10]], ptr [[TMP12]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP13:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[TMP13]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1000, 1000
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 1
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP5:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_lshr1_step1_start0_ind1
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; VF2-NEXT:    [[TMP1:%.*]] = lshr <2 x i64> [[VEC_IND]], <i64 1, i64 1>
+; VF2-NEXT:    [[TMP2:%.*]] = extractelement <2 x i64> [[TMP1]], i32 0
+; VF2-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP2]]
+; VF2-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP1]], i32 1
+; VF2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
+; VF2-NEXT:    [[TMP6:%.*]] = load i64, ptr [[TMP3]], align 8
+; VF2-NEXT:    [[TMP7:%.*]] = load i64, ptr [[TMP5]], align 8
+; VF2-NEXT:    [[TMP8:%.*]] = insertelement <2 x i64> poison, i64 [[TMP6]], i32 0
+; VF2-NEXT:    [[TMP9:%.*]] = insertelement <2 x i64> [[TMP8]], i64 [[TMP7]], i32 1
+; VF2-NEXT:    [[TMP10:%.*]] = add nsw <2 x i64> [[TMP9]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF2-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[TMP11]], i32 0
+; VF2-NEXT:    store <2 x i64> [[TMP10]], ptr [[TMP12]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP13:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
+; VF2-NEXT:    br i1 [[TMP13]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1000, 1000
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 1
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP5:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_lshr1_step1_start0_ind1
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; VF4-NEXT:    [[TMP1:%.*]] = lshr <4 x i64> [[VEC_IND]], <i64 1, i64 1, i64 1, i64 1>
+; VF4-NEXT:    [[TMP2:%.*]] = extractelement <4 x i64> [[TMP1]], i32 0
+; VF4-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP2]]
+; VF4-NEXT:    [[TMP4:%.*]] = extractelement <4 x i64> [[TMP1]], i32 1
+; VF4-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
+; VF4-NEXT:    [[TMP6:%.*]] = extractelement <4 x i64> [[TMP1]], i32 2
+; VF4-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF4-NEXT:    [[TMP8:%.*]] = extractelement <4 x i64> [[TMP1]], i32 3
+; VF4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF4-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP3]], align 8
+; VF4-NEXT:    [[TMP11:%.*]] = load i64, ptr [[TMP5]], align 8
+; VF4-NEXT:    [[TMP12:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF4-NEXT:    [[TMP13:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF4-NEXT:    [[TMP14:%.*]] = insertelement <4 x i64> poison, i64 [[TMP10]], i32 0
+; VF4-NEXT:    [[TMP15:%.*]] = insertelement <4 x i64> [[TMP14]], i64 [[TMP11]], i32 1
+; VF4-NEXT:    [[TMP16:%.*]] = insertelement <4 x i64> [[TMP15]], i64 [[TMP12]], i32 2
+; VF4-NEXT:    [[TMP17:%.*]] = insertelement <4 x i64> [[TMP16]], i64 [[TMP13]], i32 3
+; VF4-NEXT:    [[TMP18:%.*]] = add nsw <4 x i64> [[TMP17]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP19:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF4-NEXT:    [[TMP20:%.*]] = getelementptr inbounds i64, ptr [[TMP19]], i32 0
+; VF4-NEXT:    store <4 x i64> [[TMP18]], ptr [[TMP20]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[TMP21:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
+; VF4-NEXT:    br i1 [[TMP21]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1000, 1000
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 1
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP5:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -130,52 +226,107 @@ exit:
 
 ; for (iv = 0 ; ; iv += 1) B[iv] = A[iv>>2] + 42;
 define void @ld_lshr2_step1_start0_ind1(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_lshr2_step1_start0_ind1
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
-; CHECK-NEXT:    [[TMP1:%.*]] = lshr <2 x i64> [[VEC_IND]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP2:%.*]] = extractelement <2 x i64> [[TMP1]], i32 0
-; CHECK-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP2]]
-; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP1]], i32 1
-; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
-; CHECK-NEXT:    [[TMP6:%.*]] = load i64, ptr [[TMP3]], align 8
-; CHECK-NEXT:    [[TMP7:%.*]] = load i64, ptr [[TMP5]], align 8
-; CHECK-NEXT:    [[TMP8:%.*]] = insertelement <2 x i64> poison, i64 [[TMP6]], i32 0
-; CHECK-NEXT:    [[TMP9:%.*]] = insertelement <2 x i64> [[TMP8]], i64 [[TMP7]], i32 1
-; CHECK-NEXT:    [[TMP10:%.*]] = add nsw <2 x i64> [[TMP9]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[TMP11]], i32 0
-; CHECK-NEXT:    store <2 x i64> [[TMP10]], ptr [[TMP12]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP13:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
-; 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 1000, 1000
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 2
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP7:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_lshr2_step1_start0_ind1
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; VF2-NEXT:    [[TMP1:%.*]] = lshr <2 x i64> [[VEC_IND]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP2:%.*]] = extractelement <2 x i64> [[TMP1]], i32 0
+; VF2-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP2]]
+; VF2-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP1]], i32 1
+; VF2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
+; VF2-NEXT:    [[TMP6:%.*]] = load i64, ptr [[TMP3]], align 8
+; VF2-NEXT:    [[TMP7:%.*]] = load i64, ptr [[TMP5]], align 8
+; VF2-NEXT:    [[TMP8:%.*]] = insertelement <2 x i64> poison, i64 [[TMP6]], i32 0
+; VF2-NEXT:    [[TMP9:%.*]] = insertelement <2 x i64> [[TMP8]], i64 [[TMP7]], i32 1
+; VF2-NEXT:    [[TMP10:%.*]] = add nsw <2 x i64> [[TMP9]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF2-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[TMP11]], i32 0
+; VF2-NEXT:    store <2 x i64> [[TMP10]], ptr [[TMP12]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP13:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
+; VF2-NEXT:    br i1 [[TMP13]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1000, 1000
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 2
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP7:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_lshr2_step1_start0_ind1
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; VF4-NEXT:    [[TMP1:%.*]] = lshr <4 x i64> [[VEC_IND]], <i64 2, i64 2, i64 2, i64 2>
+; VF4-NEXT:    [[TMP2:%.*]] = extractelement <4 x i64> [[TMP1]], i32 0
+; VF4-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP2]]
+; VF4-NEXT:    [[TMP4:%.*]] = extractelement <4 x i64> [[TMP1]], i32 1
+; VF4-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
+; VF4-NEXT:    [[TMP6:%.*]] = extractelement <4 x i64> [[TMP1]], i32 2
+; VF4-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF4-NEXT:    [[TMP8:%.*]] = extractelement <4 x i64> [[TMP1]], i32 3
+; VF4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF4-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP3]], align 8
+; VF4-NEXT:    [[TMP11:%.*]] = load i64, ptr [[TMP5]], align 8
+; VF4-NEXT:    [[TMP12:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF4-NEXT:    [[TMP13:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF4-NEXT:    [[TMP14:%.*]] = insertelement <4 x i64> poison, i64 [[TMP10]], i32 0
+; VF4-NEXT:    [[TMP15:%.*]] = insertelement <4 x i64> [[TMP14]], i64 [[TMP11]], i32 1
+; VF4-NEXT:    [[TMP16:%.*]] = insertelement <4 x i64> [[TMP15]], i64 [[TMP12]], i32 2
+; VF4-NEXT:    [[TMP17:%.*]] = insertelement <4 x i64> [[TMP16]], i64 [[TMP13]], i32 3
+; VF4-NEXT:    [[TMP18:%.*]] = add nsw <4 x i64> [[TMP17]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP19:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF4-NEXT:    [[TMP20:%.*]] = getelementptr inbounds i64, ptr [[TMP19]], i32 0
+; VF4-NEXT:    store <4 x i64> [[TMP18]], ptr [[TMP20]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[TMP21:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
+; VF4-NEXT:    br i1 [[TMP21]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1000, 1000
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 2
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP7:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -196,57 +347,125 @@ exit:
 
 ; for (iv = 0 ; ; iv += 2) B[iv] = A[iv>>0] + 42;
 define void @ld_lshr0_step2_start0_ind1(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_lshr0_step2_start0_ind1
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 2>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 2
-; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 2
-; CHECK-NEXT:    [[TMP2:%.*]] = lshr <2 x i64> [[VEC_IND]], zeroinitializer
-; CHECK-NEXT:    [[TMP3:%.*]] = extractelement <2 x i64> [[TMP2]], i32 0
-; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP3]]
-; CHECK-NEXT:    [[TMP5:%.*]] = extractelement <2 x i64> [[TMP2]], i32 1
-; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP5]]
-; CHECK-NEXT:    [[TMP7:%.*]] = load i64, ptr [[TMP4]], align 8
-; CHECK-NEXT:    [[TMP8:%.*]] = load i64, ptr [[TMP6]], align 8
-; CHECK-NEXT:    [[TMP9:%.*]] = insertelement <2 x i64> poison, i64 [[TMP7]], i32 0
-; CHECK-NEXT:    [[TMP10:%.*]] = insertelement <2 x i64> [[TMP9]], i64 [[TMP8]], i32 1
-; CHECK-NEXT:    [[TMP11:%.*]] = add nsw <2 x i64> [[TMP10]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
-; CHECK-NEXT:    [[TMP14:%.*]] = extractelement <2 x i64> [[TMP11]], i32 0
-; CHECK-NEXT:    store i64 [[TMP14]], ptr [[TMP12]], align 8
-; CHECK-NEXT:    [[TMP15:%.*]] = extractelement <2 x i64> [[TMP11]], i32 1
-; CHECK-NEXT:    store i64 [[TMP15]], ptr [[TMP13]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 4, i64 4>
-; CHECK-NEXT:    [[TMP16:%.*]] = icmp eq i64 [[INDEX_NEXT]], 500
-; CHECK-NEXT:    br i1 [[TMP16]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 500, 500
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 0
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP9:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_lshr0_step2_start0_ind1
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 2>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 2
+; VF2-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 2
+; VF2-NEXT:    [[TMP2:%.*]] = lshr <2 x i64> [[VEC_IND]], zeroinitializer
+; VF2-NEXT:    [[TMP3:%.*]] = extractelement <2 x i64> [[TMP2]], i32 0
+; VF2-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP3]]
+; VF2-NEXT:    [[TMP5:%.*]] = extractelement <2 x i64> [[TMP2]], i32 1
+; VF2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP5]]
+; VF2-NEXT:    [[TMP7:%.*]] = load i64, ptr [[TMP4]], align 8
+; VF2-NEXT:    [[TMP8:%.*]] = load i64, ptr [[TMP6]], align 8
+; VF2-NEXT:    [[TMP9:%.*]] = insertelement <2 x i64> poison, i64 [[TMP7]], i32 0
+; VF2-NEXT:    [[TMP10:%.*]] = insertelement <2 x i64> [[TMP9]], i64 [[TMP8]], i32 1
+; VF2-NEXT:    [[TMP11:%.*]] = add nsw <2 x i64> [[TMP10]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF2-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF2-NEXT:    [[TMP14:%.*]] = extractelement <2 x i64> [[TMP11]], i32 0
+; VF2-NEXT:    store i64 [[TMP14]], ptr [[TMP12]], align 8
+; VF2-NEXT:    [[TMP15:%.*]] = extractelement <2 x i64> [[TMP11]], i32 1
+; VF2-NEXT:    store i64 [[TMP15]], ptr [[TMP13]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 4, i64 4>
+; VF2-NEXT:    [[TMP16:%.*]] = icmp eq i64 [[INDEX_NEXT]], 500
+; VF2-NEXT:    br i1 [[TMP16]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 500, 500
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 0
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP9:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_lshr0_step2_start0_ind1
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 2, i64 4, i64 6>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 2
+; VF4-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 2
+; VF4-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 4
+; VF4-NEXT:    [[TMP3:%.*]] = add i64 [[OFFSET_IDX]], 6
+; VF4-NEXT:    [[TMP4:%.*]] = lshr <4 x i64> [[VEC_IND]], zeroinitializer
+; VF4-NEXT:    [[TMP5:%.*]] = extractelement <4 x i64> [[TMP4]], i32 0
+; VF4-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP5]]
+; VF4-NEXT:    [[TMP7:%.*]] = extractelement <4 x i64> [[TMP4]], i32 1
+; VF4-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP7]]
+; VF4-NEXT:    [[TMP9:%.*]] = extractelement <4 x i64> [[TMP4]], i32 2
+; VF4-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP9]]
+; VF4-NEXT:    [[TMP11:%.*]] = extractelement <4 x i64> [[TMP4]], i32 3
+; VF4-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP11]]
+; VF4-NEXT:    [[TMP13:%.*]] = load i64, ptr [[TMP6]], align 8
+; VF4-NEXT:    [[TMP14:%.*]] = load i64, ptr [[TMP8]], align 8
+; VF4-NEXT:    [[TMP15:%.*]] = load i64, ptr [[TMP10]], align 8
+; VF4-NEXT:    [[TMP16:%.*]] = load i64, ptr [[TMP12]], align 8
+; VF4-NEXT:    [[TMP17:%.*]] = insertelement <4 x i64> poison, i64 [[TMP13]], i32 0
+; VF4-NEXT:    [[TMP18:%.*]] = insertelement <4 x i64> [[TMP17]], i64 [[TMP14]], i32 1
+; VF4-NEXT:    [[TMP19:%.*]] = insertelement <4 x i64> [[TMP18]], i64 [[TMP15]], i32 2
+; VF4-NEXT:    [[TMP20:%.*]] = insertelement <4 x i64> [[TMP19]], i64 [[TMP16]], i32 3
+; VF4-NEXT:    [[TMP21:%.*]] = add nsw <4 x i64> [[TMP20]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP22:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF4-NEXT:    [[TMP23:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF4-NEXT:    [[TMP24:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF4-NEXT:    [[TMP25:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP3]]
+; VF4-NEXT:    [[TMP26:%.*]] = extractelement <4 x i64> [[TMP21]], i32 0
+; VF4-NEXT:    store i64 [[TMP26]], ptr [[TMP22]], align 8
+; VF4-NEXT:    [[TMP27:%.*]] = extractelement <4 x i64> [[TMP21]], i32 1
+; VF4-NEXT:    store i64 [[TMP27]], ptr [[TMP23]], align 8
+; VF4-NEXT:    [[TMP28:%.*]] = extractelement <4 x i64> [[TMP21]], i32 2
+; VF4-NEXT:    store i64 [[TMP28]], ptr [[TMP24]], align 8
+; VF4-NEXT:    [[TMP29:%.*]] = extractelement <4 x i64> [[TMP21]], i32 3
+; VF4-NEXT:    store i64 [[TMP29]], ptr [[TMP25]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 8, i64 8, i64 8, i64 8>
+; VF4-NEXT:    [[TMP30:%.*]] = icmp eq i64 [[INDEX_NEXT]], 500
+; VF4-NEXT:    br i1 [[TMP30]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 500, 500
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 0
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP9:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -267,50 +486,103 @@ exit:
 
 ; for (iv = 0 ; ; iv += 2) B[iv] = A[iv>>1] + 42;
 define void @ld_lshr1_step2_start0_ind1(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_lshr1_step2_start0_ind1
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 2
-; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 2
-; CHECK-NEXT:    [[TMP2:%.*]] = lshr i64 [[TMP0]], 1
-; CHECK-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP2]]
-; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[TMP3]], i32 0
-; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP4]], align 8
-; CHECK-NEXT:    [[TMP5:%.*]] = add nsw <2 x i64> [[WIDE_LOAD]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
-; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <2 x i64> [[TMP5]], i32 0
-; CHECK-NEXT:    store i64 [[TMP8]], ptr [[TMP6]], align 8
-; CHECK-NEXT:    [[TMP9:%.*]] = extractelement <2 x i64> [[TMP5]], i32 1
-; CHECK-NEXT:    store i64 [[TMP9]], ptr [[TMP7]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], 500
-; CHECK-NEXT:    br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 500, 500
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 1
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP11:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_lshr1_step2_start0_ind1
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 2
+; VF2-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 2
+; VF2-NEXT:    [[TMP2:%.*]] = lshr i64 [[TMP0]], 1
+; VF2-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP2]]
+; VF2-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[TMP3]], i32 0
+; VF2-NEXT:    [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP4]], align 8
+; VF2-NEXT:    [[TMP5:%.*]] = add nsw <2 x i64> [[WIDE_LOAD]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF2-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF2-NEXT:    [[TMP8:%.*]] = extractelement <2 x i64> [[TMP5]], i32 0
+; VF2-NEXT:    store i64 [[TMP8]], ptr [[TMP6]], align 8
+; VF2-NEXT:    [[TMP9:%.*]] = extractelement <2 x i64> [[TMP5]], i32 1
+; VF2-NEXT:    store i64 [[TMP9]], ptr [[TMP7]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], 500
+; VF2-NEXT:    br i1 [[TMP10]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 500, 500
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 1
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP11:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_lshr1_step2_start0_ind1
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 2
+; VF4-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 2
+; VF4-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 4
+; VF4-NEXT:    [[TMP3:%.*]] = add i64 [[OFFSET_IDX]], 6
+; VF4-NEXT:    [[TMP4:%.*]] = lshr i64 [[TMP0]], 1
+; VF4-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
+; VF4-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[TMP5]], i32 0
+; VF4-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i64>, ptr [[TMP6]], align 8
+; VF4-NEXT:    [[TMP7:%.*]] = add nsw <4 x i64> [[WIDE_LOAD]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF4-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF4-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP3]]
+; VF4-NEXT:    [[TMP12:%.*]] = extractelement <4 x i64> [[TMP7]], i32 0
+; VF4-NEXT:    store i64 [[TMP12]], ptr [[TMP8]], align 8
+; VF4-NEXT:    [[TMP13:%.*]] = extractelement <4 x i64> [[TMP7]], i32 1
+; VF4-NEXT:    store i64 [[TMP13]], ptr [[TMP9]], align 8
+; VF4-NEXT:    [[TMP14:%.*]] = extractelement <4 x i64> [[TMP7]], i32 2
+; VF4-NEXT:    store i64 [[TMP14]], ptr [[TMP10]], align 8
+; VF4-NEXT:    [[TMP15:%.*]] = extractelement <4 x i64> [[TMP7]], i32 3
+; VF4-NEXT:    store i64 [[TMP15]], ptr [[TMP11]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[TMP16:%.*]] = icmp eq i64 [[INDEX_NEXT]], 500
+; VF4-NEXT:    br i1 [[TMP16]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 500, 500
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 1
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP11:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -331,57 +603,125 @@ exit:
 
 ; for (iv = 0 ; ; iv += 3) B[iv] = A[iv>>0] + 42;
 define void @ld_lshr0_step3_start0_ind1(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_lshr0_step3_start0_ind1
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 3
-; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 3
-; CHECK-NEXT:    [[TMP2:%.*]] = lshr <2 x i64> [[VEC_IND]], zeroinitializer
-; CHECK-NEXT:    [[TMP3:%.*]] = extractelement <2 x i64> [[TMP2]], i32 0
-; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP3]]
-; CHECK-NEXT:    [[TMP5:%.*]] = extractelement <2 x i64> [[TMP2]], i32 1
-; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP5]]
-; CHECK-NEXT:    [[TMP7:%.*]] = load i64, ptr [[TMP4]], align 8
-; CHECK-NEXT:    [[TMP8:%.*]] = load i64, ptr [[TMP6]], align 8
-; CHECK-NEXT:    [[TMP9:%.*]] = insertelement <2 x i64> poison, i64 [[TMP7]], i32 0
-; CHECK-NEXT:    [[TMP10:%.*]] = insertelement <2 x i64> [[TMP9]], i64 [[TMP8]], i32 1
-; CHECK-NEXT:    [[TMP11:%.*]] = add nsw <2 x i64> [[TMP10]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
-; CHECK-NEXT:    [[TMP14:%.*]] = extractelement <2 x i64> [[TMP11]], i32 0
-; CHECK-NEXT:    store i64 [[TMP14]], ptr [[TMP12]], align 8
-; CHECK-NEXT:    [[TMP15:%.*]] = extractelement <2 x i64> [[TMP11]], i32 1
-; CHECK-NEXT:    store i64 [[TMP15]], ptr [[TMP13]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 6, i64 6>
-; CHECK-NEXT:    [[TMP16:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
-; CHECK-NEXT:    br i1 [[TMP16]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 996, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 0
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP13:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_lshr0_step3_start0_ind1
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 3
+; VF2-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 3
+; VF2-NEXT:    [[TMP2:%.*]] = lshr <2 x i64> [[VEC_IND]], zeroinitializer
+; VF2-NEXT:    [[TMP3:%.*]] = extractelement <2 x i64> [[TMP2]], i32 0
+; VF2-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP3]]
+; VF2-NEXT:    [[TMP5:%.*]] = extractelement <2 x i64> [[TMP2]], i32 1
+; VF2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP5]]
+; VF2-NEXT:    [[TMP7:%.*]] = load i64, ptr [[TMP4]], align 8
+; VF2-NEXT:    [[TMP8:%.*]] = load i64, ptr [[TMP6]], align 8
+; VF2-NEXT:    [[TMP9:%.*]] = insertelement <2 x i64> poison, i64 [[TMP7]], i32 0
+; VF2-NEXT:    [[TMP10:%.*]] = insertelement <2 x i64> [[TMP9]], i64 [[TMP8]], i32 1
+; VF2-NEXT:    [[TMP11:%.*]] = add nsw <2 x i64> [[TMP10]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF2-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF2-NEXT:    [[TMP14:%.*]] = extractelement <2 x i64> [[TMP11]], i32 0
+; VF2-NEXT:    store i64 [[TMP14]], ptr [[TMP12]], align 8
+; VF2-NEXT:    [[TMP15:%.*]] = extractelement <2 x i64> [[TMP11]], i32 1
+; VF2-NEXT:    store i64 [[TMP15]], ptr [[TMP13]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 6, i64 6>
+; VF2-NEXT:    [[TMP16:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
+; VF2-NEXT:    br i1 [[TMP16]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 996, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 0
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP13:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_lshr0_step3_start0_ind1
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 3, i64 6, i64 9>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 3
+; VF4-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 3
+; VF4-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 6
+; VF4-NEXT:    [[TMP3:%.*]] = add i64 [[OFFSET_IDX]], 9
+; VF4-NEXT:    [[TMP4:%.*]] = lshr <4 x i64> [[VEC_IND]], zeroinitializer
+; VF4-NEXT:    [[TMP5:%.*]] = extractelement <4 x i64> [[TMP4]], i32 0
+; VF4-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP5]]
+; VF4-NEXT:    [[TMP7:%.*]] = extractelement <4 x i64> [[TMP4]], i32 1
+; VF4-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP7]]
+; VF4-NEXT:    [[TMP9:%.*]] = extractelement <4 x i64> [[TMP4]], i32 2
+; VF4-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP9]]
+; VF4-NEXT:    [[TMP11:%.*]] = extractelement <4 x i64> [[TMP4]], i32 3
+; VF4-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP11]]
+; VF4-NEXT:    [[TMP13:%.*]] = load i64, ptr [[TMP6]], align 8
+; VF4-NEXT:    [[TMP14:%.*]] = load i64, ptr [[TMP8]], align 8
+; VF4-NEXT:    [[TMP15:%.*]] = load i64, ptr [[TMP10]], align 8
+; VF4-NEXT:    [[TMP16:%.*]] = load i64, ptr [[TMP12]], align 8
+; VF4-NEXT:    [[TMP17:%.*]] = insertelement <4 x i64> poison, i64 [[TMP13]], i32 0
+; VF4-NEXT:    [[TMP18:%.*]] = insertelement <4 x i64> [[TMP17]], i64 [[TMP14]], i32 1
+; VF4-NEXT:    [[TMP19:%.*]] = insertelement <4 x i64> [[TMP18]], i64 [[TMP15]], i32 2
+; VF4-NEXT:    [[TMP20:%.*]] = insertelement <4 x i64> [[TMP19]], i64 [[TMP16]], i32 3
+; VF4-NEXT:    [[TMP21:%.*]] = add nsw <4 x i64> [[TMP20]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP22:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF4-NEXT:    [[TMP23:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF4-NEXT:    [[TMP24:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF4-NEXT:    [[TMP25:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP3]]
+; VF4-NEXT:    [[TMP26:%.*]] = extractelement <4 x i64> [[TMP21]], i32 0
+; VF4-NEXT:    store i64 [[TMP26]], ptr [[TMP22]], align 8
+; VF4-NEXT:    [[TMP27:%.*]] = extractelement <4 x i64> [[TMP21]], i32 1
+; VF4-NEXT:    store i64 [[TMP27]], ptr [[TMP23]], align 8
+; VF4-NEXT:    [[TMP28:%.*]] = extractelement <4 x i64> [[TMP21]], i32 2
+; VF4-NEXT:    store i64 [[TMP28]], ptr [[TMP24]], align 8
+; VF4-NEXT:    [[TMP29:%.*]] = extractelement <4 x i64> [[TMP21]], i32 3
+; VF4-NEXT:    store i64 [[TMP29]], ptr [[TMP25]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 12, i64 12, i64 12, i64 12>
+; VF4-NEXT:    [[TMP30:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
+; VF4-NEXT:    br i1 [[TMP30]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 996, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 0
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP13:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -402,57 +742,125 @@ exit:
 
 ; for (iv = 0 ; ; iv += 3) B[iv] = A[iv>>1] + 42;
 define void @ld_lshr1_step3_start0_ind1(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_lshr1_step3_start0_ind1
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 3
-; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 3
-; CHECK-NEXT:    [[TMP2:%.*]] = lshr <2 x i64> [[VEC_IND]], <i64 1, i64 1>
-; CHECK-NEXT:    [[TMP3:%.*]] = extractelement <2 x i64> [[TMP2]], i32 0
-; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP3]]
-; CHECK-NEXT:    [[TMP5:%.*]] = extractelement <2 x i64> [[TMP2]], i32 1
-; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP5]]
-; CHECK-NEXT:    [[TMP7:%.*]] = load i64, ptr [[TMP4]], align 8
-; CHECK-NEXT:    [[TMP8:%.*]] = load i64, ptr [[TMP6]], align 8
-; CHECK-NEXT:    [[TMP9:%.*]] = insertelement <2 x i64> poison, i64 [[TMP7]], i32 0
-; CHECK-NEXT:    [[TMP10:%.*]] = insertelement <2 x i64> [[TMP9]], i64 [[TMP8]], i32 1
-; CHECK-NEXT:    [[TMP11:%.*]] = add nsw <2 x i64> [[TMP10]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
-; CHECK-NEXT:    [[TMP14:%.*]] = extractelement <2 x i64> [[TMP11]], i32 0
-; CHECK-NEXT:    store i64 [[TMP14]], ptr [[TMP12]], align 8
-; CHECK-NEXT:    [[TMP15:%.*]] = extractelement <2 x i64> [[TMP11]], i32 1
-; CHECK-NEXT:    store i64 [[TMP15]], ptr [[TMP13]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 6, i64 6>
-; CHECK-NEXT:    [[TMP16:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
-; CHECK-NEXT:    br i1 [[TMP16]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 996, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 1
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP15:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_lshr1_step3_start0_ind1
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 3
+; VF2-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 3
+; VF2-NEXT:    [[TMP2:%.*]] = lshr <2 x i64> [[VEC_IND]], <i64 1, i64 1>
+; VF2-NEXT:    [[TMP3:%.*]] = extractelement <2 x i64> [[TMP2]], i32 0
+; VF2-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP3]]
+; VF2-NEXT:    [[TMP5:%.*]] = extractelement <2 x i64> [[TMP2]], i32 1
+; VF2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP5]]
+; VF2-NEXT:    [[TMP7:%.*]] = load i64, ptr [[TMP4]], align 8
+; VF2-NEXT:    [[TMP8:%.*]] = load i64, ptr [[TMP6]], align 8
+; VF2-NEXT:    [[TMP9:%.*]] = insertelement <2 x i64> poison, i64 [[TMP7]], i32 0
+; VF2-NEXT:    [[TMP10:%.*]] = insertelement <2 x i64> [[TMP9]], i64 [[TMP8]], i32 1
+; VF2-NEXT:    [[TMP11:%.*]] = add nsw <2 x i64> [[TMP10]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF2-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF2-NEXT:    [[TMP14:%.*]] = extractelement <2 x i64> [[TMP11]], i32 0
+; VF2-NEXT:    store i64 [[TMP14]], ptr [[TMP12]], align 8
+; VF2-NEXT:    [[TMP15:%.*]] = extractelement <2 x i64> [[TMP11]], i32 1
+; VF2-NEXT:    store i64 [[TMP15]], ptr [[TMP13]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 6, i64 6>
+; VF2-NEXT:    [[TMP16:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
+; VF2-NEXT:    br i1 [[TMP16]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 996, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 1
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP15:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_lshr1_step3_start0_ind1
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 3, i64 6, i64 9>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 3
+; VF4-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 3
+; VF4-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 6
+; VF4-NEXT:    [[TMP3:%.*]] = add i64 [[OFFSET_IDX]], 9
+; VF4-NEXT:    [[TMP4:%.*]] = lshr <4 x i64> [[VEC_IND]], <i64 1, i64 1, i64 1, i64 1>
+; VF4-NEXT:    [[TMP5:%.*]] = extractelement <4 x i64> [[TMP4]], i32 0
+; VF4-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP5]]
+; VF4-NEXT:    [[TMP7:%.*]] = extractelement <4 x i64> [[TMP4]], i32 1
+; VF4-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP7]]
+; VF4-NEXT:    [[TMP9:%.*]] = extractelement <4 x i64> [[TMP4]], i32 2
+; VF4-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP9]]
+; VF4-NEXT:    [[TMP11:%.*]] = extractelement <4 x i64> [[TMP4]], i32 3
+; VF4-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP11]]
+; VF4-NEXT:    [[TMP13:%.*]] = load i64, ptr [[TMP6]], align 8
+; VF4-NEXT:    [[TMP14:%.*]] = load i64, ptr [[TMP8]], align 8
+; VF4-NEXT:    [[TMP15:%.*]] = load i64, ptr [[TMP10]], align 8
+; VF4-NEXT:    [[TMP16:%.*]] = load i64, ptr [[TMP12]], align 8
+; VF4-NEXT:    [[TMP17:%.*]] = insertelement <4 x i64> poison, i64 [[TMP13]], i32 0
+; VF4-NEXT:    [[TMP18:%.*]] = insertelement <4 x i64> [[TMP17]], i64 [[TMP14]], i32 1
+; VF4-NEXT:    [[TMP19:%.*]] = insertelement <4 x i64> [[TMP18]], i64 [[TMP15]], i32 2
+; VF4-NEXT:    [[TMP20:%.*]] = insertelement <4 x i64> [[TMP19]], i64 [[TMP16]], i32 3
+; VF4-NEXT:    [[TMP21:%.*]] = add nsw <4 x i64> [[TMP20]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP22:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF4-NEXT:    [[TMP23:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF4-NEXT:    [[TMP24:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF4-NEXT:    [[TMP25:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP3]]
+; VF4-NEXT:    [[TMP26:%.*]] = extractelement <4 x i64> [[TMP21]], i32 0
+; VF4-NEXT:    store i64 [[TMP26]], ptr [[TMP22]], align 8
+; VF4-NEXT:    [[TMP27:%.*]] = extractelement <4 x i64> [[TMP21]], i32 1
+; VF4-NEXT:    store i64 [[TMP27]], ptr [[TMP23]], align 8
+; VF4-NEXT:    [[TMP28:%.*]] = extractelement <4 x i64> [[TMP21]], i32 2
+; VF4-NEXT:    store i64 [[TMP28]], ptr [[TMP24]], align 8
+; VF4-NEXT:    [[TMP29:%.*]] = extractelement <4 x i64> [[TMP21]], i32 3
+; VF4-NEXT:    store i64 [[TMP29]], ptr [[TMP25]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 12, i64 12, i64 12, i64 12>
+; VF4-NEXT:    [[TMP30:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
+; VF4-NEXT:    br i1 [[TMP30]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 996, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 1
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP15:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -473,53 +881,109 @@ exit:
 
 ; for (iv = 1 ; ; iv += 1) B[iv] = A[iv>>1] + 42;
 define void @ld_lshr1_step1_start1_ind1(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_lshr1_step1_start1_ind1
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 2>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[INDEX]]
-; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP1:%.*]] = lshr <2 x i64> [[VEC_IND]], <i64 1, i64 1>
-; CHECK-NEXT:    [[TMP2:%.*]] = extractelement <2 x i64> [[TMP1]], i32 0
-; CHECK-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP2]]
-; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP1]], i32 1
-; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
-; CHECK-NEXT:    [[TMP6:%.*]] = load i64, ptr [[TMP3]], align 8
-; CHECK-NEXT:    [[TMP7:%.*]] = load i64, ptr [[TMP5]], align 8
-; CHECK-NEXT:    [[TMP8:%.*]] = insertelement <2 x i64> poison, i64 [[TMP6]], i32 0
-; CHECK-NEXT:    [[TMP9:%.*]] = insertelement <2 x i64> [[TMP8]], i64 [[TMP7]], i32 1
-; CHECK-NEXT:    [[TMP10:%.*]] = add nsw <2 x i64> [[TMP9]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[TMP11]], i32 0
-; CHECK-NEXT:    store <2 x i64> [[TMP10]], ptr [[TMP12]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP13:%.*]] = icmp eq i64 [[INDEX_NEXT]], 998
-; CHECK-NEXT:    br i1 [[TMP13]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 999, 998
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 999, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 1
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP17:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_lshr1_step1_start1_ind1
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 2>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[INDEX]]
+; VF2-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP1:%.*]] = lshr <2 x i64> [[VEC_IND]], <i64 1, i64 1>
+; VF2-NEXT:    [[TMP2:%.*]] = extractelement <2 x i64> [[TMP1]], i32 0
+; VF2-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP2]]
+; VF2-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP1]], i32 1
+; VF2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
+; VF2-NEXT:    [[TMP6:%.*]] = load i64, ptr [[TMP3]], align 8
+; VF2-NEXT:    [[TMP7:%.*]] = load i64, ptr [[TMP5]], align 8
+; VF2-NEXT:    [[TMP8:%.*]] = insertelement <2 x i64> poison, i64 [[TMP6]], i32 0
+; VF2-NEXT:    [[TMP9:%.*]] = insertelement <2 x i64> [[TMP8]], i64 [[TMP7]], i32 1
+; VF2-NEXT:    [[TMP10:%.*]] = add nsw <2 x i64> [[TMP9]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF2-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[TMP11]], i32 0
+; VF2-NEXT:    store <2 x i64> [[TMP10]], ptr [[TMP12]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP13:%.*]] = icmp eq i64 [[INDEX_NEXT]], 998
+; VF2-NEXT:    br i1 [[TMP13]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 999, 998
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 999, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 1
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP17:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_lshr1_step1_start1_ind1
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 1, i64 2, i64 3, i64 4>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[INDEX]]
+; VF4-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP1:%.*]] = lshr <4 x i64> [[VEC_IND]], <i64 1, i64 1, i64 1, i64 1>
+; VF4-NEXT:    [[TMP2:%.*]] = extractelement <4 x i64> [[TMP1]], i32 0
+; VF4-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP2]]
+; VF4-NEXT:    [[TMP4:%.*]] = extractelement <4 x i64> [[TMP1]], i32 1
+; VF4-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
+; VF4-NEXT:    [[TMP6:%.*]] = extractelement <4 x i64> [[TMP1]], i32 2
+; VF4-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF4-NEXT:    [[TMP8:%.*]] = extractelement <4 x i64> [[TMP1]], i32 3
+; VF4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF4-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP3]], align 8
+; VF4-NEXT:    [[TMP11:%.*]] = load i64, ptr [[TMP5]], align 8
+; VF4-NEXT:    [[TMP12:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF4-NEXT:    [[TMP13:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF4-NEXT:    [[TMP14:%.*]] = insertelement <4 x i64> poison, i64 [[TMP10]], i32 0
+; VF4-NEXT:    [[TMP15:%.*]] = insertelement <4 x i64> [[TMP14]], i64 [[TMP11]], i32 1
+; VF4-NEXT:    [[TMP16:%.*]] = insertelement <4 x i64> [[TMP15]], i64 [[TMP12]], i32 2
+; VF4-NEXT:    [[TMP17:%.*]] = insertelement <4 x i64> [[TMP16]], i64 [[TMP13]], i32 3
+; VF4-NEXT:    [[TMP18:%.*]] = add nsw <4 x i64> [[TMP17]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP19:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF4-NEXT:    [[TMP20:%.*]] = getelementptr inbounds i64, ptr [[TMP19]], i32 0
+; VF4-NEXT:    store <4 x i64> [[TMP18]], ptr [[TMP20]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[TMP21:%.*]] = icmp eq i64 [[INDEX_NEXT]], 996
+; VF4-NEXT:    br i1 [[TMP21]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 999, 996
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 1
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP17:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -540,51 +1004,105 @@ exit:
 
 ; for (iv = 1 ; ; iv += 2) B[iv] = A[iv>>1] + 42;
 define void @ld_lshr1_step2_start1_ind1(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_lshr1_step2_start1_ind1
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 2
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
-; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 2
-; CHECK-NEXT:    [[TMP3:%.*]] = lshr i64 [[TMP1]], 1
-; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP3]]
-; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[TMP4]], i32 0
-; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP5]], align 8
-; CHECK-NEXT:    [[TMP6:%.*]] = add nsw <2 x i64> [[WIDE_LOAD]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
-; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
-; CHECK-NEXT:    [[TMP9:%.*]] = extractelement <2 x i64> [[TMP6]], i32 0
-; CHECK-NEXT:    store i64 [[TMP9]], ptr [[TMP7]], align 8
-; CHECK-NEXT:    [[TMP10:%.*]] = extractelement <2 x i64> [[TMP6]], i32 1
-; CHECK-NEXT:    store i64 [[TMP10]], ptr [[TMP8]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], 498
-; CHECK-NEXT:    br i1 [[TMP11]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 499, 498
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 1
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP19:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_lshr1_step2_start1_ind1
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 2
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
+; VF2-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 2
+; VF2-NEXT:    [[TMP3:%.*]] = lshr i64 [[TMP1]], 1
+; VF2-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP3]]
+; VF2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[TMP4]], i32 0
+; VF2-NEXT:    [[WIDE_LOAD:%.*]] = load <2 x i64>, ptr [[TMP5]], align 8
+; VF2-NEXT:    [[TMP6:%.*]] = add nsw <2 x i64> [[WIDE_LOAD]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF2-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF2-NEXT:    [[TMP9:%.*]] = extractelement <2 x i64> [[TMP6]], i32 0
+; VF2-NEXT:    store i64 [[TMP9]], ptr [[TMP7]], align 8
+; VF2-NEXT:    [[TMP10:%.*]] = extractelement <2 x i64> [[TMP6]], i32 1
+; VF2-NEXT:    store i64 [[TMP10]], ptr [[TMP8]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], 498
+; VF2-NEXT:    br i1 [[TMP11]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 499, 498
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 1
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP19:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_lshr1_step2_start1_ind1
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 2
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
+; VF4-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 2
+; VF4-NEXT:    [[TMP3:%.*]] = add i64 [[OFFSET_IDX]], 4
+; VF4-NEXT:    [[TMP4:%.*]] = add i64 [[OFFSET_IDX]], 6
+; VF4-NEXT:    [[TMP5:%.*]] = lshr i64 [[TMP1]], 1
+; VF4-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP5]]
+; VF4-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[TMP6]], i32 0
+; VF4-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i64>, ptr [[TMP7]], align 8
+; VF4-NEXT:    [[TMP8:%.*]] = add nsw <4 x i64> [[WIDE_LOAD]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF4-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF4-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP3]]
+; VF4-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP4]]
+; VF4-NEXT:    [[TMP13:%.*]] = extractelement <4 x i64> [[TMP8]], i32 0
+; VF4-NEXT:    store i64 [[TMP13]], ptr [[TMP9]], align 8
+; VF4-NEXT:    [[TMP14:%.*]] = extractelement <4 x i64> [[TMP8]], i32 1
+; VF4-NEXT:    store i64 [[TMP14]], ptr [[TMP10]], align 8
+; VF4-NEXT:    [[TMP15:%.*]] = extractelement <4 x i64> [[TMP8]], i32 2
+; VF4-NEXT:    store i64 [[TMP15]], ptr [[TMP11]], align 8
+; VF4-NEXT:    [[TMP16:%.*]] = extractelement <4 x i64> [[TMP8]], i32 3
+; VF4-NEXT:    store i64 [[TMP16]], ptr [[TMP12]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[TMP17:%.*]] = icmp eq i64 [[INDEX_NEXT]], 496
+; VF4-NEXT:    br i1 [[TMP17]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 499, 496
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 993, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 1
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP19:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -605,58 +1123,127 @@ exit:
 
 ; for (iv = 1 ; ; iv += 3) B[iv] = A[iv>>1] + 42;
 define void @ld_lshr1_step3_start1_ind1(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_lshr1_step3_start1_ind1
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 4>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 3
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
-; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 3
-; CHECK-NEXT:    [[TMP3:%.*]] = lshr <2 x i64> [[VEC_IND]], <i64 1, i64 1>
-; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP3]], i32 0
-; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP3]], i32 1
-; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
-; CHECK-NEXT:    [[TMP8:%.*]] = load i64, ptr [[TMP5]], align 8
-; CHECK-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP7]], align 8
-; CHECK-NEXT:    [[TMP10:%.*]] = insertelement <2 x i64> poison, i64 [[TMP8]], i32 0
-; CHECK-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> [[TMP10]], i64 [[TMP9]], i32 1
-; CHECK-NEXT:    [[TMP12:%.*]] = add nsw <2 x i64> [[TMP11]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
-; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
-; CHECK-NEXT:    [[TMP15:%.*]] = extractelement <2 x i64> [[TMP12]], i32 0
-; CHECK-NEXT:    store i64 [[TMP15]], ptr [[TMP13]], align 8
-; CHECK-NEXT:    [[TMP16:%.*]] = extractelement <2 x i64> [[TMP12]], i32 1
-; CHECK-NEXT:    store i64 [[TMP16]], ptr [[TMP14]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 6, i64 6>
-; CHECK-NEXT:    [[TMP17:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
-; CHECK-NEXT:    br i1 [[TMP17]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP20:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 1
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP21:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_lshr1_step3_start1_ind1
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 4>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 3
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
+; VF2-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 3
+; VF2-NEXT:    [[TMP3:%.*]] = lshr <2 x i64> [[VEC_IND]], <i64 1, i64 1>
+; VF2-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP3]], i32 0
+; VF2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
+; VF2-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP3]], i32 1
+; VF2-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF2-NEXT:    [[TMP8:%.*]] = load i64, ptr [[TMP5]], align 8
+; VF2-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF2-NEXT:    [[TMP10:%.*]] = insertelement <2 x i64> poison, i64 [[TMP8]], i32 0
+; VF2-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> [[TMP10]], i64 [[TMP9]], i32 1
+; VF2-NEXT:    [[TMP12:%.*]] = add nsw <2 x i64> [[TMP11]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF2-NEXT:    [[TMP15:%.*]] = extractelement <2 x i64> [[TMP12]], i32 0
+; VF2-NEXT:    store i64 [[TMP15]], ptr [[TMP13]], align 8
+; VF2-NEXT:    [[TMP16:%.*]] = extractelement <2 x i64> [[TMP12]], i32 1
+; VF2-NEXT:    store i64 [[TMP16]], ptr [[TMP14]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 6, i64 6>
+; VF2-NEXT:    [[TMP17:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
+; VF2-NEXT:    br i1 [[TMP17]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP20:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 1
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP21:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_lshr1_step3_start1_ind1
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 1, i64 4, i64 7, i64 10>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 3
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
+; VF4-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 3
+; VF4-NEXT:    [[TMP3:%.*]] = add i64 [[OFFSET_IDX]], 6
+; VF4-NEXT:    [[TMP4:%.*]] = add i64 [[OFFSET_IDX]], 9
+; VF4-NEXT:    [[TMP5:%.*]] = lshr <4 x i64> [[VEC_IND]], <i64 1, i64 1, i64 1, i64 1>
+; VF4-NEXT:    [[TMP6:%.*]] = extractelement <4 x i64> [[TMP5]], i32 0
+; VF4-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF4-NEXT:    [[TMP8:%.*]] = extractelement <4 x i64> [[TMP5]], i32 1
+; VF4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF4-NEXT:    [[TMP10:%.*]] = extractelement <4 x i64> [[TMP5]], i32 2
+; VF4-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP10]]
+; VF4-NEXT:    [[TMP12:%.*]] = extractelement <4 x i64> [[TMP5]], i32 3
+; VF4-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP12]]
+; VF4-NEXT:    [[TMP14:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF4-NEXT:    [[TMP15:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF4-NEXT:    [[TMP16:%.*]] = load i64, ptr [[TMP11]], align 8
+; VF4-NEXT:    [[TMP17:%.*]] = load i64, ptr [[TMP13]], align 8
+; VF4-NEXT:    [[TMP18:%.*]] = insertelement <4 x i64> poison, i64 [[TMP14]], i32 0
+; VF4-NEXT:    [[TMP19:%.*]] = insertelement <4 x i64> [[TMP18]], i64 [[TMP15]], i32 1
+; VF4-NEXT:    [[TMP20:%.*]] = insertelement <4 x i64> [[TMP19]], i64 [[TMP16]], i32 2
+; VF4-NEXT:    [[TMP21:%.*]] = insertelement <4 x i64> [[TMP20]], i64 [[TMP17]], i32 3
+; VF4-NEXT:    [[TMP22:%.*]] = add nsw <4 x i64> [[TMP21]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP23:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF4-NEXT:    [[TMP24:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF4-NEXT:    [[TMP25:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP3]]
+; VF4-NEXT:    [[TMP26:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP4]]
+; VF4-NEXT:    [[TMP27:%.*]] = extractelement <4 x i64> [[TMP22]], i32 0
+; VF4-NEXT:    store i64 [[TMP27]], ptr [[TMP23]], align 8
+; VF4-NEXT:    [[TMP28:%.*]] = extractelement <4 x i64> [[TMP22]], i32 1
+; VF4-NEXT:    store i64 [[TMP28]], ptr [[TMP24]], align 8
+; VF4-NEXT:    [[TMP29:%.*]] = extractelement <4 x i64> [[TMP22]], i32 2
+; VF4-NEXT:    store i64 [[TMP29]], ptr [[TMP25]], align 8
+; VF4-NEXT:    [[TMP30:%.*]] = extractelement <4 x i64> [[TMP22]], i32 3
+; VF4-NEXT:    store i64 [[TMP30]], ptr [[TMP26]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 12, i64 12, i64 12, i64 12>
+; VF4-NEXT:    [[TMP31:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
+; VF4-NEXT:    br i1 [[TMP31]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP20:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 1
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP21:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -677,58 +1264,127 @@ exit:
 
 ; for (iv = 1 ; ; iv += 3) B[iv] = A[iv>>2] + 42;
 define void @ld_lshr2_step3_start1_ind1(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_lshr2_step3_start1_ind1
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 4>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 3
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
-; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 3
-; CHECK-NEXT:    [[TMP3:%.*]] = lshr <2 x i64> [[VEC_IND]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP3]], i32 0
-; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP3]], i32 1
-; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
-; CHECK-NEXT:    [[TMP8:%.*]] = load i64, ptr [[TMP5]], align 8
-; CHECK-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP7]], align 8
-; CHECK-NEXT:    [[TMP10:%.*]] = insertelement <2 x i64> poison, i64 [[TMP8]], i32 0
-; CHECK-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> [[TMP10]], i64 [[TMP9]], i32 1
-; CHECK-NEXT:    [[TMP12:%.*]] = add nsw <2 x i64> [[TMP11]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
-; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
-; CHECK-NEXT:    [[TMP15:%.*]] = extractelement <2 x i64> [[TMP12]], i32 0
-; CHECK-NEXT:    store i64 [[TMP15]], ptr [[TMP13]], align 8
-; CHECK-NEXT:    [[TMP16:%.*]] = extractelement <2 x i64> [[TMP12]], i32 1
-; CHECK-NEXT:    store i64 [[TMP16]], ptr [[TMP14]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 6, i64 6>
-; CHECK-NEXT:    [[TMP17:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
-; CHECK-NEXT:    br i1 [[TMP17]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP22:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 2
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP23:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_lshr2_step3_start1_ind1
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 4>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 3
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
+; VF2-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 3
+; VF2-NEXT:    [[TMP3:%.*]] = lshr <2 x i64> [[VEC_IND]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP3]], i32 0
+; VF2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
+; VF2-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP3]], i32 1
+; VF2-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF2-NEXT:    [[TMP8:%.*]] = load i64, ptr [[TMP5]], align 8
+; VF2-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF2-NEXT:    [[TMP10:%.*]] = insertelement <2 x i64> poison, i64 [[TMP8]], i32 0
+; VF2-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> [[TMP10]], i64 [[TMP9]], i32 1
+; VF2-NEXT:    [[TMP12:%.*]] = add nsw <2 x i64> [[TMP11]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF2-NEXT:    [[TMP15:%.*]] = extractelement <2 x i64> [[TMP12]], i32 0
+; VF2-NEXT:    store i64 [[TMP15]], ptr [[TMP13]], align 8
+; VF2-NEXT:    [[TMP16:%.*]] = extractelement <2 x i64> [[TMP12]], i32 1
+; VF2-NEXT:    store i64 [[TMP16]], ptr [[TMP14]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 6, i64 6>
+; VF2-NEXT:    [[TMP17:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
+; VF2-NEXT:    br i1 [[TMP17]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP22:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 2
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP23:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_lshr2_step3_start1_ind1
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 1, i64 4, i64 7, i64 10>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 3
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
+; VF4-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 3
+; VF4-NEXT:    [[TMP3:%.*]] = add i64 [[OFFSET_IDX]], 6
+; VF4-NEXT:    [[TMP4:%.*]] = add i64 [[OFFSET_IDX]], 9
+; VF4-NEXT:    [[TMP5:%.*]] = lshr <4 x i64> [[VEC_IND]], <i64 2, i64 2, i64 2, i64 2>
+; VF4-NEXT:    [[TMP6:%.*]] = extractelement <4 x i64> [[TMP5]], i32 0
+; VF4-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF4-NEXT:    [[TMP8:%.*]] = extractelement <4 x i64> [[TMP5]], i32 1
+; VF4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF4-NEXT:    [[TMP10:%.*]] = extractelement <4 x i64> [[TMP5]], i32 2
+; VF4-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP10]]
+; VF4-NEXT:    [[TMP12:%.*]] = extractelement <4 x i64> [[TMP5]], i32 3
+; VF4-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP12]]
+; VF4-NEXT:    [[TMP14:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF4-NEXT:    [[TMP15:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF4-NEXT:    [[TMP16:%.*]] = load i64, ptr [[TMP11]], align 8
+; VF4-NEXT:    [[TMP17:%.*]] = load i64, ptr [[TMP13]], align 8
+; VF4-NEXT:    [[TMP18:%.*]] = insertelement <4 x i64> poison, i64 [[TMP14]], i32 0
+; VF4-NEXT:    [[TMP19:%.*]] = insertelement <4 x i64> [[TMP18]], i64 [[TMP15]], i32 1
+; VF4-NEXT:    [[TMP20:%.*]] = insertelement <4 x i64> [[TMP19]], i64 [[TMP16]], i32 2
+; VF4-NEXT:    [[TMP21:%.*]] = insertelement <4 x i64> [[TMP20]], i64 [[TMP17]], i32 3
+; VF4-NEXT:    [[TMP22:%.*]] = add nsw <4 x i64> [[TMP21]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP23:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF4-NEXT:    [[TMP24:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF4-NEXT:    [[TMP25:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP3]]
+; VF4-NEXT:    [[TMP26:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP4]]
+; VF4-NEXT:    [[TMP27:%.*]] = extractelement <4 x i64> [[TMP22]], i32 0
+; VF4-NEXT:    store i64 [[TMP27]], ptr [[TMP23]], align 8
+; VF4-NEXT:    [[TMP28:%.*]] = extractelement <4 x i64> [[TMP22]], i32 1
+; VF4-NEXT:    store i64 [[TMP28]], ptr [[TMP24]], align 8
+; VF4-NEXT:    [[TMP29:%.*]] = extractelement <4 x i64> [[TMP22]], i32 2
+; VF4-NEXT:    store i64 [[TMP29]], ptr [[TMP25]], align 8
+; VF4-NEXT:    [[TMP30:%.*]] = extractelement <4 x i64> [[TMP22]], i32 3
+; VF4-NEXT:    store i64 [[TMP30]], ptr [[TMP26]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 12, i64 12, i64 12, i64 12>
+; VF4-NEXT:    [[TMP31:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
+; VF4-NEXT:    br i1 [[TMP31]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP22:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[LSHR:%.*]] = lshr i64 [[IV]], 2
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[LSHR]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP23:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop

diff --git a/llvm/test/Transforms/LoopVectorize/uniform_across_vf_induction2.ll b/llvm/test/Transforms/LoopVectorize/uniform_across_vf_induction2.ll
index 2569e17..f603357 100644 (file)
--- a/llvm/test/Transforms/LoopVectorize/uniform_across_vf_induction2.ll
+++ b/llvm/test/Transforms/LoopVectorize/uniform_across_vf_induction2.ll
@@ -1,63 +1,130 @@
 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 2
-; RUN: opt -passes=loop-vectorize -force-vector-interleave=1 -force-vector-width=2 %s -S | FileCheck %s
+; RUN: opt -passes=loop-vectorize -force-vector-interleave=1 -force-vector-width=2 %s -S | FileCheck --check-prefix=VF2 %s
+; RUN: opt -passes=loop-vectorize -force-vector-interleave=1 -force-vector-width=4 %s -S | FileCheck --check-prefix=VF4 %s
 
 ; for (iv = 0, iv2 = 0 ; ; iv += 1, iv2 += 1) B[iv] = A[iv/1 + iv2/1] + 42;
 define void @ld_div1_step1_start0_ind2(ptr noalias %A, ptr noalias %B) {
 ; CHECK-LABEL: define void @ld_div1_step1_start0_ind2
 ; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
-; CHECK-NEXT:    [[TMP1:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 1, i64 1>
-; CHECK-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 1, i64 1>
-; CHECK-NEXT:    [[TMP3:%.*]] = add <2 x i64> [[TMP1]], [[TMP2]]
-; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP3]], i32 0
-; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP3]], i32 1
-; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
-; CHECK-NEXT:    [[TMP8:%.*]] = load i64, ptr [[TMP5]], align 8
-; CHECK-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP7]], align 8
-; CHECK-NEXT:    [[TMP10:%.*]] = insertelement <2 x i64> poison, i64 [[TMP8]], i32 0
-; CHECK-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> [[TMP10]], i64 [[TMP9]], i32 1
-; CHECK-NEXT:    [[TMP12:%.*]] = add nsw <2 x i64> [[TMP11]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[TMP13]], i32 0
-; CHECK-NEXT:    store <2 x i64> [[TMP12]], ptr [[TMP14]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 2, i64 2>
-; CHECK-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1000, 1000
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 1
-; CHECK-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 1
-; CHECK-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP3:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; ; VF2-LABEL: define void @ld_div1_step1_start0_ind2
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; VF2-NEXT:    [[TMP1:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 1, i64 1>
+; VF2-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 1, i64 1>
+; VF2-NEXT:    [[TMP3:%.*]] = add <2 x i64> [[TMP1]], [[TMP2]]
+; VF2-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP3]], i32 0
+; VF2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
+; VF2-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP3]], i32 1
+; VF2-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF2-NEXT:    [[TMP8:%.*]] = load i64, ptr [[TMP5]], align 8
+; VF2-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF2-NEXT:    [[TMP10:%.*]] = insertelement <2 x i64> poison, i64 [[TMP8]], i32 0
+; VF2-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> [[TMP10]], i64 [[TMP9]], i32 1
+; VF2-NEXT:    [[TMP12:%.*]] = add nsw <2 x i64> [[TMP11]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[TMP13]], i32 0
+; VF2-NEXT:    store <2 x i64> [[TMP12]], ptr [[TMP14]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 2, i64 2>
+; VF2-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
+; VF2-NEXT:    br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1000, 1000
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF2-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 1
+; VF2-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 1
+; VF2-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP3:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_div1_step1_start0_ind2
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND2:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; VF4-NEXT:    [[TMP1:%.*]] = udiv <4 x i64> [[VEC_IND]], <i64 1, i64 1, i64 1, i64 1>
+; VF4-NEXT:    [[TMP2:%.*]] = udiv <4 x i64> [[VEC_IND2]], <i64 1, i64 1, i64 1, i64 1>
+; VF4-NEXT:    [[TMP3:%.*]] = add <4 x i64> [[TMP1]], [[TMP2]]
+; VF4-NEXT:    [[TMP4:%.*]] = extractelement <4 x i64> [[TMP3]], i32 0
+; VF4-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
+; VF4-NEXT:    [[TMP6:%.*]] = extractelement <4 x i64> [[TMP3]], i32 1
+; VF4-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF4-NEXT:    [[TMP8:%.*]] = extractelement <4 x i64> [[TMP3]], i32 2
+; VF4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF4-NEXT:    [[TMP10:%.*]] = extractelement <4 x i64> [[TMP3]], i32 3
+; VF4-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP10]]
+; VF4-NEXT:    [[TMP12:%.*]] = load i64, ptr [[TMP5]], align 8
+; VF4-NEXT:    [[TMP13:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF4-NEXT:    [[TMP14:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF4-NEXT:    [[TMP15:%.*]] = load i64, ptr [[TMP11]], align 8
+; VF4-NEXT:    [[TMP16:%.*]] = insertelement <4 x i64> poison, i64 [[TMP12]], i32 0
+; VF4-NEXT:    [[TMP17:%.*]] = insertelement <4 x i64> [[TMP16]], i64 [[TMP13]], i32 1
+; VF4-NEXT:    [[TMP18:%.*]] = insertelement <4 x i64> [[TMP17]], i64 [[TMP14]], i32 2
+; VF4-NEXT:    [[TMP19:%.*]] = insertelement <4 x i64> [[TMP18]], i64 [[TMP15]], i32 3
+; VF4-NEXT:    [[TMP20:%.*]] = add nsw <4 x i64> [[TMP19]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP21:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF4-NEXT:    [[TMP22:%.*]] = getelementptr inbounds i64, ptr [[TMP21]], i32 0
+; VF4-NEXT:    store <4 x i64> [[TMP20]], ptr [[TMP22]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[VEC_IND_NEXT3]] = add <4 x i64> [[VEC_IND2]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[TMP23:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
+; VF4-NEXT:    br i1 [[TMP23]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1000, 1000
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF4-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 1
+; VF4-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 1
+; VF4-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP3:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -84,59 +151,125 @@ exit:
 define void @ld_div2_step1_start0_ind2(ptr noalias %A, ptr noalias %B) {
 ; CHECK-LABEL: define void @ld_div2_step1_start0_ind2
 ; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
-; CHECK-NEXT:    [[TMP1:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP3:%.*]] = add <2 x i64> [[TMP1]], [[TMP2]]
-; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP3]], i32 0
-; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP3]], i32 1
-; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
-; CHECK-NEXT:    [[TMP8:%.*]] = load i64, ptr [[TMP5]], align 8
-; CHECK-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP7]], align 8
-; CHECK-NEXT:    [[TMP10:%.*]] = insertelement <2 x i64> poison, i64 [[TMP8]], i32 0
-; CHECK-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> [[TMP10]], i64 [[TMP9]], i32 1
-; CHECK-NEXT:    [[TMP12:%.*]] = add nsw <2 x i64> [[TMP11]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[TMP13]], i32 0
-; CHECK-NEXT:    store <2 x i64> [[TMP12]], ptr [[TMP14]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 2, i64 2>
-; CHECK-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1000, 1000
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 2
-; CHECK-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 2
-; CHECK-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP5:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_div2_step1_start0_ind2
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; VF2-NEXT:    [[TMP1:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP3:%.*]] = add <2 x i64> [[TMP1]], [[TMP2]]
+; VF2-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP3]], i32 0
+; VF2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
+; VF2-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP3]], i32 1
+; VF2-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF2-NEXT:    [[TMP8:%.*]] = load i64, ptr [[TMP5]], align 8
+; VF2-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF2-NEXT:    [[TMP10:%.*]] = insertelement <2 x i64> poison, i64 [[TMP8]], i32 0
+; VF2-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> [[TMP10]], i64 [[TMP9]], i32 1
+; VF2-NEXT:    [[TMP12:%.*]] = add nsw <2 x i64> [[TMP11]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[TMP13]], i32 0
+; VF2-NEXT:    store <2 x i64> [[TMP12]], ptr [[TMP14]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 2, i64 2>
+; VF2-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
+; VF2-NEXT:    br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1000, 1000
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF2-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 2
+; VF2-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 2
+; VF2-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP5:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_div2_step1_start0_ind2
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND2:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; VF4-NEXT:    [[TMP1:%.*]] = udiv <4 x i64> [[VEC_IND]], <i64 2, i64 2, i64 2, i64 2>
+; VF4-NEXT:    [[TMP2:%.*]] = udiv <4 x i64> [[VEC_IND2]], <i64 2, i64 2, i64 2, i64 2>
+; VF4-NEXT:    [[TMP3:%.*]] = add <4 x i64> [[TMP1]], [[TMP2]]
+; VF4-NEXT:    [[TMP4:%.*]] = extractelement <4 x i64> [[TMP3]], i32 0
+; VF4-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
+; VF4-NEXT:    [[TMP6:%.*]] = extractelement <4 x i64> [[TMP3]], i32 1
+; VF4-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF4-NEXT:    [[TMP8:%.*]] = extractelement <4 x i64> [[TMP3]], i32 2
+; VF4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF4-NEXT:    [[TMP10:%.*]] = extractelement <4 x i64> [[TMP3]], i32 3
+; VF4-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP10]]
+; VF4-NEXT:    [[TMP12:%.*]] = load i64, ptr [[TMP5]], align 8
+; VF4-NEXT:    [[TMP13:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF4-NEXT:    [[TMP14:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF4-NEXT:    [[TMP15:%.*]] = load i64, ptr [[TMP11]], align 8
+; VF4-NEXT:    [[TMP16:%.*]] = insertelement <4 x i64> poison, i64 [[TMP12]], i32 0
+; VF4-NEXT:    [[TMP17:%.*]] = insertelement <4 x i64> [[TMP16]], i64 [[TMP13]], i32 1
+; VF4-NEXT:    [[TMP18:%.*]] = insertelement <4 x i64> [[TMP17]], i64 [[TMP14]], i32 2
+; VF4-NEXT:    [[TMP19:%.*]] = insertelement <4 x i64> [[TMP18]], i64 [[TMP15]], i32 3
+; VF4-NEXT:    [[TMP20:%.*]] = add nsw <4 x i64> [[TMP19]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP21:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF4-NEXT:    [[TMP22:%.*]] = getelementptr inbounds i64, ptr [[TMP21]], i32 0
+; VF4-NEXT:    store <4 x i64> [[TMP20]], ptr [[TMP22]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[VEC_IND_NEXT3]] = add <4 x i64> [[VEC_IND2]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[TMP23:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
+; VF4-NEXT:    br i1 [[TMP23]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1000, 1000
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF4-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 2
+; VF4-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 2
+; VF4-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP5:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -161,61 +294,125 @@ exit:
 
 ; for (iv = 0, iv2 = 0 ; ; iv += 1, iv2 += 1) B[iv] = A[iv/3 + iv2/3] + 42;
 define void @ld_div3_step1_start0_ind2(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_div3_step1_start0_ind2
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
-; CHECK-NEXT:    [[TMP1:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 3, i64 3>
-; CHECK-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 3, i64 3>
-; CHECK-NEXT:    [[TMP3:%.*]] = add <2 x i64> [[TMP1]], [[TMP2]]
-; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP3]], i32 0
-; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP3]], i32 1
-; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
-; CHECK-NEXT:    [[TMP8:%.*]] = load i64, ptr [[TMP5]], align 8
-; CHECK-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP7]], align 8
-; CHECK-NEXT:    [[TMP10:%.*]] = insertelement <2 x i64> poison, i64 [[TMP8]], i32 0
-; CHECK-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> [[TMP10]], i64 [[TMP9]], i32 1
-; CHECK-NEXT:    [[TMP12:%.*]] = add nsw <2 x i64> [[TMP11]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[TMP13]], i32 0
-; CHECK-NEXT:    store <2 x i64> [[TMP12]], ptr [[TMP14]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 2, i64 2>
-; CHECK-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1000, 1000
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 3
-; CHECK-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 3
-; CHECK-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP7:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_div3_step1_start0_ind2
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; VF2-NEXT:    [[TMP1:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 3, i64 3>
+; VF2-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 3, i64 3>
+; VF2-NEXT:    [[TMP3:%.*]] = add <2 x i64> [[TMP1]], [[TMP2]]
+; VF2-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP3]], i32 0
+; VF2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
+; VF2-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP3]], i32 1
+; VF2-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF2-NEXT:    [[TMP8:%.*]] = load i64, ptr [[TMP5]], align 8
+; VF2-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF2-NEXT:    [[TMP10:%.*]] = insertelement <2 x i64> poison, i64 [[TMP8]], i32 0
+; VF2-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> [[TMP10]], i64 [[TMP9]], i32 1
+; VF2-NEXT:    [[TMP12:%.*]] = add nsw <2 x i64> [[TMP11]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[TMP13]], i32 0
+; VF2-NEXT:    store <2 x i64> [[TMP12]], ptr [[TMP14]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 2, i64 2>
+; VF2-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
+; VF2-NEXT:    br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1000, 1000
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF2-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 3
+; VF2-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 3
+; VF2-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP7:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_div3_step1_start0_ind2
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND2:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[TMP0:%.*]] = add i64 [[INDEX]], 0
+; VF4-NEXT:    [[TMP1:%.*]] = udiv <4 x i64> [[VEC_IND]], <i64 3, i64 3, i64 3, i64 3>
+; VF4-NEXT:    [[TMP2:%.*]] = udiv <4 x i64> [[VEC_IND2]], <i64 3, i64 3, i64 3, i64 3>
+; VF4-NEXT:    [[TMP3:%.*]] = add <4 x i64> [[TMP1]], [[TMP2]]
+; VF4-NEXT:    [[TMP4:%.*]] = extractelement <4 x i64> [[TMP3]], i32 0
+; VF4-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
+; VF4-NEXT:    [[TMP6:%.*]] = extractelement <4 x i64> [[TMP3]], i32 1
+; VF4-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF4-NEXT:    [[TMP8:%.*]] = extractelement <4 x i64> [[TMP3]], i32 2
+; VF4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF4-NEXT:    [[TMP10:%.*]] = extractelement <4 x i64> [[TMP3]], i32 3
+; VF4-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP10]]
+; VF4-NEXT:    [[TMP12:%.*]] = load i64, ptr [[TMP5]], align 8
+; VF4-NEXT:    [[TMP13:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF4-NEXT:    [[TMP14:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF4-NEXT:    [[TMP15:%.*]] = load i64, ptr [[TMP11]], align 8
+; VF4-NEXT:    [[TMP16:%.*]] = insertelement <4 x i64> poison, i64 [[TMP12]], i32 0
+; VF4-NEXT:    [[TMP17:%.*]] = insertelement <4 x i64> [[TMP16]], i64 [[TMP13]], i32 1
+; VF4-NEXT:    [[TMP18:%.*]] = insertelement <4 x i64> [[TMP17]], i64 [[TMP14]], i32 2
+; VF4-NEXT:    [[TMP19:%.*]] = insertelement <4 x i64> [[TMP18]], i64 [[TMP15]], i32 3
+; VF4-NEXT:    [[TMP20:%.*]] = add nsw <4 x i64> [[TMP19]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP21:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF4-NEXT:    [[TMP22:%.*]] = getelementptr inbounds i64, ptr [[TMP21]], i32 0
+; VF4-NEXT:    store <4 x i64> [[TMP20]], ptr [[TMP22]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[VEC_IND_NEXT3]] = add <4 x i64> [[VEC_IND2]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[TMP23:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000
+; VF4-NEXT:    br i1 [[TMP23]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 1000, 1000
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF4-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 3
+; VF4-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 3
+; VF4-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP7:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -240,66 +437,143 @@ exit:
 
 ; for (iv = 0, iv2 = 0 ; ; iv += 2, iv2 += 1) B[iv] = A[iv/1 + iv2/1] + 42;
 define void @ld_div1_step2_start0_ind2(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_div1_step2_start0_ind2
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 2>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 2
-; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 2
-; CHECK-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 1, i64 1>
-; CHECK-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 1, i64 1>
-; CHECK-NEXT:    [[TMP4:%.*]] = add <2 x i64> [[TMP2]], [[TMP3]]
-; CHECK-NEXT:    [[TMP5:%.*]] = extractelement <2 x i64> [[TMP4]], i32 0
-; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP5]]
-; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <2 x i64> [[TMP4]], i32 1
-; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP7]]
-; CHECK-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP6]], align 8
-; CHECK-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP8]], align 8
-; CHECK-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> poison, i64 [[TMP9]], i32 0
-; CHECK-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> [[TMP11]], i64 [[TMP10]], i32 1
-; CHECK-NEXT:    [[TMP13:%.*]] = add nsw <2 x i64> [[TMP12]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
-; CHECK-NEXT:    [[TMP16:%.*]] = extractelement <2 x i64> [[TMP13]], i32 0
-; CHECK-NEXT:    store i64 [[TMP16]], ptr [[TMP14]], align 8
-; CHECK-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP13]], i32 1
-; CHECK-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 4, i64 4>
-; CHECK-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], 500
-; CHECK-NEXT:    br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 500, 500
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 500, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 1
-; CHECK-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 1
-; CHECK-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP9:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_div1_step2_start0_ind2
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 2>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 2
+; VF2-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 2
+; VF2-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 1, i64 1>
+; VF2-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 1, i64 1>
+; VF2-NEXT:    [[TMP4:%.*]] = add <2 x i64> [[TMP2]], [[TMP3]]
+; VF2-NEXT:    [[TMP5:%.*]] = extractelement <2 x i64> [[TMP4]], i32 0
+; VF2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP5]]
+; VF2-NEXT:    [[TMP7:%.*]] = extractelement <2 x i64> [[TMP4]], i32 1
+; VF2-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP7]]
+; VF2-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP6]], align 8
+; VF2-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP8]], align 8
+; VF2-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> poison, i64 [[TMP9]], i32 0
+; VF2-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> [[TMP11]], i64 [[TMP10]], i32 1
+; VF2-NEXT:    [[TMP13:%.*]] = add nsw <2 x i64> [[TMP12]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF2-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF2-NEXT:    [[TMP16:%.*]] = extractelement <2 x i64> [[TMP13]], i32 0
+; VF2-NEXT:    store i64 [[TMP16]], ptr [[TMP14]], align 8
+; VF2-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP13]], i32 1
+; VF2-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 4, i64 4>
+; VF2-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], 500
+; VF2-NEXT:    br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 500, 500
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF2-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 500, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 1
+; VF2-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 1
+; VF2-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP9:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_div1_step2_start0_ind2
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 2, i64 4, i64 6>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND2:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 2
+; VF4-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 2
+; VF4-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 4
+; VF4-NEXT:    [[TMP3:%.*]] = add i64 [[OFFSET_IDX]], 6
+; VF4-NEXT:    [[TMP4:%.*]] = udiv <4 x i64> [[VEC_IND]], <i64 1, i64 1, i64 1, i64 1>
+; VF4-NEXT:    [[TMP5:%.*]] = udiv <4 x i64> [[VEC_IND2]], <i64 1, i64 1, i64 1, i64 1>
+; VF4-NEXT:    [[TMP6:%.*]] = add <4 x i64> [[TMP4]], [[TMP5]]
+; VF4-NEXT:    [[TMP7:%.*]] = extractelement <4 x i64> [[TMP6]], i32 0
+; VF4-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP7]]
+; VF4-NEXT:    [[TMP9:%.*]] = extractelement <4 x i64> [[TMP6]], i32 1
+; VF4-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP9]]
+; VF4-NEXT:    [[TMP11:%.*]] = extractelement <4 x i64> [[TMP6]], i32 2
+; VF4-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP11]]
+; VF4-NEXT:    [[TMP13:%.*]] = extractelement <4 x i64> [[TMP6]], i32 3
+; VF4-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP13]]
+; VF4-NEXT:    [[TMP15:%.*]] = load i64, ptr [[TMP8]], align 8
+; VF4-NEXT:    [[TMP16:%.*]] = load i64, ptr [[TMP10]], align 8
+; VF4-NEXT:    [[TMP17:%.*]] = load i64, ptr [[TMP12]], align 8
+; VF4-NEXT:    [[TMP18:%.*]] = load i64, ptr [[TMP14]], align 8
+; VF4-NEXT:    [[TMP19:%.*]] = insertelement <4 x i64> poison, i64 [[TMP15]], i32 0
+; VF4-NEXT:    [[TMP20:%.*]] = insertelement <4 x i64> [[TMP19]], i64 [[TMP16]], i32 1
+; VF4-NEXT:    [[TMP21:%.*]] = insertelement <4 x i64> [[TMP20]], i64 [[TMP17]], i32 2
+; VF4-NEXT:    [[TMP22:%.*]] = insertelement <4 x i64> [[TMP21]], i64 [[TMP18]], i32 3
+; VF4-NEXT:    [[TMP23:%.*]] = add nsw <4 x i64> [[TMP22]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP24:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF4-NEXT:    [[TMP25:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF4-NEXT:    [[TMP26:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF4-NEXT:    [[TMP27:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP3]]
+; VF4-NEXT:    [[TMP28:%.*]] = extractelement <4 x i64> [[TMP23]], i32 0
+; VF4-NEXT:    store i64 [[TMP28]], ptr [[TMP24]], align 8
+; VF4-NEXT:    [[TMP29:%.*]] = extractelement <4 x i64> [[TMP23]], i32 1
+; VF4-NEXT:    store i64 [[TMP29]], ptr [[TMP25]], align 8
+; VF4-NEXT:    [[TMP30:%.*]] = extractelement <4 x i64> [[TMP23]], i32 2
+; VF4-NEXT:    store i64 [[TMP30]], ptr [[TMP26]], align 8
+; VF4-NEXT:    [[TMP31:%.*]] = extractelement <4 x i64> [[TMP23]], i32 3
+; VF4-NEXT:    store i64 [[TMP31]], ptr [[TMP27]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 8, i64 8, i64 8, i64 8>
+; VF4-NEXT:    [[VEC_IND_NEXT3]] = add <4 x i64> [[VEC_IND2]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[TMP32:%.*]] = icmp eq i64 [[INDEX_NEXT]], 500
+; VF4-NEXT:    br i1 [[TMP32]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 500, 500
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF4-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 500, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 1
+; VF4-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 1
+; VF4-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP9:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -324,66 +598,143 @@ exit:
 
 ; for (iv = 0, iv2 = 0 ; ; iv += 2, iv2 += 1) B[iv] = A[iv/2 + iv2/2] + 42;
 define void @ld_div2_step2_start0_ind2(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_div2_step2_start0_ind2
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 2>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 2
-; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 2
-; CHECK-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP4:%.*]] = add <2 x i64> [[TMP2]], [[TMP3]]
-; CHECK-NEXT:    [[TMP5:%.*]] = extractelement <2 x i64> [[TMP4]], i32 0
-; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP5]]
-; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <2 x i64> [[TMP4]], i32 1
-; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP7]]
-; CHECK-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP6]], align 8
-; CHECK-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP8]], align 8
-; CHECK-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> poison, i64 [[TMP9]], i32 0
-; CHECK-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> [[TMP11]], i64 [[TMP10]], i32 1
-; CHECK-NEXT:    [[TMP13:%.*]] = add nsw <2 x i64> [[TMP12]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
-; CHECK-NEXT:    [[TMP16:%.*]] = extractelement <2 x i64> [[TMP13]], i32 0
-; CHECK-NEXT:    store i64 [[TMP16]], ptr [[TMP14]], align 8
-; CHECK-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP13]], i32 1
-; CHECK-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 4, i64 4>
-; CHECK-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], 500
-; CHECK-NEXT:    br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 500, 500
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 500, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 2
-; CHECK-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 2
-; CHECK-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP11:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_div2_step2_start0_ind2
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 2>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 2
+; VF2-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 2
+; VF2-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP4:%.*]] = add <2 x i64> [[TMP2]], [[TMP3]]
+; VF2-NEXT:    [[TMP5:%.*]] = extractelement <2 x i64> [[TMP4]], i32 0
+; VF2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP5]]
+; VF2-NEXT:    [[TMP7:%.*]] = extractelement <2 x i64> [[TMP4]], i32 1
+; VF2-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP7]]
+; VF2-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP6]], align 8
+; VF2-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP8]], align 8
+; VF2-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> poison, i64 [[TMP9]], i32 0
+; VF2-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> [[TMP11]], i64 [[TMP10]], i32 1
+; VF2-NEXT:    [[TMP13:%.*]] = add nsw <2 x i64> [[TMP12]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF2-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF2-NEXT:    [[TMP16:%.*]] = extractelement <2 x i64> [[TMP13]], i32 0
+; VF2-NEXT:    store i64 [[TMP16]], ptr [[TMP14]], align 8
+; VF2-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP13]], i32 1
+; VF2-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 4, i64 4>
+; VF2-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], 500
+; VF2-NEXT:    br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 500, 500
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF2-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 500, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 2
+; VF2-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 2
+; VF2-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP11:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_div2_step2_start0_ind2
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 2, i64 4, i64 6>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND2:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 2
+; VF4-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 2
+; VF4-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 4
+; VF4-NEXT:    [[TMP3:%.*]] = add i64 [[OFFSET_IDX]], 6
+; VF4-NEXT:    [[TMP4:%.*]] = udiv <4 x i64> [[VEC_IND]], <i64 2, i64 2, i64 2, i64 2>
+; VF4-NEXT:    [[TMP5:%.*]] = udiv <4 x i64> [[VEC_IND2]], <i64 2, i64 2, i64 2, i64 2>
+; VF4-NEXT:    [[TMP6:%.*]] = add <4 x i64> [[TMP4]], [[TMP5]]
+; VF4-NEXT:    [[TMP7:%.*]] = extractelement <4 x i64> [[TMP6]], i32 0
+; VF4-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP7]]
+; VF4-NEXT:    [[TMP9:%.*]] = extractelement <4 x i64> [[TMP6]], i32 1
+; VF4-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP9]]
+; VF4-NEXT:    [[TMP11:%.*]] = extractelement <4 x i64> [[TMP6]], i32 2
+; VF4-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP11]]
+; VF4-NEXT:    [[TMP13:%.*]] = extractelement <4 x i64> [[TMP6]], i32 3
+; VF4-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP13]]
+; VF4-NEXT:    [[TMP15:%.*]] = load i64, ptr [[TMP8]], align 8
+; VF4-NEXT:    [[TMP16:%.*]] = load i64, ptr [[TMP10]], align 8
+; VF4-NEXT:    [[TMP17:%.*]] = load i64, ptr [[TMP12]], align 8
+; VF4-NEXT:    [[TMP18:%.*]] = load i64, ptr [[TMP14]], align 8
+; VF4-NEXT:    [[TMP19:%.*]] = insertelement <4 x i64> poison, i64 [[TMP15]], i32 0
+; VF4-NEXT:    [[TMP20:%.*]] = insertelement <4 x i64> [[TMP19]], i64 [[TMP16]], i32 1
+; VF4-NEXT:    [[TMP21:%.*]] = insertelement <4 x i64> [[TMP20]], i64 [[TMP17]], i32 2
+; VF4-NEXT:    [[TMP22:%.*]] = insertelement <4 x i64> [[TMP21]], i64 [[TMP18]], i32 3
+; VF4-NEXT:    [[TMP23:%.*]] = add nsw <4 x i64> [[TMP22]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP24:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF4-NEXT:    [[TMP25:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF4-NEXT:    [[TMP26:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF4-NEXT:    [[TMP27:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP3]]
+; VF4-NEXT:    [[TMP28:%.*]] = extractelement <4 x i64> [[TMP23]], i32 0
+; VF4-NEXT:    store i64 [[TMP28]], ptr [[TMP24]], align 8
+; VF4-NEXT:    [[TMP29:%.*]] = extractelement <4 x i64> [[TMP23]], i32 1
+; VF4-NEXT:    store i64 [[TMP29]], ptr [[TMP25]], align 8
+; VF4-NEXT:    [[TMP30:%.*]] = extractelement <4 x i64> [[TMP23]], i32 2
+; VF4-NEXT:    store i64 [[TMP30]], ptr [[TMP26]], align 8
+; VF4-NEXT:    [[TMP31:%.*]] = extractelement <4 x i64> [[TMP23]], i32 3
+; VF4-NEXT:    store i64 [[TMP31]], ptr [[TMP27]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 8, i64 8, i64 8, i64 8>
+; VF4-NEXT:    [[VEC_IND_NEXT3]] = add <4 x i64> [[VEC_IND2]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[TMP32:%.*]] = icmp eq i64 [[INDEX_NEXT]], 500
+; VF4-NEXT:    br i1 [[TMP32]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 500, 500
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF4-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 500, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 2
+; VF4-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 2
+; VF4-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP11:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -408,66 +759,143 @@ exit:
 
 ; for (iv = 0, iv2 = 0 ; ; iv += 2, iv2 += 1) B[iv] = A[iv/3 + iv2/3] + 42;
 define void @ld_div3_step2_start0_ind2(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_div3_step2_start0_ind2
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 2>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 2
-; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 2
-; CHECK-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 3, i64 3>
-; CHECK-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 3, i64 3>
-; CHECK-NEXT:    [[TMP4:%.*]] = add <2 x i64> [[TMP2]], [[TMP3]]
-; CHECK-NEXT:    [[TMP5:%.*]] = extractelement <2 x i64> [[TMP4]], i32 0
-; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP5]]
-; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <2 x i64> [[TMP4]], i32 1
-; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP7]]
-; CHECK-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP6]], align 8
-; CHECK-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP8]], align 8
-; CHECK-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> poison, i64 [[TMP9]], i32 0
-; CHECK-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> [[TMP11]], i64 [[TMP10]], i32 1
-; CHECK-NEXT:    [[TMP13:%.*]] = add nsw <2 x i64> [[TMP12]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
-; CHECK-NEXT:    [[TMP16:%.*]] = extractelement <2 x i64> [[TMP13]], i32 0
-; CHECK-NEXT:    store i64 [[TMP16]], ptr [[TMP14]], align 8
-; CHECK-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP13]], i32 1
-; CHECK-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 4, i64 4>
-; CHECK-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], 500
-; CHECK-NEXT:    br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 500, 500
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 500, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 3
-; CHECK-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 3
-; CHECK-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP13:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_div3_step2_start0_ind2
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 2>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 2
+; VF2-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 2
+; VF2-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 3, i64 3>
+; VF2-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 3, i64 3>
+; VF2-NEXT:    [[TMP4:%.*]] = add <2 x i64> [[TMP2]], [[TMP3]]
+; VF2-NEXT:    [[TMP5:%.*]] = extractelement <2 x i64> [[TMP4]], i32 0
+; VF2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP5]]
+; VF2-NEXT:    [[TMP7:%.*]] = extractelement <2 x i64> [[TMP4]], i32 1
+; VF2-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP7]]
+; VF2-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP6]], align 8
+; VF2-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP8]], align 8
+; VF2-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> poison, i64 [[TMP9]], i32 0
+; VF2-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> [[TMP11]], i64 [[TMP10]], i32 1
+; VF2-NEXT:    [[TMP13:%.*]] = add nsw <2 x i64> [[TMP12]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF2-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF2-NEXT:    [[TMP16:%.*]] = extractelement <2 x i64> [[TMP13]], i32 0
+; VF2-NEXT:    store i64 [[TMP16]], ptr [[TMP14]], align 8
+; VF2-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP13]], i32 1
+; VF2-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 4, i64 4>
+; VF2-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], 500
+; VF2-NEXT:    br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 500, 500
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF2-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 500, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 3
+; VF2-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 3
+; VF2-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP13:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_div3_step2_start0_ind2
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 2, i64 4, i64 6>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND2:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 2
+; VF4-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 2
+; VF4-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 4
+; VF4-NEXT:    [[TMP3:%.*]] = add i64 [[OFFSET_IDX]], 6
+; VF4-NEXT:    [[TMP4:%.*]] = udiv <4 x i64> [[VEC_IND]], <i64 3, i64 3, i64 3, i64 3>
+; VF4-NEXT:    [[TMP5:%.*]] = udiv <4 x i64> [[VEC_IND2]], <i64 3, i64 3, i64 3, i64 3>
+; VF4-NEXT:    [[TMP6:%.*]] = add <4 x i64> [[TMP4]], [[TMP5]]
+; VF4-NEXT:    [[TMP7:%.*]] = extractelement <4 x i64> [[TMP6]], i32 0
+; VF4-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP7]]
+; VF4-NEXT:    [[TMP9:%.*]] = extractelement <4 x i64> [[TMP6]], i32 1
+; VF4-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP9]]
+; VF4-NEXT:    [[TMP11:%.*]] = extractelement <4 x i64> [[TMP6]], i32 2
+; VF4-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP11]]
+; VF4-NEXT:    [[TMP13:%.*]] = extractelement <4 x i64> [[TMP6]], i32 3
+; VF4-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP13]]
+; VF4-NEXT:    [[TMP15:%.*]] = load i64, ptr [[TMP8]], align 8
+; VF4-NEXT:    [[TMP16:%.*]] = load i64, ptr [[TMP10]], align 8
+; VF4-NEXT:    [[TMP17:%.*]] = load i64, ptr [[TMP12]], align 8
+; VF4-NEXT:    [[TMP18:%.*]] = load i64, ptr [[TMP14]], align 8
+; VF4-NEXT:    [[TMP19:%.*]] = insertelement <4 x i64> poison, i64 [[TMP15]], i32 0
+; VF4-NEXT:    [[TMP20:%.*]] = insertelement <4 x i64> [[TMP19]], i64 [[TMP16]], i32 1
+; VF4-NEXT:    [[TMP21:%.*]] = insertelement <4 x i64> [[TMP20]], i64 [[TMP17]], i32 2
+; VF4-NEXT:    [[TMP22:%.*]] = insertelement <4 x i64> [[TMP21]], i64 [[TMP18]], i32 3
+; VF4-NEXT:    [[TMP23:%.*]] = add nsw <4 x i64> [[TMP22]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP24:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF4-NEXT:    [[TMP25:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF4-NEXT:    [[TMP26:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF4-NEXT:    [[TMP27:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP3]]
+; VF4-NEXT:    [[TMP28:%.*]] = extractelement <4 x i64> [[TMP23]], i32 0
+; VF4-NEXT:    store i64 [[TMP28]], ptr [[TMP24]], align 8
+; VF4-NEXT:    [[TMP29:%.*]] = extractelement <4 x i64> [[TMP23]], i32 1
+; VF4-NEXT:    store i64 [[TMP29]], ptr [[TMP25]], align 8
+; VF4-NEXT:    [[TMP30:%.*]] = extractelement <4 x i64> [[TMP23]], i32 2
+; VF4-NEXT:    store i64 [[TMP30]], ptr [[TMP26]], align 8
+; VF4-NEXT:    [[TMP31:%.*]] = extractelement <4 x i64> [[TMP23]], i32 3
+; VF4-NEXT:    store i64 [[TMP31]], ptr [[TMP27]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 8, i64 8, i64 8, i64 8>
+; VF4-NEXT:    [[VEC_IND_NEXT3]] = add <4 x i64> [[VEC_IND2]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[TMP32:%.*]] = icmp eq i64 [[INDEX_NEXT]], 500
+; VF4-NEXT:    br i1 [[TMP32]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 500, 500
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 1000, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF4-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 500, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 3
+; VF4-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 3
+; VF4-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP13:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -492,66 +920,143 @@ exit:
 
 ; for (iv = 0, iv2 = 0 ; ; iv += 3, iv2 += 1) B[iv] = A[iv/1 + iv2/1] + 42;
 define void @ld_div1_step3_start0_ind2(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_div1_step3_start0_ind2
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 3
-; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 3
-; CHECK-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 1, i64 1>
-; CHECK-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 1, i64 1>
-; CHECK-NEXT:    [[TMP4:%.*]] = add <2 x i64> [[TMP2]], [[TMP3]]
-; CHECK-NEXT:    [[TMP5:%.*]] = extractelement <2 x i64> [[TMP4]], i32 0
-; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP5]]
-; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <2 x i64> [[TMP4]], i32 1
-; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP7]]
-; CHECK-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP6]], align 8
-; CHECK-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP8]], align 8
-; CHECK-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> poison, i64 [[TMP9]], i32 0
-; CHECK-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> [[TMP11]], i64 [[TMP10]], i32 1
-; CHECK-NEXT:    [[TMP13:%.*]] = add nsw <2 x i64> [[TMP12]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
-; CHECK-NEXT:    [[TMP16:%.*]] = extractelement <2 x i64> [[TMP13]], i32 0
-; CHECK-NEXT:    store i64 [[TMP16]], ptr [[TMP14]], align 8
-; CHECK-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP13]], i32 1
-; CHECK-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 6, i64 6>
-; CHECK-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
-; CHECK-NEXT:    br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 996, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 332, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 1
-; CHECK-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 1
-; CHECK-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP15:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_div1_step3_start0_ind2
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 3
+; VF2-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 3
+; VF2-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 1, i64 1>
+; VF2-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 1, i64 1>
+; VF2-NEXT:    [[TMP4:%.*]] = add <2 x i64> [[TMP2]], [[TMP3]]
+; VF2-NEXT:    [[TMP5:%.*]] = extractelement <2 x i64> [[TMP4]], i32 0
+; VF2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP5]]
+; VF2-NEXT:    [[TMP7:%.*]] = extractelement <2 x i64> [[TMP4]], i32 1
+; VF2-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP7]]
+; VF2-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP6]], align 8
+; VF2-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP8]], align 8
+; VF2-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> poison, i64 [[TMP9]], i32 0
+; VF2-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> [[TMP11]], i64 [[TMP10]], i32 1
+; VF2-NEXT:    [[TMP13:%.*]] = add nsw <2 x i64> [[TMP12]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF2-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF2-NEXT:    [[TMP16:%.*]] = extractelement <2 x i64> [[TMP13]], i32 0
+; VF2-NEXT:    store i64 [[TMP16]], ptr [[TMP14]], align 8
+; VF2-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP13]], i32 1
+; VF2-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 6, i64 6>
+; VF2-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
+; VF2-NEXT:    br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 996, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF2-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 332, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 1
+; VF2-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 1
+; VF2-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP15:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_div1_step3_start0_ind2
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 3, i64 6, i64 9>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND2:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 3
+; VF4-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 3
+; VF4-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 6
+; VF4-NEXT:    [[TMP3:%.*]] = add i64 [[OFFSET_IDX]], 9
+; VF4-NEXT:    [[TMP4:%.*]] = udiv <4 x i64> [[VEC_IND]], <i64 1, i64 1, i64 1, i64 1>
+; VF4-NEXT:    [[TMP5:%.*]] = udiv <4 x i64> [[VEC_IND2]], <i64 1, i64 1, i64 1, i64 1>
+; VF4-NEXT:    [[TMP6:%.*]] = add <4 x i64> [[TMP4]], [[TMP5]]
+; VF4-NEXT:    [[TMP7:%.*]] = extractelement <4 x i64> [[TMP6]], i32 0
+; VF4-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP7]]
+; VF4-NEXT:    [[TMP9:%.*]] = extractelement <4 x i64> [[TMP6]], i32 1
+; VF4-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP9]]
+; VF4-NEXT:    [[TMP11:%.*]] = extractelement <4 x i64> [[TMP6]], i32 2
+; VF4-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP11]]
+; VF4-NEXT:    [[TMP13:%.*]] = extractelement <4 x i64> [[TMP6]], i32 3
+; VF4-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP13]]
+; VF4-NEXT:    [[TMP15:%.*]] = load i64, ptr [[TMP8]], align 8
+; VF4-NEXT:    [[TMP16:%.*]] = load i64, ptr [[TMP10]], align 8
+; VF4-NEXT:    [[TMP17:%.*]] = load i64, ptr [[TMP12]], align 8
+; VF4-NEXT:    [[TMP18:%.*]] = load i64, ptr [[TMP14]], align 8
+; VF4-NEXT:    [[TMP19:%.*]] = insertelement <4 x i64> poison, i64 [[TMP15]], i32 0
+; VF4-NEXT:    [[TMP20:%.*]] = insertelement <4 x i64> [[TMP19]], i64 [[TMP16]], i32 1
+; VF4-NEXT:    [[TMP21:%.*]] = insertelement <4 x i64> [[TMP20]], i64 [[TMP17]], i32 2
+; VF4-NEXT:    [[TMP22:%.*]] = insertelement <4 x i64> [[TMP21]], i64 [[TMP18]], i32 3
+; VF4-NEXT:    [[TMP23:%.*]] = add nsw <4 x i64> [[TMP22]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP24:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF4-NEXT:    [[TMP25:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF4-NEXT:    [[TMP26:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF4-NEXT:    [[TMP27:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP3]]
+; VF4-NEXT:    [[TMP28:%.*]] = extractelement <4 x i64> [[TMP23]], i32 0
+; VF4-NEXT:    store i64 [[TMP28]], ptr [[TMP24]], align 8
+; VF4-NEXT:    [[TMP29:%.*]] = extractelement <4 x i64> [[TMP23]], i32 1
+; VF4-NEXT:    store i64 [[TMP29]], ptr [[TMP25]], align 8
+; VF4-NEXT:    [[TMP30:%.*]] = extractelement <4 x i64> [[TMP23]], i32 2
+; VF4-NEXT:    store i64 [[TMP30]], ptr [[TMP26]], align 8
+; VF4-NEXT:    [[TMP31:%.*]] = extractelement <4 x i64> [[TMP23]], i32 3
+; VF4-NEXT:    store i64 [[TMP31]], ptr [[TMP27]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 12, i64 12, i64 12, i64 12>
+; VF4-NEXT:    [[VEC_IND_NEXT3]] = add <4 x i64> [[VEC_IND2]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[TMP32:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
+; VF4-NEXT:    br i1 [[TMP32]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 996, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF4-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 332, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 1
+; VF4-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 1
+; VF4-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP15:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -576,66 +1081,143 @@ exit:
 
 ; for (iv = 0, iv2 = 0 ; ; iv += 3, iv2 += 1) B[iv] = A[iv/2 + iv2/2] + 42;
 define void @ld_div2_step3_start0_ind2(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_div2_step3_start0_ind2
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 3
-; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 3
-; CHECK-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP4:%.*]] = add <2 x i64> [[TMP2]], [[TMP3]]
-; CHECK-NEXT:    [[TMP5:%.*]] = extractelement <2 x i64> [[TMP4]], i32 0
-; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP5]]
-; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <2 x i64> [[TMP4]], i32 1
-; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP7]]
-; CHECK-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP6]], align 8
-; CHECK-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP8]], align 8
-; CHECK-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> poison, i64 [[TMP9]], i32 0
-; CHECK-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> [[TMP11]], i64 [[TMP10]], i32 1
-; CHECK-NEXT:    [[TMP13:%.*]] = add nsw <2 x i64> [[TMP12]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
-; CHECK-NEXT:    [[TMP16:%.*]] = extractelement <2 x i64> [[TMP13]], i32 0
-; CHECK-NEXT:    store i64 [[TMP16]], ptr [[TMP14]], align 8
-; CHECK-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP13]], i32 1
-; CHECK-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 6, i64 6>
-; CHECK-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
-; CHECK-NEXT:    br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 996, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 332, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 2
-; CHECK-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 2
-; CHECK-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP17:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_div2_step3_start0_ind2
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 3
+; VF2-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 3
+; VF2-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP4:%.*]] = add <2 x i64> [[TMP2]], [[TMP3]]
+; VF2-NEXT:    [[TMP5:%.*]] = extractelement <2 x i64> [[TMP4]], i32 0
+; VF2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP5]]
+; VF2-NEXT:    [[TMP7:%.*]] = extractelement <2 x i64> [[TMP4]], i32 1
+; VF2-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP7]]
+; VF2-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP6]], align 8
+; VF2-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP8]], align 8
+; VF2-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> poison, i64 [[TMP9]], i32 0
+; VF2-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> [[TMP11]], i64 [[TMP10]], i32 1
+; VF2-NEXT:    [[TMP13:%.*]] = add nsw <2 x i64> [[TMP12]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF2-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF2-NEXT:    [[TMP16:%.*]] = extractelement <2 x i64> [[TMP13]], i32 0
+; VF2-NEXT:    store i64 [[TMP16]], ptr [[TMP14]], align 8
+; VF2-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP13]], i32 1
+; VF2-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 6, i64 6>
+; VF2-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
+; VF2-NEXT:    br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 996, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF2-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 332, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 2
+; VF2-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 2
+; VF2-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP17:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_div2_step3_start0_ind2
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 3, i64 6, i64 9>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND2:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 3
+; VF4-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 3
+; VF4-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 6
+; VF4-NEXT:    [[TMP3:%.*]] = add i64 [[OFFSET_IDX]], 9
+; VF4-NEXT:    [[TMP4:%.*]] = udiv <4 x i64> [[VEC_IND]], <i64 2, i64 2, i64 2, i64 2>
+; VF4-NEXT:    [[TMP5:%.*]] = udiv <4 x i64> [[VEC_IND2]], <i64 2, i64 2, i64 2, i64 2>
+; VF4-NEXT:    [[TMP6:%.*]] = add <4 x i64> [[TMP4]], [[TMP5]]
+; VF4-NEXT:    [[TMP7:%.*]] = extractelement <4 x i64> [[TMP6]], i32 0
+; VF4-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP7]]
+; VF4-NEXT:    [[TMP9:%.*]] = extractelement <4 x i64> [[TMP6]], i32 1
+; VF4-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP9]]
+; VF4-NEXT:    [[TMP11:%.*]] = extractelement <4 x i64> [[TMP6]], i32 2
+; VF4-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP11]]
+; VF4-NEXT:    [[TMP13:%.*]] = extractelement <4 x i64> [[TMP6]], i32 3
+; VF4-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP13]]
+; VF4-NEXT:    [[TMP15:%.*]] = load i64, ptr [[TMP8]], align 8
+; VF4-NEXT:    [[TMP16:%.*]] = load i64, ptr [[TMP10]], align 8
+; VF4-NEXT:    [[TMP17:%.*]] = load i64, ptr [[TMP12]], align 8
+; VF4-NEXT:    [[TMP18:%.*]] = load i64, ptr [[TMP14]], align 8
+; VF4-NEXT:    [[TMP19:%.*]] = insertelement <4 x i64> poison, i64 [[TMP15]], i32 0
+; VF4-NEXT:    [[TMP20:%.*]] = insertelement <4 x i64> [[TMP19]], i64 [[TMP16]], i32 1
+; VF4-NEXT:    [[TMP21:%.*]] = insertelement <4 x i64> [[TMP20]], i64 [[TMP17]], i32 2
+; VF4-NEXT:    [[TMP22:%.*]] = insertelement <4 x i64> [[TMP21]], i64 [[TMP18]], i32 3
+; VF4-NEXT:    [[TMP23:%.*]] = add nsw <4 x i64> [[TMP22]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP24:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF4-NEXT:    [[TMP25:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF4-NEXT:    [[TMP26:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF4-NEXT:    [[TMP27:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP3]]
+; VF4-NEXT:    [[TMP28:%.*]] = extractelement <4 x i64> [[TMP23]], i32 0
+; VF4-NEXT:    store i64 [[TMP28]], ptr [[TMP24]], align 8
+; VF4-NEXT:    [[TMP29:%.*]] = extractelement <4 x i64> [[TMP23]], i32 1
+; VF4-NEXT:    store i64 [[TMP29]], ptr [[TMP25]], align 8
+; VF4-NEXT:    [[TMP30:%.*]] = extractelement <4 x i64> [[TMP23]], i32 2
+; VF4-NEXT:    store i64 [[TMP30]], ptr [[TMP26]], align 8
+; VF4-NEXT:    [[TMP31:%.*]] = extractelement <4 x i64> [[TMP23]], i32 3
+; VF4-NEXT:    store i64 [[TMP31]], ptr [[TMP27]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 12, i64 12, i64 12, i64 12>
+; VF4-NEXT:    [[VEC_IND_NEXT3]] = add <4 x i64> [[VEC_IND2]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[TMP32:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
+; VF4-NEXT:    br i1 [[TMP32]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 996, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF4-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 332, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 2
+; VF4-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 2
+; VF4-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP17:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -660,66 +1242,143 @@ exit:
 
 ; for (iv = 0, iv2 = 0 ; ; iv += 3, iv2 += 1) B[iv] = A[iv/3 + iv2/3] + 42;
 define void @ld_div3_step3_start0_ind2(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_div3_step3_start0_ind2
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 3
-; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 3
-; CHECK-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 3, i64 3>
-; CHECK-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 3, i64 3>
-; CHECK-NEXT:    [[TMP4:%.*]] = add <2 x i64> [[TMP2]], [[TMP3]]
-; CHECK-NEXT:    [[TMP5:%.*]] = extractelement <2 x i64> [[TMP4]], i32 0
-; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP5]]
-; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <2 x i64> [[TMP4]], i32 1
-; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP7]]
-; CHECK-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP6]], align 8
-; CHECK-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP8]], align 8
-; CHECK-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> poison, i64 [[TMP9]], i32 0
-; CHECK-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> [[TMP11]], i64 [[TMP10]], i32 1
-; CHECK-NEXT:    [[TMP13:%.*]] = add nsw <2 x i64> [[TMP12]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
-; CHECK-NEXT:    [[TMP16:%.*]] = extractelement <2 x i64> [[TMP13]], i32 0
-; CHECK-NEXT:    store i64 [[TMP16]], ptr [[TMP14]], align 8
-; CHECK-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP13]], i32 1
-; CHECK-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 6, i64 6>
-; CHECK-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
-; CHECK-NEXT:    br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 996, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 332, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 3
-; CHECK-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 3
-; CHECK-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP19:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_div3_step3_start0_ind2
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 0, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 3
+; VF2-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 3
+; VF2-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 3, i64 3>
+; VF2-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 3, i64 3>
+; VF2-NEXT:    [[TMP4:%.*]] = add <2 x i64> [[TMP2]], [[TMP3]]
+; VF2-NEXT:    [[TMP5:%.*]] = extractelement <2 x i64> [[TMP4]], i32 0
+; VF2-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP5]]
+; VF2-NEXT:    [[TMP7:%.*]] = extractelement <2 x i64> [[TMP4]], i32 1
+; VF2-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP7]]
+; VF2-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP6]], align 8
+; VF2-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP8]], align 8
+; VF2-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> poison, i64 [[TMP9]], i32 0
+; VF2-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> [[TMP11]], i64 [[TMP10]], i32 1
+; VF2-NEXT:    [[TMP13:%.*]] = add nsw <2 x i64> [[TMP12]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF2-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF2-NEXT:    [[TMP16:%.*]] = extractelement <2 x i64> [[TMP13]], i32 0
+; VF2-NEXT:    store i64 [[TMP16]], ptr [[TMP14]], align 8
+; VF2-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP13]], i32 1
+; VF2-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 6, i64 6>
+; VF2-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP18:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
+; VF2-NEXT:    br i1 [[TMP18]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 996, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF2-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 332, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 3
+; VF2-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 3
+; VF2-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP19:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_div3_step3_start0_ind2
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 3, i64 6, i64 9>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND2:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 3
+; VF4-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 3
+; VF4-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 6
+; VF4-NEXT:    [[TMP3:%.*]] = add i64 [[OFFSET_IDX]], 9
+; VF4-NEXT:    [[TMP4:%.*]] = udiv <4 x i64> [[VEC_IND]], <i64 3, i64 3, i64 3, i64 3>
+; VF4-NEXT:    [[TMP5:%.*]] = udiv <4 x i64> [[VEC_IND2]], <i64 3, i64 3, i64 3, i64 3>
+; VF4-NEXT:    [[TMP6:%.*]] = add <4 x i64> [[TMP4]], [[TMP5]]
+; VF4-NEXT:    [[TMP7:%.*]] = extractelement <4 x i64> [[TMP6]], i32 0
+; VF4-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP7]]
+; VF4-NEXT:    [[TMP9:%.*]] = extractelement <4 x i64> [[TMP6]], i32 1
+; VF4-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP9]]
+; VF4-NEXT:    [[TMP11:%.*]] = extractelement <4 x i64> [[TMP6]], i32 2
+; VF4-NEXT:    [[TMP12:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP11]]
+; VF4-NEXT:    [[TMP13:%.*]] = extractelement <4 x i64> [[TMP6]], i32 3
+; VF4-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP13]]
+; VF4-NEXT:    [[TMP15:%.*]] = load i64, ptr [[TMP8]], align 8
+; VF4-NEXT:    [[TMP16:%.*]] = load i64, ptr [[TMP10]], align 8
+; VF4-NEXT:    [[TMP17:%.*]] = load i64, ptr [[TMP12]], align 8
+; VF4-NEXT:    [[TMP18:%.*]] = load i64, ptr [[TMP14]], align 8
+; VF4-NEXT:    [[TMP19:%.*]] = insertelement <4 x i64> poison, i64 [[TMP15]], i32 0
+; VF4-NEXT:    [[TMP20:%.*]] = insertelement <4 x i64> [[TMP19]], i64 [[TMP16]], i32 1
+; VF4-NEXT:    [[TMP21:%.*]] = insertelement <4 x i64> [[TMP20]], i64 [[TMP17]], i32 2
+; VF4-NEXT:    [[TMP22:%.*]] = insertelement <4 x i64> [[TMP21]], i64 [[TMP18]], i32 3
+; VF4-NEXT:    [[TMP23:%.*]] = add nsw <4 x i64> [[TMP22]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP24:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF4-NEXT:    [[TMP25:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF4-NEXT:    [[TMP26:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF4-NEXT:    [[TMP27:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP3]]
+; VF4-NEXT:    [[TMP28:%.*]] = extractelement <4 x i64> [[TMP23]], i32 0
+; VF4-NEXT:    store i64 [[TMP28]], ptr [[TMP24]], align 8
+; VF4-NEXT:    [[TMP29:%.*]] = extractelement <4 x i64> [[TMP23]], i32 1
+; VF4-NEXT:    store i64 [[TMP29]], ptr [[TMP25]], align 8
+; VF4-NEXT:    [[TMP30:%.*]] = extractelement <4 x i64> [[TMP23]], i32 2
+; VF4-NEXT:    store i64 [[TMP30]], ptr [[TMP26]], align 8
+; VF4-NEXT:    [[TMP31:%.*]] = extractelement <4 x i64> [[TMP23]], i32 3
+; VF4-NEXT:    store i64 [[TMP31]], ptr [[TMP27]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 12, i64 12, i64 12, i64 12>
+; VF4-NEXT:    [[VEC_IND_NEXT3]] = add <4 x i64> [[VEC_IND2]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[TMP32:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
+; VF4-NEXT:    br i1 [[TMP32]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 996, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ]
+; VF4-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 332, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 3
+; VF4-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 3
+; VF4-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP19:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -744,62 +1403,127 @@ exit:
 
 ; for (iv = 1, iv2 = 0 ; ; iv += 1, iv2 += 1) B[iv] = A[iv/1 + iv2/1] + 42;
 define void @ld_div1_step1_start1_ind2(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_div1_step1_start1_ind2
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 2>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[INDEX]]
-; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP1:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 1, i64 1>
-; CHECK-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 1, i64 1>
-; CHECK-NEXT:    [[TMP3:%.*]] = add <2 x i64> [[TMP1]], [[TMP2]]
-; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP3]], i32 0
-; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP3]], i32 1
-; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
-; CHECK-NEXT:    [[TMP8:%.*]] = load i64, ptr [[TMP5]], align 8
-; CHECK-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP7]], align 8
-; CHECK-NEXT:    [[TMP10:%.*]] = insertelement <2 x i64> poison, i64 [[TMP8]], i32 0
-; CHECK-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> [[TMP10]], i64 [[TMP9]], i32 1
-; CHECK-NEXT:    [[TMP12:%.*]] = add nsw <2 x i64> [[TMP11]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[TMP13]], i32 0
-; CHECK-NEXT:    store <2 x i64> [[TMP12]], ptr [[TMP14]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 2, i64 2>
-; CHECK-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], 998
-; CHECK-NEXT:    br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP20:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 999, 998
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 999, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 998, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 1
-; CHECK-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 1
-; CHECK-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP21:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_div1_step1_start1_ind2
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 2>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[INDEX]]
+; VF2-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP1:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 1, i64 1>
+; VF2-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 1, i64 1>
+; VF2-NEXT:    [[TMP3:%.*]] = add <2 x i64> [[TMP1]], [[TMP2]]
+; VF2-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP3]], i32 0
+; VF2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
+; VF2-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP3]], i32 1
+; VF2-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF2-NEXT:    [[TMP8:%.*]] = load i64, ptr [[TMP5]], align 8
+; VF2-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF2-NEXT:    [[TMP10:%.*]] = insertelement <2 x i64> poison, i64 [[TMP8]], i32 0
+; VF2-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> [[TMP10]], i64 [[TMP9]], i32 1
+; VF2-NEXT:    [[TMP12:%.*]] = add nsw <2 x i64> [[TMP11]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[TMP13]], i32 0
+; VF2-NEXT:    store <2 x i64> [[TMP12]], ptr [[TMP14]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 2, i64 2>
+; VF2-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], 998
+; VF2-NEXT:    br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP20:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 999, 998
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 999, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF2-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 998, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 1
+; VF2-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 1
+; VF2-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP21:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_div1_step1_start1_ind2
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 1, i64 2, i64 3, i64 4>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND2:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[INDEX]]
+; VF4-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP1:%.*]] = udiv <4 x i64> [[VEC_IND]], <i64 1, i64 1, i64 1, i64 1>
+; VF4-NEXT:    [[TMP2:%.*]] = udiv <4 x i64> [[VEC_IND2]], <i64 1, i64 1, i64 1, i64 1>
+; VF4-NEXT:    [[TMP3:%.*]] = add <4 x i64> [[TMP1]], [[TMP2]]
+; VF4-NEXT:    [[TMP4:%.*]] = extractelement <4 x i64> [[TMP3]], i32 0
+; VF4-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
+; VF4-NEXT:    [[TMP6:%.*]] = extractelement <4 x i64> [[TMP3]], i32 1
+; VF4-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF4-NEXT:    [[TMP8:%.*]] = extractelement <4 x i64> [[TMP3]], i32 2
+; VF4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF4-NEXT:    [[TMP10:%.*]] = extractelement <4 x i64> [[TMP3]], i32 3
+; VF4-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP10]]
+; VF4-NEXT:    [[TMP12:%.*]] = load i64, ptr [[TMP5]], align 8
+; VF4-NEXT:    [[TMP13:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF4-NEXT:    [[TMP14:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF4-NEXT:    [[TMP15:%.*]] = load i64, ptr [[TMP11]], align 8
+; VF4-NEXT:    [[TMP16:%.*]] = insertelement <4 x i64> poison, i64 [[TMP12]], i32 0
+; VF4-NEXT:    [[TMP17:%.*]] = insertelement <4 x i64> [[TMP16]], i64 [[TMP13]], i32 1
+; VF4-NEXT:    [[TMP18:%.*]] = insertelement <4 x i64> [[TMP17]], i64 [[TMP14]], i32 2
+; VF4-NEXT:    [[TMP19:%.*]] = insertelement <4 x i64> [[TMP18]], i64 [[TMP15]], i32 3
+; VF4-NEXT:    [[TMP20:%.*]] = add nsw <4 x i64> [[TMP19]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP21:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF4-NEXT:    [[TMP22:%.*]] = getelementptr inbounds i64, ptr [[TMP21]], i32 0
+; VF4-NEXT:    store <4 x i64> [[TMP20]], ptr [[TMP22]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[VEC_IND_NEXT3]] = add <4 x i64> [[VEC_IND2]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[TMP23:%.*]] = icmp eq i64 [[INDEX_NEXT]], 996
+; VF4-NEXT:    br i1 [[TMP23]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP20:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 999, 996
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF4-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 996, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 1
+; VF4-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 1
+; VF4-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP21:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -824,62 +1548,127 @@ exit:
 
 ; for (iv = 1, iv2 = 0 ; ; iv += 1, iv2 += 1) B[iv] = A[iv/2 + iv2/2] + 42;
 define void @ld_div2_step1_start1_ind2(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_div2_step1_start1_ind2
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 2>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[INDEX]]
-; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP1:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP3:%.*]] = add <2 x i64> [[TMP1]], [[TMP2]]
-; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP3]], i32 0
-; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP3]], i32 1
-; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
-; CHECK-NEXT:    [[TMP8:%.*]] = load i64, ptr [[TMP5]], align 8
-; CHECK-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP7]], align 8
-; CHECK-NEXT:    [[TMP10:%.*]] = insertelement <2 x i64> poison, i64 [[TMP8]], i32 0
-; CHECK-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> [[TMP10]], i64 [[TMP9]], i32 1
-; CHECK-NEXT:    [[TMP12:%.*]] = add nsw <2 x i64> [[TMP11]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[TMP13]], i32 0
-; CHECK-NEXT:    store <2 x i64> [[TMP12]], ptr [[TMP14]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 2, i64 2>
-; CHECK-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], 998
-; CHECK-NEXT:    br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP22:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 999, 998
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 999, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 998, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 2
-; CHECK-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 2
-; CHECK-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP23:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_div2_step1_start1_ind2
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 2>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[INDEX]]
+; VF2-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP1:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP3:%.*]] = add <2 x i64> [[TMP1]], [[TMP2]]
+; VF2-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP3]], i32 0
+; VF2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
+; VF2-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP3]], i32 1
+; VF2-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF2-NEXT:    [[TMP8:%.*]] = load i64, ptr [[TMP5]], align 8
+; VF2-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF2-NEXT:    [[TMP10:%.*]] = insertelement <2 x i64> poison, i64 [[TMP8]], i32 0
+; VF2-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> [[TMP10]], i64 [[TMP9]], i32 1
+; VF2-NEXT:    [[TMP12:%.*]] = add nsw <2 x i64> [[TMP11]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[TMP13]], i32 0
+; VF2-NEXT:    store <2 x i64> [[TMP12]], ptr [[TMP14]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 2, i64 2>
+; VF2-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], 998
+; VF2-NEXT:    br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP22:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 999, 998
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 999, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF2-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 998, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 2
+; VF2-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 2
+; VF2-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP23:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_div2_step1_start1_ind2
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 1, i64 2, i64 3, i64 4>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND2:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[INDEX]]
+; VF4-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP1:%.*]] = udiv <4 x i64> [[VEC_IND]], <i64 2, i64 2, i64 2, i64 2>
+; VF4-NEXT:    [[TMP2:%.*]] = udiv <4 x i64> [[VEC_IND2]], <i64 2, i64 2, i64 2, i64 2>
+; VF4-NEXT:    [[TMP3:%.*]] = add <4 x i64> [[TMP1]], [[TMP2]]
+; VF4-NEXT:    [[TMP4:%.*]] = extractelement <4 x i64> [[TMP3]], i32 0
+; VF4-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
+; VF4-NEXT:    [[TMP6:%.*]] = extractelement <4 x i64> [[TMP3]], i32 1
+; VF4-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF4-NEXT:    [[TMP8:%.*]] = extractelement <4 x i64> [[TMP3]], i32 2
+; VF4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF4-NEXT:    [[TMP10:%.*]] = extractelement <4 x i64> [[TMP3]], i32 3
+; VF4-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP10]]
+; VF4-NEXT:    [[TMP12:%.*]] = load i64, ptr [[TMP5]], align 8
+; VF4-NEXT:    [[TMP13:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF4-NEXT:    [[TMP14:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF4-NEXT:    [[TMP15:%.*]] = load i64, ptr [[TMP11]], align 8
+; VF4-NEXT:    [[TMP16:%.*]] = insertelement <4 x i64> poison, i64 [[TMP12]], i32 0
+; VF4-NEXT:    [[TMP17:%.*]] = insertelement <4 x i64> [[TMP16]], i64 [[TMP13]], i32 1
+; VF4-NEXT:    [[TMP18:%.*]] = insertelement <4 x i64> [[TMP17]], i64 [[TMP14]], i32 2
+; VF4-NEXT:    [[TMP19:%.*]] = insertelement <4 x i64> [[TMP18]], i64 [[TMP15]], i32 3
+; VF4-NEXT:    [[TMP20:%.*]] = add nsw <4 x i64> [[TMP19]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP21:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF4-NEXT:    [[TMP22:%.*]] = getelementptr inbounds i64, ptr [[TMP21]], i32 0
+; VF4-NEXT:    store <4 x i64> [[TMP20]], ptr [[TMP22]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[VEC_IND_NEXT3]] = add <4 x i64> [[VEC_IND2]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[TMP23:%.*]] = icmp eq i64 [[INDEX_NEXT]], 996
+; VF4-NEXT:    br i1 [[TMP23]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP22:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 999, 996
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF4-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 996, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 2
+; VF4-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 2
+; VF4-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP23:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -904,62 +1693,127 @@ exit:
 
 ; for (iv = 1, iv2 = 0 ; ; iv += 1, iv2 += 1) B[iv] = A[iv/3 + iv2/3] + 42;
 define void @ld_div3_step1_start1_ind2(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_div3_step1_start1_ind2
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 2>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[INDEX]]
-; CHECK-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP1:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 3, i64 3>
-; CHECK-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 3, i64 3>
-; CHECK-NEXT:    [[TMP3:%.*]] = add <2 x i64> [[TMP1]], [[TMP2]]
-; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP3]], i32 0
-; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP3]], i32 1
-; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
-; CHECK-NEXT:    [[TMP8:%.*]] = load i64, ptr [[TMP5]], align 8
-; CHECK-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP7]], align 8
-; CHECK-NEXT:    [[TMP10:%.*]] = insertelement <2 x i64> poison, i64 [[TMP8]], i32 0
-; CHECK-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> [[TMP10]], i64 [[TMP9]], i32 1
-; CHECK-NEXT:    [[TMP12:%.*]] = add nsw <2 x i64> [[TMP11]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
-; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[TMP13]], i32 0
-; CHECK-NEXT:    store <2 x i64> [[TMP12]], ptr [[TMP14]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 2, i64 2>
-; CHECK-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], 998
-; CHECK-NEXT:    br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP24:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 999, 998
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 999, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 998, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 3
-; CHECK-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 3
-; CHECK-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP25:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_div3_step1_start1_ind2
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 2>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[INDEX]]
+; VF2-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP1:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 3, i64 3>
+; VF2-NEXT:    [[TMP2:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 3, i64 3>
+; VF2-NEXT:    [[TMP3:%.*]] = add <2 x i64> [[TMP1]], [[TMP2]]
+; VF2-NEXT:    [[TMP4:%.*]] = extractelement <2 x i64> [[TMP3]], i32 0
+; VF2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
+; VF2-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP3]], i32 1
+; VF2-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF2-NEXT:    [[TMP8:%.*]] = load i64, ptr [[TMP5]], align 8
+; VF2-NEXT:    [[TMP9:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF2-NEXT:    [[TMP10:%.*]] = insertelement <2 x i64> poison, i64 [[TMP8]], i32 0
+; VF2-NEXT:    [[TMP11:%.*]] = insertelement <2 x i64> [[TMP10]], i64 [[TMP9]], i32 1
+; VF2-NEXT:    [[TMP12:%.*]] = add nsw <2 x i64> [[TMP11]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF2-NEXT:    [[TMP14:%.*]] = getelementptr inbounds i64, ptr [[TMP13]], i32 0
+; VF2-NEXT:    store <2 x i64> [[TMP12]], ptr [[TMP14]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 2, i64 2>
+; VF2-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], 998
+; VF2-NEXT:    br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP24:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 999, 998
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 999, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF2-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 998, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 3
+; VF2-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 3
+; VF2-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP25:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_div3_step1_start1_ind2
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 1, i64 2, i64 3, i64 4>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND2:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[INDEX]]
+; VF4-NEXT:    [[TMP0:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP1:%.*]] = udiv <4 x i64> [[VEC_IND]], <i64 3, i64 3, i64 3, i64 3>
+; VF4-NEXT:    [[TMP2:%.*]] = udiv <4 x i64> [[VEC_IND2]], <i64 3, i64 3, i64 3, i64 3>
+; VF4-NEXT:    [[TMP3:%.*]] = add <4 x i64> [[TMP1]], [[TMP2]]
+; VF4-NEXT:    [[TMP4:%.*]] = extractelement <4 x i64> [[TMP3]], i32 0
+; VF4-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP4]]
+; VF4-NEXT:    [[TMP6:%.*]] = extractelement <4 x i64> [[TMP3]], i32 1
+; VF4-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF4-NEXT:    [[TMP8:%.*]] = extractelement <4 x i64> [[TMP3]], i32 2
+; VF4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF4-NEXT:    [[TMP10:%.*]] = extractelement <4 x i64> [[TMP3]], i32 3
+; VF4-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP10]]
+; VF4-NEXT:    [[TMP12:%.*]] = load i64, ptr [[TMP5]], align 8
+; VF4-NEXT:    [[TMP13:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF4-NEXT:    [[TMP14:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF4-NEXT:    [[TMP15:%.*]] = load i64, ptr [[TMP11]], align 8
+; VF4-NEXT:    [[TMP16:%.*]] = insertelement <4 x i64> poison, i64 [[TMP12]], i32 0
+; VF4-NEXT:    [[TMP17:%.*]] = insertelement <4 x i64> [[TMP16]], i64 [[TMP13]], i32 1
+; VF4-NEXT:    [[TMP18:%.*]] = insertelement <4 x i64> [[TMP17]], i64 [[TMP14]], i32 2
+; VF4-NEXT:    [[TMP19:%.*]] = insertelement <4 x i64> [[TMP18]], i64 [[TMP15]], i32 3
+; VF4-NEXT:    [[TMP20:%.*]] = add nsw <4 x i64> [[TMP19]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP21:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP0]]
+; VF4-NEXT:    [[TMP22:%.*]] = getelementptr inbounds i64, ptr [[TMP21]], i32 0
+; VF4-NEXT:    store <4 x i64> [[TMP20]], ptr [[TMP22]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[VEC_IND_NEXT3]] = add <4 x i64> [[VEC_IND2]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[TMP23:%.*]] = icmp eq i64 [[INDEX_NEXT]], 996
+; VF4-NEXT:    br i1 [[TMP23]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP24:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 999, 996
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF4-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 996, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 3
+; VF4-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 3
+; VF4-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 1
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP25:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -984,67 +1838,145 @@ exit:
 
 ; for (iv = 1, iv2 = 0 ; ; iv += 2, iv2 += 1) B[iv] = A[iv/1 + iv2/1] + 42;
 define void @ld_div1_step2_start1_ind2(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_div1_step2_start1_ind2
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 2
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
-; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 2
-; CHECK-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 1, i64 1>
-; CHECK-NEXT:    [[TMP4:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 1, i64 1>
-; CHECK-NEXT:    [[TMP5:%.*]] = add <2 x i64> [[TMP3]], [[TMP4]]
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP5]], i32 0
-; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
-; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <2 x i64> [[TMP5]], i32 1
-; CHECK-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
-; CHECK-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP7]], align 8
-; CHECK-NEXT:    [[TMP11:%.*]] = load i64, ptr [[TMP9]], align 8
-; CHECK-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> poison, i64 [[TMP10]], i32 0
-; CHECK-NEXT:    [[TMP13:%.*]] = insertelement <2 x i64> [[TMP12]], i64 [[TMP11]], i32 1
-; CHECK-NEXT:    [[TMP14:%.*]] = add nsw <2 x i64> [[TMP13]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
-; CHECK-NEXT:    [[TMP16:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
-; CHECK-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP14]], i32 0
-; CHECK-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
-; CHECK-NEXT:    [[TMP18:%.*]] = extractelement <2 x i64> [[TMP14]], i32 1
-; CHECK-NEXT:    store i64 [[TMP18]], ptr [[TMP16]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 4, i64 4>
-; CHECK-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP19:%.*]] = icmp eq i64 [[INDEX_NEXT]], 498
-; CHECK-NEXT:    br i1 [[TMP19]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP26:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 499, 498
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 498, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 1
-; CHECK-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 1
-; CHECK-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP27:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_div1_step2_start1_ind2
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 2
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
+; VF2-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 2
+; VF2-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 1, i64 1>
+; VF2-NEXT:    [[TMP4:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 1, i64 1>
+; VF2-NEXT:    [[TMP5:%.*]] = add <2 x i64> [[TMP3]], [[TMP4]]
+; VF2-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP5]], i32 0
+; VF2-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF2-NEXT:    [[TMP8:%.*]] = extractelement <2 x i64> [[TMP5]], i32 1
+; VF2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF2-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF2-NEXT:    [[TMP11:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF2-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> poison, i64 [[TMP10]], i32 0
+; VF2-NEXT:    [[TMP13:%.*]] = insertelement <2 x i64> [[TMP12]], i64 [[TMP11]], i32 1
+; VF2-NEXT:    [[TMP14:%.*]] = add nsw <2 x i64> [[TMP13]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF2-NEXT:    [[TMP16:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF2-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP14]], i32 0
+; VF2-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
+; VF2-NEXT:    [[TMP18:%.*]] = extractelement <2 x i64> [[TMP14]], i32 1
+; VF2-NEXT:    store i64 [[TMP18]], ptr [[TMP16]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 4, i64 4>
+; VF2-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP19:%.*]] = icmp eq i64 [[INDEX_NEXT]], 498
+; VF2-NEXT:    br i1 [[TMP19]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP26:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 499, 498
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF2-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 498, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 1
+; VF2-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 1
+; VF2-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP27:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_div1_step2_start1_ind2
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 1, i64 3, i64 5, i64 7>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND2:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 2
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
+; VF4-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 2
+; VF4-NEXT:    [[TMP3:%.*]] = add i64 [[OFFSET_IDX]], 4
+; VF4-NEXT:    [[TMP4:%.*]] = add i64 [[OFFSET_IDX]], 6
+; VF4-NEXT:    [[TMP5:%.*]] = udiv <4 x i64> [[VEC_IND]], <i64 1, i64 1, i64 1, i64 1>
+; VF4-NEXT:    [[TMP6:%.*]] = udiv <4 x i64> [[VEC_IND2]], <i64 1, i64 1, i64 1, i64 1>
+; VF4-NEXT:    [[TMP7:%.*]] = add <4 x i64> [[TMP5]], [[TMP6]]
+; VF4-NEXT:    [[TMP8:%.*]] = extractelement <4 x i64> [[TMP7]], i32 0
+; VF4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF4-NEXT:    [[TMP10:%.*]] = extractelement <4 x i64> [[TMP7]], i32 1
+; VF4-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP10]]
+; VF4-NEXT:    [[TMP12:%.*]] = extractelement <4 x i64> [[TMP7]], i32 2
+; VF4-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP12]]
+; VF4-NEXT:    [[TMP14:%.*]] = extractelement <4 x i64> [[TMP7]], i32 3
+; VF4-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP14]]
+; VF4-NEXT:    [[TMP16:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF4-NEXT:    [[TMP17:%.*]] = load i64, ptr [[TMP11]], align 8
+; VF4-NEXT:    [[TMP18:%.*]] = load i64, ptr [[TMP13]], align 8
+; VF4-NEXT:    [[TMP19:%.*]] = load i64, ptr [[TMP15]], align 8
+; VF4-NEXT:    [[TMP20:%.*]] = insertelement <4 x i64> poison, i64 [[TMP16]], i32 0
+; VF4-NEXT:    [[TMP21:%.*]] = insertelement <4 x i64> [[TMP20]], i64 [[TMP17]], i32 1
+; VF4-NEXT:    [[TMP22:%.*]] = insertelement <4 x i64> [[TMP21]], i64 [[TMP18]], i32 2
+; VF4-NEXT:    [[TMP23:%.*]] = insertelement <4 x i64> [[TMP22]], i64 [[TMP19]], i32 3
+; VF4-NEXT:    [[TMP24:%.*]] = add nsw <4 x i64> [[TMP23]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP25:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF4-NEXT:    [[TMP26:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF4-NEXT:    [[TMP27:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP3]]
+; VF4-NEXT:    [[TMP28:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP4]]
+; VF4-NEXT:    [[TMP29:%.*]] = extractelement <4 x i64> [[TMP24]], i32 0
+; VF4-NEXT:    store i64 [[TMP29]], ptr [[TMP25]], align 8
+; VF4-NEXT:    [[TMP30:%.*]] = extractelement <4 x i64> [[TMP24]], i32 1
+; VF4-NEXT:    store i64 [[TMP30]], ptr [[TMP26]], align 8
+; VF4-NEXT:    [[TMP31:%.*]] = extractelement <4 x i64> [[TMP24]], i32 2
+; VF4-NEXT:    store i64 [[TMP31]], ptr [[TMP27]], align 8
+; VF4-NEXT:    [[TMP32:%.*]] = extractelement <4 x i64> [[TMP24]], i32 3
+; VF4-NEXT:    store i64 [[TMP32]], ptr [[TMP28]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 8, i64 8, i64 8, i64 8>
+; VF4-NEXT:    [[VEC_IND_NEXT3]] = add <4 x i64> [[VEC_IND2]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[TMP33:%.*]] = icmp eq i64 [[INDEX_NEXT]], 496
+; VF4-NEXT:    br i1 [[TMP33]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP26:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 499, 496
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 993, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF4-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 496, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 1
+; VF4-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 1
+; VF4-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP27:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -1069,67 +2001,145 @@ exit:
 
 ; for (iv = 1, iv2 = 0 ; ; iv += 2, iv2 += 1) B[iv] = A[iv/2 + iv2/2] + 42;
 define void @ld_div2_step2_start1_ind2(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_div2_step2_start1_ind2
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 2
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
-; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 2
-; CHECK-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP4:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP5:%.*]] = add <2 x i64> [[TMP3]], [[TMP4]]
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP5]], i32 0
-; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
-; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <2 x i64> [[TMP5]], i32 1
-; CHECK-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
-; CHECK-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP7]], align 8
-; CHECK-NEXT:    [[TMP11:%.*]] = load i64, ptr [[TMP9]], align 8
-; CHECK-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> poison, i64 [[TMP10]], i32 0
-; CHECK-NEXT:    [[TMP13:%.*]] = insertelement <2 x i64> [[TMP12]], i64 [[TMP11]], i32 1
-; CHECK-NEXT:    [[TMP14:%.*]] = add nsw <2 x i64> [[TMP13]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
-; CHECK-NEXT:    [[TMP16:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
-; CHECK-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP14]], i32 0
-; CHECK-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
-; CHECK-NEXT:    [[TMP18:%.*]] = extractelement <2 x i64> [[TMP14]], i32 1
-; CHECK-NEXT:    store i64 [[TMP18]], ptr [[TMP16]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 4, i64 4>
-; CHECK-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP19:%.*]] = icmp eq i64 [[INDEX_NEXT]], 498
-; CHECK-NEXT:    br i1 [[TMP19]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP28:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 499, 498
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 498, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 2
-; CHECK-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 2
-; CHECK-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP29:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_div2_step2_start1_ind2
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 2
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
+; VF2-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 2
+; VF2-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP4:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP5:%.*]] = add <2 x i64> [[TMP3]], [[TMP4]]
+; VF2-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP5]], i32 0
+; VF2-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF2-NEXT:    [[TMP8:%.*]] = extractelement <2 x i64> [[TMP5]], i32 1
+; VF2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF2-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF2-NEXT:    [[TMP11:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF2-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> poison, i64 [[TMP10]], i32 0
+; VF2-NEXT:    [[TMP13:%.*]] = insertelement <2 x i64> [[TMP12]], i64 [[TMP11]], i32 1
+; VF2-NEXT:    [[TMP14:%.*]] = add nsw <2 x i64> [[TMP13]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF2-NEXT:    [[TMP16:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF2-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP14]], i32 0
+; VF2-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
+; VF2-NEXT:    [[TMP18:%.*]] = extractelement <2 x i64> [[TMP14]], i32 1
+; VF2-NEXT:    store i64 [[TMP18]], ptr [[TMP16]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 4, i64 4>
+; VF2-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP19:%.*]] = icmp eq i64 [[INDEX_NEXT]], 498
+; VF2-NEXT:    br i1 [[TMP19]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP28:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 499, 498
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF2-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 498, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 2
+; VF2-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 2
+; VF2-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP29:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_div2_step2_start1_ind2
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 1, i64 3, i64 5, i64 7>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND2:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 2
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
+; VF4-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 2
+; VF4-NEXT:    [[TMP3:%.*]] = add i64 [[OFFSET_IDX]], 4
+; VF4-NEXT:    [[TMP4:%.*]] = add i64 [[OFFSET_IDX]], 6
+; VF4-NEXT:    [[TMP5:%.*]] = udiv <4 x i64> [[VEC_IND]], <i64 2, i64 2, i64 2, i64 2>
+; VF4-NEXT:    [[TMP6:%.*]] = udiv <4 x i64> [[VEC_IND2]], <i64 2, i64 2, i64 2, i64 2>
+; VF4-NEXT:    [[TMP7:%.*]] = add <4 x i64> [[TMP5]], [[TMP6]]
+; VF4-NEXT:    [[TMP8:%.*]] = extractelement <4 x i64> [[TMP7]], i32 0
+; VF4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF4-NEXT:    [[TMP10:%.*]] = extractelement <4 x i64> [[TMP7]], i32 1
+; VF4-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP10]]
+; VF4-NEXT:    [[TMP12:%.*]] = extractelement <4 x i64> [[TMP7]], i32 2
+; VF4-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP12]]
+; VF4-NEXT:    [[TMP14:%.*]] = extractelement <4 x i64> [[TMP7]], i32 3
+; VF4-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP14]]
+; VF4-NEXT:    [[TMP16:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF4-NEXT:    [[TMP17:%.*]] = load i64, ptr [[TMP11]], align 8
+; VF4-NEXT:    [[TMP18:%.*]] = load i64, ptr [[TMP13]], align 8
+; VF4-NEXT:    [[TMP19:%.*]] = load i64, ptr [[TMP15]], align 8
+; VF4-NEXT:    [[TMP20:%.*]] = insertelement <4 x i64> poison, i64 [[TMP16]], i32 0
+; VF4-NEXT:    [[TMP21:%.*]] = insertelement <4 x i64> [[TMP20]], i64 [[TMP17]], i32 1
+; VF4-NEXT:    [[TMP22:%.*]] = insertelement <4 x i64> [[TMP21]], i64 [[TMP18]], i32 2
+; VF4-NEXT:    [[TMP23:%.*]] = insertelement <4 x i64> [[TMP22]], i64 [[TMP19]], i32 3
+; VF4-NEXT:    [[TMP24:%.*]] = add nsw <4 x i64> [[TMP23]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP25:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF4-NEXT:    [[TMP26:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF4-NEXT:    [[TMP27:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP3]]
+; VF4-NEXT:    [[TMP28:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP4]]
+; VF4-NEXT:    [[TMP29:%.*]] = extractelement <4 x i64> [[TMP24]], i32 0
+; VF4-NEXT:    store i64 [[TMP29]], ptr [[TMP25]], align 8
+; VF4-NEXT:    [[TMP30:%.*]] = extractelement <4 x i64> [[TMP24]], i32 1
+; VF4-NEXT:    store i64 [[TMP30]], ptr [[TMP26]], align 8
+; VF4-NEXT:    [[TMP31:%.*]] = extractelement <4 x i64> [[TMP24]], i32 2
+; VF4-NEXT:    store i64 [[TMP31]], ptr [[TMP27]], align 8
+; VF4-NEXT:    [[TMP32:%.*]] = extractelement <4 x i64> [[TMP24]], i32 3
+; VF4-NEXT:    store i64 [[TMP32]], ptr [[TMP28]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 8, i64 8, i64 8, i64 8>
+; VF4-NEXT:    [[VEC_IND_NEXT3]] = add <4 x i64> [[VEC_IND2]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[TMP33:%.*]] = icmp eq i64 [[INDEX_NEXT]], 496
+; VF4-NEXT:    br i1 [[TMP33]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP28:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 499, 496
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 993, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF4-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 496, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 2
+; VF4-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 2
+; VF4-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP29:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -1154,67 +2164,145 @@ exit:
 
 ; for (iv = 1, iv2 = 0 ; ; iv += 2, iv2 += 1) B[iv] = A[iv/3 + iv2/3] + 42;
 define void @ld_div3_step2_start1_ind2(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_div3_step2_start1_ind2
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 2
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
-; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 2
-; CHECK-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 3, i64 3>
-; CHECK-NEXT:    [[TMP4:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 3, i64 3>
-; CHECK-NEXT:    [[TMP5:%.*]] = add <2 x i64> [[TMP3]], [[TMP4]]
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP5]], i32 0
-; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
-; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <2 x i64> [[TMP5]], i32 1
-; CHECK-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
-; CHECK-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP7]], align 8
-; CHECK-NEXT:    [[TMP11:%.*]] = load i64, ptr [[TMP9]], align 8
-; CHECK-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> poison, i64 [[TMP10]], i32 0
-; CHECK-NEXT:    [[TMP13:%.*]] = insertelement <2 x i64> [[TMP12]], i64 [[TMP11]], i32 1
-; CHECK-NEXT:    [[TMP14:%.*]] = add nsw <2 x i64> [[TMP13]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
-; CHECK-NEXT:    [[TMP16:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
-; CHECK-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP14]], i32 0
-; CHECK-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
-; CHECK-NEXT:    [[TMP18:%.*]] = extractelement <2 x i64> [[TMP14]], i32 1
-; CHECK-NEXT:    store i64 [[TMP18]], ptr [[TMP16]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 4, i64 4>
-; CHECK-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP19:%.*]] = icmp eq i64 [[INDEX_NEXT]], 498
-; CHECK-NEXT:    br i1 [[TMP19]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP30:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 499, 498
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 498, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 3
-; CHECK-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 3
-; CHECK-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP31:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_div3_step2_start1_ind2
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 2
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
+; VF2-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 2
+; VF2-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 3, i64 3>
+; VF2-NEXT:    [[TMP4:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 3, i64 3>
+; VF2-NEXT:    [[TMP5:%.*]] = add <2 x i64> [[TMP3]], [[TMP4]]
+; VF2-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP5]], i32 0
+; VF2-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF2-NEXT:    [[TMP8:%.*]] = extractelement <2 x i64> [[TMP5]], i32 1
+; VF2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF2-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF2-NEXT:    [[TMP11:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF2-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> poison, i64 [[TMP10]], i32 0
+; VF2-NEXT:    [[TMP13:%.*]] = insertelement <2 x i64> [[TMP12]], i64 [[TMP11]], i32 1
+; VF2-NEXT:    [[TMP14:%.*]] = add nsw <2 x i64> [[TMP13]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF2-NEXT:    [[TMP16:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF2-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP14]], i32 0
+; VF2-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
+; VF2-NEXT:    [[TMP18:%.*]] = extractelement <2 x i64> [[TMP14]], i32 1
+; VF2-NEXT:    store i64 [[TMP18]], ptr [[TMP16]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 4, i64 4>
+; VF2-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP19:%.*]] = icmp eq i64 [[INDEX_NEXT]], 498
+; VF2-NEXT:    br i1 [[TMP19]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP30:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 499, 498
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF2-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 498, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 3
+; VF2-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 3
+; VF2-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP31:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_div3_step2_start1_ind2
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 1, i64 3, i64 5, i64 7>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND2:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 2
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
+; VF4-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 2
+; VF4-NEXT:    [[TMP3:%.*]] = add i64 [[OFFSET_IDX]], 4
+; VF4-NEXT:    [[TMP4:%.*]] = add i64 [[OFFSET_IDX]], 6
+; VF4-NEXT:    [[TMP5:%.*]] = udiv <4 x i64> [[VEC_IND]], <i64 3, i64 3, i64 3, i64 3>
+; VF4-NEXT:    [[TMP6:%.*]] = udiv <4 x i64> [[VEC_IND2]], <i64 3, i64 3, i64 3, i64 3>
+; VF4-NEXT:    [[TMP7:%.*]] = add <4 x i64> [[TMP5]], [[TMP6]]
+; VF4-NEXT:    [[TMP8:%.*]] = extractelement <4 x i64> [[TMP7]], i32 0
+; VF4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF4-NEXT:    [[TMP10:%.*]] = extractelement <4 x i64> [[TMP7]], i32 1
+; VF4-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP10]]
+; VF4-NEXT:    [[TMP12:%.*]] = extractelement <4 x i64> [[TMP7]], i32 2
+; VF4-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP12]]
+; VF4-NEXT:    [[TMP14:%.*]] = extractelement <4 x i64> [[TMP7]], i32 3
+; VF4-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP14]]
+; VF4-NEXT:    [[TMP16:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF4-NEXT:    [[TMP17:%.*]] = load i64, ptr [[TMP11]], align 8
+; VF4-NEXT:    [[TMP18:%.*]] = load i64, ptr [[TMP13]], align 8
+; VF4-NEXT:    [[TMP19:%.*]] = load i64, ptr [[TMP15]], align 8
+; VF4-NEXT:    [[TMP20:%.*]] = insertelement <4 x i64> poison, i64 [[TMP16]], i32 0
+; VF4-NEXT:    [[TMP21:%.*]] = insertelement <4 x i64> [[TMP20]], i64 [[TMP17]], i32 1
+; VF4-NEXT:    [[TMP22:%.*]] = insertelement <4 x i64> [[TMP21]], i64 [[TMP18]], i32 2
+; VF4-NEXT:    [[TMP23:%.*]] = insertelement <4 x i64> [[TMP22]], i64 [[TMP19]], i32 3
+; VF4-NEXT:    [[TMP24:%.*]] = add nsw <4 x i64> [[TMP23]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP25:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF4-NEXT:    [[TMP26:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF4-NEXT:    [[TMP27:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP3]]
+; VF4-NEXT:    [[TMP28:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP4]]
+; VF4-NEXT:    [[TMP29:%.*]] = extractelement <4 x i64> [[TMP24]], i32 0
+; VF4-NEXT:    store i64 [[TMP29]], ptr [[TMP25]], align 8
+; VF4-NEXT:    [[TMP30:%.*]] = extractelement <4 x i64> [[TMP24]], i32 1
+; VF4-NEXT:    store i64 [[TMP30]], ptr [[TMP26]], align 8
+; VF4-NEXT:    [[TMP31:%.*]] = extractelement <4 x i64> [[TMP24]], i32 2
+; VF4-NEXT:    store i64 [[TMP31]], ptr [[TMP27]], align 8
+; VF4-NEXT:    [[TMP32:%.*]] = extractelement <4 x i64> [[TMP24]], i32 3
+; VF4-NEXT:    store i64 [[TMP32]], ptr [[TMP28]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 8, i64 8, i64 8, i64 8>
+; VF4-NEXT:    [[VEC_IND_NEXT3]] = add <4 x i64> [[VEC_IND2]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[TMP33:%.*]] = icmp eq i64 [[INDEX_NEXT]], 496
+; VF4-NEXT:    br i1 [[TMP33]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP30:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 499, 496
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 993, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF4-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 496, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 3
+; VF4-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 3
+; VF4-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 2
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP31:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -1239,67 +2327,145 @@ exit:
 
 ; for (iv = 1, iv2 = 0 ; ; iv += 3, iv2 += 1) B[iv] = A[iv/1 + iv2/1] + 42;
 define void @ld_div1_step3_start1_ind2(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_div1_step3_start1_ind2
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 4>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 3
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
-; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 3
-; CHECK-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 1, i64 1>
-; CHECK-NEXT:    [[TMP4:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 1, i64 1>
-; CHECK-NEXT:    [[TMP5:%.*]] = add <2 x i64> [[TMP3]], [[TMP4]]
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP5]], i32 0
-; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
-; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <2 x i64> [[TMP5]], i32 1
-; CHECK-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
-; CHECK-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP7]], align 8
-; CHECK-NEXT:    [[TMP11:%.*]] = load i64, ptr [[TMP9]], align 8
-; CHECK-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> poison, i64 [[TMP10]], i32 0
-; CHECK-NEXT:    [[TMP13:%.*]] = insertelement <2 x i64> [[TMP12]], i64 [[TMP11]], i32 1
-; CHECK-NEXT:    [[TMP14:%.*]] = add nsw <2 x i64> [[TMP13]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
-; CHECK-NEXT:    [[TMP16:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
-; CHECK-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP14]], i32 0
-; CHECK-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
-; CHECK-NEXT:    [[TMP18:%.*]] = extractelement <2 x i64> [[TMP14]], i32 1
-; CHECK-NEXT:    store i64 [[TMP18]], ptr [[TMP16]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 6, i64 6>
-; CHECK-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP19:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
-; CHECK-NEXT:    br i1 [[TMP19]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP32:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 332, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 1
-; CHECK-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 1
-; CHECK-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP33:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_div1_step3_start1_ind2
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 4>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 3
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
+; VF2-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 3
+; VF2-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 1, i64 1>
+; VF2-NEXT:    [[TMP4:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 1, i64 1>
+; VF2-NEXT:    [[TMP5:%.*]] = add <2 x i64> [[TMP3]], [[TMP4]]
+; VF2-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP5]], i32 0
+; VF2-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF2-NEXT:    [[TMP8:%.*]] = extractelement <2 x i64> [[TMP5]], i32 1
+; VF2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF2-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF2-NEXT:    [[TMP11:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF2-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> poison, i64 [[TMP10]], i32 0
+; VF2-NEXT:    [[TMP13:%.*]] = insertelement <2 x i64> [[TMP12]], i64 [[TMP11]], i32 1
+; VF2-NEXT:    [[TMP14:%.*]] = add nsw <2 x i64> [[TMP13]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF2-NEXT:    [[TMP16:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF2-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP14]], i32 0
+; VF2-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
+; VF2-NEXT:    [[TMP18:%.*]] = extractelement <2 x i64> [[TMP14]], i32 1
+; VF2-NEXT:    store i64 [[TMP18]], ptr [[TMP16]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 6, i64 6>
+; VF2-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP19:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
+; VF2-NEXT:    br i1 [[TMP19]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP32:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF2-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 332, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 1
+; VF2-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 1
+; VF2-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP33:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_div1_step3_start1_ind2
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 1, i64 4, i64 7, i64 10>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND2:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 3
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
+; VF4-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 3
+; VF4-NEXT:    [[TMP3:%.*]] = add i64 [[OFFSET_IDX]], 6
+; VF4-NEXT:    [[TMP4:%.*]] = add i64 [[OFFSET_IDX]], 9
+; VF4-NEXT:    [[TMP5:%.*]] = udiv <4 x i64> [[VEC_IND]], <i64 1, i64 1, i64 1, i64 1>
+; VF4-NEXT:    [[TMP6:%.*]] = udiv <4 x i64> [[VEC_IND2]], <i64 1, i64 1, i64 1, i64 1>
+; VF4-NEXT:    [[TMP7:%.*]] = add <4 x i64> [[TMP5]], [[TMP6]]
+; VF4-NEXT:    [[TMP8:%.*]] = extractelement <4 x i64> [[TMP7]], i32 0
+; VF4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF4-NEXT:    [[TMP10:%.*]] = extractelement <4 x i64> [[TMP7]], i32 1
+; VF4-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP10]]
+; VF4-NEXT:    [[TMP12:%.*]] = extractelement <4 x i64> [[TMP7]], i32 2
+; VF4-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP12]]
+; VF4-NEXT:    [[TMP14:%.*]] = extractelement <4 x i64> [[TMP7]], i32 3
+; VF4-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP14]]
+; VF4-NEXT:    [[TMP16:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF4-NEXT:    [[TMP17:%.*]] = load i64, ptr [[TMP11]], align 8
+; VF4-NEXT:    [[TMP18:%.*]] = load i64, ptr [[TMP13]], align 8
+; VF4-NEXT:    [[TMP19:%.*]] = load i64, ptr [[TMP15]], align 8
+; VF4-NEXT:    [[TMP20:%.*]] = insertelement <4 x i64> poison, i64 [[TMP16]], i32 0
+; VF4-NEXT:    [[TMP21:%.*]] = insertelement <4 x i64> [[TMP20]], i64 [[TMP17]], i32 1
+; VF4-NEXT:    [[TMP22:%.*]] = insertelement <4 x i64> [[TMP21]], i64 [[TMP18]], i32 2
+; VF4-NEXT:    [[TMP23:%.*]] = insertelement <4 x i64> [[TMP22]], i64 [[TMP19]], i32 3
+; VF4-NEXT:    [[TMP24:%.*]] = add nsw <4 x i64> [[TMP23]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP25:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF4-NEXT:    [[TMP26:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF4-NEXT:    [[TMP27:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP3]]
+; VF4-NEXT:    [[TMP28:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP4]]
+; VF4-NEXT:    [[TMP29:%.*]] = extractelement <4 x i64> [[TMP24]], i32 0
+; VF4-NEXT:    store i64 [[TMP29]], ptr [[TMP25]], align 8
+; VF4-NEXT:    [[TMP30:%.*]] = extractelement <4 x i64> [[TMP24]], i32 1
+; VF4-NEXT:    store i64 [[TMP30]], ptr [[TMP26]], align 8
+; VF4-NEXT:    [[TMP31:%.*]] = extractelement <4 x i64> [[TMP24]], i32 2
+; VF4-NEXT:    store i64 [[TMP31]], ptr [[TMP27]], align 8
+; VF4-NEXT:    [[TMP32:%.*]] = extractelement <4 x i64> [[TMP24]], i32 3
+; VF4-NEXT:    store i64 [[TMP32]], ptr [[TMP28]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 12, i64 12, i64 12, i64 12>
+; VF4-NEXT:    [[VEC_IND_NEXT3]] = add <4 x i64> [[VEC_IND2]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[TMP33:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
+; VF4-NEXT:    br i1 [[TMP33]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP32:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF4-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 332, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 1
+; VF4-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 1
+; VF4-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP33:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -1324,67 +2490,145 @@ exit:
 
 ; for (iv = 1, iv2 = 0 ; ; iv += 3, iv2 += 1) B[iv] = A[iv/2 + iv2/2] + 42;
 define void @ld_div2_step3_start1_ind2(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_div2_step3_start1_ind2
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 4>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 3
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
-; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 3
-; CHECK-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP4:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP5:%.*]] = add <2 x i64> [[TMP3]], [[TMP4]]
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP5]], i32 0
-; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
-; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <2 x i64> [[TMP5]], i32 1
-; CHECK-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
-; CHECK-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP7]], align 8
-; CHECK-NEXT:    [[TMP11:%.*]] = load i64, ptr [[TMP9]], align 8
-; CHECK-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> poison, i64 [[TMP10]], i32 0
-; CHECK-NEXT:    [[TMP13:%.*]] = insertelement <2 x i64> [[TMP12]], i64 [[TMP11]], i32 1
-; CHECK-NEXT:    [[TMP14:%.*]] = add nsw <2 x i64> [[TMP13]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
-; CHECK-NEXT:    [[TMP16:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
-; CHECK-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP14]], i32 0
-; CHECK-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
-; CHECK-NEXT:    [[TMP18:%.*]] = extractelement <2 x i64> [[TMP14]], i32 1
-; CHECK-NEXT:    store i64 [[TMP18]], ptr [[TMP16]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 6, i64 6>
-; CHECK-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP19:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
-; CHECK-NEXT:    br i1 [[TMP19]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP34:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 332, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 2
-; CHECK-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 2
-; CHECK-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP35:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_div2_step3_start1_ind2
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 4>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 3
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
+; VF2-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 3
+; VF2-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP4:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP5:%.*]] = add <2 x i64> [[TMP3]], [[TMP4]]
+; VF2-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP5]], i32 0
+; VF2-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF2-NEXT:    [[TMP8:%.*]] = extractelement <2 x i64> [[TMP5]], i32 1
+; VF2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF2-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF2-NEXT:    [[TMP11:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF2-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> poison, i64 [[TMP10]], i32 0
+; VF2-NEXT:    [[TMP13:%.*]] = insertelement <2 x i64> [[TMP12]], i64 [[TMP11]], i32 1
+; VF2-NEXT:    [[TMP14:%.*]] = add nsw <2 x i64> [[TMP13]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF2-NEXT:    [[TMP16:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF2-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP14]], i32 0
+; VF2-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
+; VF2-NEXT:    [[TMP18:%.*]] = extractelement <2 x i64> [[TMP14]], i32 1
+; VF2-NEXT:    store i64 [[TMP18]], ptr [[TMP16]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 6, i64 6>
+; VF2-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP19:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
+; VF2-NEXT:    br i1 [[TMP19]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP34:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF2-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 332, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 2
+; VF2-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 2
+; VF2-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP35:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_div2_step3_start1_ind2
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 1, i64 4, i64 7, i64 10>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND2:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 3
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
+; VF4-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 3
+; VF4-NEXT:    [[TMP3:%.*]] = add i64 [[OFFSET_IDX]], 6
+; VF4-NEXT:    [[TMP4:%.*]] = add i64 [[OFFSET_IDX]], 9
+; VF4-NEXT:    [[TMP5:%.*]] = udiv <4 x i64> [[VEC_IND]], <i64 2, i64 2, i64 2, i64 2>
+; VF4-NEXT:    [[TMP6:%.*]] = udiv <4 x i64> [[VEC_IND2]], <i64 2, i64 2, i64 2, i64 2>
+; VF4-NEXT:    [[TMP7:%.*]] = add <4 x i64> [[TMP5]], [[TMP6]]
+; VF4-NEXT:    [[TMP8:%.*]] = extractelement <4 x i64> [[TMP7]], i32 0
+; VF4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF4-NEXT:    [[TMP10:%.*]] = extractelement <4 x i64> [[TMP7]], i32 1
+; VF4-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP10]]
+; VF4-NEXT:    [[TMP12:%.*]] = extractelement <4 x i64> [[TMP7]], i32 2
+; VF4-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP12]]
+; VF4-NEXT:    [[TMP14:%.*]] = extractelement <4 x i64> [[TMP7]], i32 3
+; VF4-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP14]]
+; VF4-NEXT:    [[TMP16:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF4-NEXT:    [[TMP17:%.*]] = load i64, ptr [[TMP11]], align 8
+; VF4-NEXT:    [[TMP18:%.*]] = load i64, ptr [[TMP13]], align 8
+; VF4-NEXT:    [[TMP19:%.*]] = load i64, ptr [[TMP15]], align 8
+; VF4-NEXT:    [[TMP20:%.*]] = insertelement <4 x i64> poison, i64 [[TMP16]], i32 0
+; VF4-NEXT:    [[TMP21:%.*]] = insertelement <4 x i64> [[TMP20]], i64 [[TMP17]], i32 1
+; VF4-NEXT:    [[TMP22:%.*]] = insertelement <4 x i64> [[TMP21]], i64 [[TMP18]], i32 2
+; VF4-NEXT:    [[TMP23:%.*]] = insertelement <4 x i64> [[TMP22]], i64 [[TMP19]], i32 3
+; VF4-NEXT:    [[TMP24:%.*]] = add nsw <4 x i64> [[TMP23]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP25:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF4-NEXT:    [[TMP26:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF4-NEXT:    [[TMP27:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP3]]
+; VF4-NEXT:    [[TMP28:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP4]]
+; VF4-NEXT:    [[TMP29:%.*]] = extractelement <4 x i64> [[TMP24]], i32 0
+; VF4-NEXT:    store i64 [[TMP29]], ptr [[TMP25]], align 8
+; VF4-NEXT:    [[TMP30:%.*]] = extractelement <4 x i64> [[TMP24]], i32 1
+; VF4-NEXT:    store i64 [[TMP30]], ptr [[TMP26]], align 8
+; VF4-NEXT:    [[TMP31:%.*]] = extractelement <4 x i64> [[TMP24]], i32 2
+; VF4-NEXT:    store i64 [[TMP31]], ptr [[TMP27]], align 8
+; VF4-NEXT:    [[TMP32:%.*]] = extractelement <4 x i64> [[TMP24]], i32 3
+; VF4-NEXT:    store i64 [[TMP32]], ptr [[TMP28]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 12, i64 12, i64 12, i64 12>
+; VF4-NEXT:    [[VEC_IND_NEXT3]] = add <4 x i64> [[VEC_IND2]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[TMP33:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
+; VF4-NEXT:    br i1 [[TMP33]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP34:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF4-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 332, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 2
+; VF4-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 2
+; VF4-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP35:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop
@@ -1409,67 +2653,145 @@ exit:
 
 ; for (iv = 1, iv2 = 0 ; ; iv += 3, iv2 += 1) B[iv] = A[iv/3 + iv2/3] + 42;
 define void @ld_div3_step3_start1_ind2(ptr noalias %A, ptr noalias %B) {
-; CHECK-LABEL: define void @ld_div3_step3_start1_ind2
-; CHECK-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
-; 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 i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 4>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
-; CHECK-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 3
-; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
-; CHECK-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
-; CHECK-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 3
-; CHECK-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 3, i64 3>
-; CHECK-NEXT:    [[TMP4:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 3, i64 3>
-; CHECK-NEXT:    [[TMP5:%.*]] = add <2 x i64> [[TMP3]], [[TMP4]]
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP5]], i32 0
-; CHECK-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
-; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <2 x i64> [[TMP5]], i32 1
-; CHECK-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
-; CHECK-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP7]], align 8
-; CHECK-NEXT:    [[TMP11:%.*]] = load i64, ptr [[TMP9]], align 8
-; CHECK-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> poison, i64 [[TMP10]], i32 0
-; CHECK-NEXT:    [[TMP13:%.*]] = insertelement <2 x i64> [[TMP12]], i64 [[TMP11]], i32 1
-; CHECK-NEXT:    [[TMP14:%.*]] = add nsw <2 x i64> [[TMP13]], <i64 42, i64 42>
-; CHECK-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
-; CHECK-NEXT:    [[TMP16:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
-; CHECK-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP14]], i32 0
-; CHECK-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
-; CHECK-NEXT:    [[TMP18:%.*]] = extractelement <2 x i64> [[TMP14]], i32 1
-; CHECK-NEXT:    store i64 [[TMP18]], ptr [[TMP16]], align 8
-; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
-; CHECK-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 6, i64 6>
-; CHECK-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
-; CHECK-NEXT:    [[TMP19:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
-; CHECK-NEXT:    br i1 [[TMP19]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP36:![0-9]+]]
-; CHECK:       middle.block:
-; CHECK-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
-; CHECK-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
-; CHECK:       scalar.ph:
-; CHECK-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
-; CHECK-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 332, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
-; CHECK-NEXT:    br label [[LOOP:%.*]]
-; CHECK:       loop:
-; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
-; CHECK-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 3
-; CHECK-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 3
-; CHECK-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
-; CHECK-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
-; CHECK-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
-; CHECK-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
-; CHECK-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
-; CHECK-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
-; CHECK-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
-; CHECK-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
-; CHECK-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
-; CHECK-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP37:![0-9]+]]
-; CHECK:       exit:
-; CHECK-NEXT:    ret void
+; VF2-LABEL: define void @ld_div3_step3_start1_ind2
+; VF2-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF2-NEXT:  entry:
+; VF2-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF2:       vector.ph:
+; VF2-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF2:       vector.body:
+; VF2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND:%.*]] = phi <2 x i64> [ <i64 1, i64 4>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[VEC_IND2:%.*]] = phi <2 x i64> [ <i64 0, i64 1>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF2-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 3
+; VF2-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
+; VF2-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF2-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 3
+; VF2-NEXT:    [[TMP3:%.*]] = udiv <2 x i64> [[VEC_IND]], <i64 3, i64 3>
+; VF2-NEXT:    [[TMP4:%.*]] = udiv <2 x i64> [[VEC_IND2]], <i64 3, i64 3>
+; VF2-NEXT:    [[TMP5:%.*]] = add <2 x i64> [[TMP3]], [[TMP4]]
+; VF2-NEXT:    [[TMP6:%.*]] = extractelement <2 x i64> [[TMP5]], i32 0
+; VF2-NEXT:    [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP6]]
+; VF2-NEXT:    [[TMP8:%.*]] = extractelement <2 x i64> [[TMP5]], i32 1
+; VF2-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF2-NEXT:    [[TMP10:%.*]] = load i64, ptr [[TMP7]], align 8
+; VF2-NEXT:    [[TMP11:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF2-NEXT:    [[TMP12:%.*]] = insertelement <2 x i64> poison, i64 [[TMP10]], i32 0
+; VF2-NEXT:    [[TMP13:%.*]] = insertelement <2 x i64> [[TMP12]], i64 [[TMP11]], i32 1
+; VF2-NEXT:    [[TMP14:%.*]] = add nsw <2 x i64> [[TMP13]], <i64 42, i64 42>
+; VF2-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF2-NEXT:    [[TMP16:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF2-NEXT:    [[TMP17:%.*]] = extractelement <2 x i64> [[TMP14]], i32 0
+; VF2-NEXT:    store i64 [[TMP17]], ptr [[TMP15]], align 8
+; VF2-NEXT:    [[TMP18:%.*]] = extractelement <2 x i64> [[TMP14]], i32 1
+; VF2-NEXT:    store i64 [[TMP18]], ptr [[TMP16]], align 8
+; VF2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 2
+; VF2-NEXT:    [[VEC_IND_NEXT]] = add <2 x i64> [[VEC_IND]], <i64 6, i64 6>
+; VF2-NEXT:    [[VEC_IND_NEXT3]] = add <2 x i64> [[VEC_IND2]], <i64 2, i64 2>
+; VF2-NEXT:    [[TMP19:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
+; VF2-NEXT:    br i1 [[TMP19]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP36:![0-9]+]]
+; VF2:       middle.block:
+; VF2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
+; VF2-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF2:       scalar.ph:
+; VF2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF2-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 332, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF2-NEXT:    br label [[LOOP:%.*]]
+; VF2:       loop:
+; VF2-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF2-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 3
+; VF2-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 3
+; VF2-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF2-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF2-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF2-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF2-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF2-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF2-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF2-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
+; VF2-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF2-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP37:![0-9]+]]
+; VF2:       exit:
+; VF2-NEXT:    ret void
+;
+; VF4-LABEL: define void @ld_div3_step3_start1_ind2
+; VF4-SAME: (ptr noalias [[A:%.*]], ptr noalias [[B:%.*]]) {
+; VF4-NEXT:  entry:
+; VF4-NEXT:    br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
+; VF4:       vector.ph:
+; VF4-NEXT:    br label [[VECTOR_BODY:%.*]]
+; VF4:       vector.body:
+; VF4-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 1, i64 4, i64 7, i64 10>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[VEC_IND2:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, [[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], [[VECTOR_BODY]] ]
+; VF4-NEXT:    [[TMP0:%.*]] = mul i64 [[INDEX]], 3
+; VF4-NEXT:    [[OFFSET_IDX:%.*]] = add i64 1, [[TMP0]]
+; VF4-NEXT:    [[TMP1:%.*]] = add i64 [[OFFSET_IDX]], 0
+; VF4-NEXT:    [[TMP2:%.*]] = add i64 [[OFFSET_IDX]], 3
+; VF4-NEXT:    [[TMP3:%.*]] = add i64 [[OFFSET_IDX]], 6
+; VF4-NEXT:    [[TMP4:%.*]] = add i64 [[OFFSET_IDX]], 9
+; VF4-NEXT:    [[TMP5:%.*]] = udiv <4 x i64> [[VEC_IND]], <i64 3, i64 3, i64 3, i64 3>
+; VF4-NEXT:    [[TMP6:%.*]] = udiv <4 x i64> [[VEC_IND2]], <i64 3, i64 3, i64 3, i64 3>
+; VF4-NEXT:    [[TMP7:%.*]] = add <4 x i64> [[TMP5]], [[TMP6]]
+; VF4-NEXT:    [[TMP8:%.*]] = extractelement <4 x i64> [[TMP7]], i32 0
+; VF4-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP8]]
+; VF4-NEXT:    [[TMP10:%.*]] = extractelement <4 x i64> [[TMP7]], i32 1
+; VF4-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP10]]
+; VF4-NEXT:    [[TMP12:%.*]] = extractelement <4 x i64> [[TMP7]], i32 2
+; VF4-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP12]]
+; VF4-NEXT:    [[TMP14:%.*]] = extractelement <4 x i64> [[TMP7]], i32 3
+; VF4-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[TMP14]]
+; VF4-NEXT:    [[TMP16:%.*]] = load i64, ptr [[TMP9]], align 8
+; VF4-NEXT:    [[TMP17:%.*]] = load i64, ptr [[TMP11]], align 8
+; VF4-NEXT:    [[TMP18:%.*]] = load i64, ptr [[TMP13]], align 8
+; VF4-NEXT:    [[TMP19:%.*]] = load i64, ptr [[TMP15]], align 8
+; VF4-NEXT:    [[TMP20:%.*]] = insertelement <4 x i64> poison, i64 [[TMP16]], i32 0
+; VF4-NEXT:    [[TMP21:%.*]] = insertelement <4 x i64> [[TMP20]], i64 [[TMP17]], i32 1
+; VF4-NEXT:    [[TMP22:%.*]] = insertelement <4 x i64> [[TMP21]], i64 [[TMP18]], i32 2
+; VF4-NEXT:    [[TMP23:%.*]] = insertelement <4 x i64> [[TMP22]], i64 [[TMP19]], i32 3
+; VF4-NEXT:    [[TMP24:%.*]] = add nsw <4 x i64> [[TMP23]], <i64 42, i64 42, i64 42, i64 42>
+; VF4-NEXT:    [[TMP25:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP1]]
+; VF4-NEXT:    [[TMP26:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP2]]
+; VF4-NEXT:    [[TMP27:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP3]]
+; VF4-NEXT:    [[TMP28:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[TMP4]]
+; VF4-NEXT:    [[TMP29:%.*]] = extractelement <4 x i64> [[TMP24]], i32 0
+; VF4-NEXT:    store i64 [[TMP29]], ptr [[TMP25]], align 8
+; VF4-NEXT:    [[TMP30:%.*]] = extractelement <4 x i64> [[TMP24]], i32 1
+; VF4-NEXT:    store i64 [[TMP30]], ptr [[TMP26]], align 8
+; VF4-NEXT:    [[TMP31:%.*]] = extractelement <4 x i64> [[TMP24]], i32 2
+; VF4-NEXT:    store i64 [[TMP31]], ptr [[TMP27]], align 8
+; VF4-NEXT:    [[TMP32:%.*]] = extractelement <4 x i64> [[TMP24]], i32 3
+; VF4-NEXT:    store i64 [[TMP32]], ptr [[TMP28]], align 8
+; VF4-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
+; VF4-NEXT:    [[VEC_IND_NEXT]] = add <4 x i64> [[VEC_IND]], <i64 12, i64 12, i64 12, i64 12>
+; VF4-NEXT:    [[VEC_IND_NEXT3]] = add <4 x i64> [[VEC_IND2]], <i64 4, i64 4, i64 4, i64 4>
+; VF4-NEXT:    [[TMP33:%.*]] = icmp eq i64 [[INDEX_NEXT]], 332
+; VF4-NEXT:    br i1 [[TMP33]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP36:![0-9]+]]
+; VF4:       middle.block:
+; VF4-NEXT:    [[CMP_N:%.*]] = icmp eq i64 333, 332
+; VF4-NEXT:    br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
+; VF4:       scalar.ph:
+; VF4-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ 997, [[MIDDLE_BLOCK]] ], [ 1, [[ENTRY:%.*]] ]
+; VF4-NEXT:    [[BC_RESUME_VAL1:%.*]] = phi i64 [ 332, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY]] ]
+; VF4-NEXT:    br label [[LOOP:%.*]]
+; VF4:       loop:
+; VF4-NEXT:    [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[IV2:%.*]] = phi i64 [ [[BC_RESUME_VAL1]], [[SCALAR_PH]] ], [ [[IV2_NEXT:%.*]], [[LOOP]] ]
+; VF4-NEXT:    [[DIV1:%.*]] = udiv i64 [[IV]], 3
+; VF4-NEXT:    [[DIV2:%.*]] = udiv i64 [[IV2]], 3
+; VF4-NEXT:    [[ADD:%.*]] = add i64 [[DIV1]], [[DIV2]]
+; VF4-NEXT:    [[GEP_LD:%.*]] = getelementptr inbounds i64, ptr [[A]], i64 [[ADD]]
+; VF4-NEXT:    [[LD:%.*]] = load i64, ptr [[GEP_LD]], align 8
+; VF4-NEXT:    [[CALC:%.*]] = add nsw i64 [[LD]], 42
+; VF4-NEXT:    [[GEP_ST:%.*]] = getelementptr inbounds i64, ptr [[B]], i64 [[IV]]
+; VF4-NEXT:    store i64 [[CALC]], ptr [[GEP_ST]], align 8
+; VF4-NEXT:    [[IV2_NEXT]] = add nsw i64 [[IV2]], 1
+; VF4-NEXT:    [[IV_NEXT]] = add nsw i64 [[IV]], 3
+; VF4-NEXT:    [[COND:%.*]] = icmp eq i64 [[IV_NEXT]], 1000
+; VF4-NEXT:    br i1 [[COND]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP37:![0-9]+]]
+; VF4:       exit:
+; VF4-NEXT:    ret void
 ;
 entry:
   br label %loop