llvm/test/Transforms/LoopVectorize/interleaved-accesses-retry-runtime-checks.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --filter-out-after "^scalar.ph:" --version 6
; RUN: opt -p loop-vectorize -enable-interleaved-mem-accesses -force-vector-width=4 -S %s | FileCheck %s

define void @dep_between_interleaved_store_and_load_with_runtime_checks(ptr %a, ptr %b, i64 %off) {
; CHECK-LABEL: define void @dep_between_interleaved_store_and_load_with_runtime_checks(
; CHECK-SAME: ptr [[A:%.*]], ptr [[B:%.*]], i64 [[OFF:%.*]]) {
; CHECK-NEXT:  [[ENTRY:.*:]]
; CHECK-NEXT:    [[B_OFF:%.*]] = getelementptr i8, ptr [[B]], i64 [[OFF]]
; CHECK-NEXT:    br label %[[VECTOR_MEMCHECK:.*]]
; CHECK:       [[VECTOR_MEMCHECK]]:
; CHECK-NEXT:    [[TMP0:%.*]] = getelementptr i8, ptr [[A]], i64 800
; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr i8, ptr [[B]], i64 400
; CHECK-NEXT:    [[TMP2:%.*]] = add i64 [[OFF]], 400
; CHECK-NEXT:    [[TMP3:%.*]] = getelementptr i8, ptr [[B]], i64 [[TMP2]]
; CHECK-NEXT:    [[BOUND0:%.*]] = icmp ult ptr [[A]], [[TMP1]]
; CHECK-NEXT:    [[BOUND1:%.*]] = icmp ult ptr [[B]], [[TMP0]]
; CHECK-NEXT:    [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
; CHECK-NEXT:    [[BOUND010:%.*]] = icmp ult ptr [[A]], [[TMP3]]
; CHECK-NEXT:    [[BOUND111:%.*]] = icmp ult ptr [[B_OFF]], [[TMP0]]
; CHECK-NEXT:    [[FOUND_CONFLICT12:%.*]] = and i1 [[BOUND010]], [[BOUND111]]
; CHECK-NEXT:    [[CONFLICT_RDX:%.*]] = or i1 [[FOUND_CONFLICT]], [[FOUND_CONFLICT12]]
; CHECK-NEXT:    [[BOUND013:%.*]] = icmp ult ptr [[B]], [[TMP3]]
; CHECK-NEXT:    [[BOUND114:%.*]] = icmp ult ptr [[B_OFF]], [[TMP1]]
; CHECK-NEXT:    [[FOUND_CONFLICT15:%.*]] = and i1 [[BOUND013]], [[BOUND114]]
; CHECK-NEXT:    [[CONFLICT_RDX16:%.*]] = or i1 [[CONFLICT_RDX]], [[FOUND_CONFLICT15]]
; CHECK-NEXT:    br i1 [[CONFLICT_RDX16]], 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:    [[TMP15:%.*]] = add i64 [[INDEX]], 1
; CHECK-NEXT:    [[TMP16:%.*]] = add i64 [[INDEX]], 2
; CHECK-NEXT:    [[TMP17:%.*]] = add i64 [[INDEX]], 3
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [2 x i32], ptr [[A]], i64 [[INDEX]]
; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [2 x i32], ptr [[A]], i64 [[TMP15]]
; CHECK-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [2 x i32], ptr [[A]], i64 [[TMP16]]
; CHECK-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [2 x i32], ptr [[A]], i64 [[TMP17]]
; CHECK-NEXT:    store i32 1, ptr [[TMP4]], align 4, !alias.scope [[META0:![0-9]+]], !noalias [[META3:![0-9]+]]
; CHECK-NEXT:    store i32 1, ptr [[TMP8]], align 4, !alias.scope [[META0]], !noalias [[META3]]
; CHECK-NEXT:    store i32 1, ptr [[TMP9]], align 4, !alias.scope [[META0]], !noalias [[META3]]
; CHECK-NEXT:    store i32 1, ptr [[TMP10]], align 4, !alias.scope [[META0]], !noalias [[META3]]
; CHECK-NEXT:    [[WIDE_VEC:%.*]] = load <8 x i32>, ptr [[TMP4]], align 4, !alias.scope [[META0]], !noalias [[META3]]
; CHECK-NEXT:    [[STRIDED_VEC:%.*]] = shufflevector <8 x i32> [[WIDE_VEC]], <8 x i32> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [2 x i32], ptr [[A]], i64 [[INDEX]], i64 1
; CHECK-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [2 x i32], ptr [[A]], i64 [[TMP15]], i64 1
; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [2 x i32], ptr [[A]], i64 [[TMP16]], i64 1
; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [2 x i32], ptr [[A]], i64 [[TMP17]], i64 1
; CHECK-NEXT:    store i32 3, ptr [[TMP11]], align 4, !alias.scope [[META0]], !noalias [[META3]]
; CHECK-NEXT:    store i32 3, ptr [[TMP12]], align 4, !alias.scope [[META0]], !noalias [[META3]]
; CHECK-NEXT:    store i32 3, ptr [[TMP13]], align 4, !alias.scope [[META0]], !noalias [[META3]]
; CHECK-NEXT:    store i32 3, ptr [[TMP14]], align 4, !alias.scope [[META0]], !noalias [[META3]]
; CHECK-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
; CHECK-NEXT:    store <4 x i32> [[STRIDED_VEC]], ptr [[TMP5]], align 4, !alias.scope [[META6:![0-9]+]], !noalias [[META7:![0-9]+]]
; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[B_OFF]], i64 [[INDEX]]
; CHECK-NEXT:    store <4 x i32> zeroinitializer, ptr [[TMP6]], align 4, !alias.scope [[META7]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP7:%.*]] = icmp eq i64 [[INDEX_NEXT]], 96
; CHECK-NEXT:    br i1 [[TMP7]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    br label %[[SCALAR_PH]]
; CHECK:       [[SCALAR_PH]]:
;
entry:
  %b.off = getelementptr i8, ptr %b, i64 %off
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %gep.a.0 = getelementptr inbounds [2 x i32], ptr %a, i64 %iv
  store i32 1, ptr %gep.a.0, align 4
  %l.a.0 = load i32, ptr %gep.a.0, align 4
  %gep.a.1 = getelementptr inbounds [2 x i32], ptr %a, i64 %iv, i64 1
  store i32 3, ptr %gep.a.1, align 4
  %gep.b = getelementptr inbounds i32, ptr %b, i64 %iv
  store i32 %l.a.0, ptr %gep.b, align 4
  %gep.b.off = getelementptr inbounds i32, ptr %b.off, i64 %iv
  store i32 0, ptr %gep.b.off, align 4
  %iv.next = add nuw nsw i64 %iv, 1
  %ec = icmp eq i64 %iv.next, 100
  br i1 %ec, label %exit, label %loop

exit:
  ret void
}

define void @dep_between_interleaved_store_and_load_with_unknown_dep_same_object(ptr %a, i64 %off) {
; CHECK-LABEL: define void @dep_between_interleaved_store_and_load_with_unknown_dep_same_object(
; CHECK-SAME: ptr [[A:%.*]], i64 [[OFF:%.*]]) {
; CHECK-NEXT:  [[ENTRY:.*:]]
; CHECK-NEXT:    [[A_OFF:%.*]] = getelementptr i8, ptr [[A]], i64 [[OFF]]
; CHECK-NEXT:    br label %[[VECTOR_MEMCHECK:.*]]
; CHECK:       [[VECTOR_MEMCHECK]]:
; CHECK-NEXT:    [[TMP0:%.*]] = getelementptr i8, ptr [[A]], i64 796
; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr nuw i8, ptr [[A]], i64 4
; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr i8, ptr [[A]], i64 800
; CHECK-NEXT:    [[TMP3:%.*]] = add i64 [[OFF]], 796
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr i8, ptr [[A]], i64 [[TMP3]]
; CHECK-NEXT:    [[BOUND0:%.*]] = icmp ult ptr [[A]], [[TMP2]]
; CHECK-NEXT:    [[BOUND1:%.*]] = icmp ult ptr [[TMP1]], [[TMP0]]
; CHECK-NEXT:    [[FOUND_CONFLICT:%.*]] = and i1 [[BOUND0]], [[BOUND1]]
; CHECK-NEXT:    [[BOUND011:%.*]] = icmp ult ptr [[A]], [[TMP4]]
; CHECK-NEXT:    [[BOUND112:%.*]] = icmp ult ptr [[A_OFF]], [[TMP0]]
; CHECK-NEXT:    [[FOUND_CONFLICT13:%.*]] = and i1 [[BOUND011]], [[BOUND112]]
; CHECK-NEXT:    [[CONFLICT_RDX:%.*]] = or i1 [[FOUND_CONFLICT]], [[FOUND_CONFLICT13]]
; CHECK-NEXT:    [[BOUND014:%.*]] = icmp ult ptr [[TMP1]], [[TMP4]]
; CHECK-NEXT:    [[BOUND115:%.*]] = icmp ult ptr [[A_OFF]], [[TMP2]]
; CHECK-NEXT:    [[FOUND_CONFLICT16:%.*]] = and i1 [[BOUND014]], [[BOUND115]]
; CHECK-NEXT:    [[CONFLICT_RDX17:%.*]] = or i1 [[CONFLICT_RDX]], [[FOUND_CONFLICT16]]
; CHECK-NEXT:    br i1 [[CONFLICT_RDX17]], 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:    [[TMP5:%.*]] = add i64 [[INDEX]], 1
; CHECK-NEXT:    [[TMP6:%.*]] = add i64 [[INDEX]], 2
; CHECK-NEXT:    [[TMP7:%.*]] = add i64 [[INDEX]], 3
; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [2 x i32], ptr [[A]], i64 [[INDEX]]
; CHECK-NEXT:    [[TMP9:%.*]] = getelementptr inbounds [2 x i32], ptr [[A]], i64 [[TMP5]]
; CHECK-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [2 x i32], ptr [[A]], i64 [[TMP6]]
; CHECK-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [2 x i32], ptr [[A]], i64 [[TMP7]]
; CHECK-NEXT:    store i32 1, ptr [[TMP8]], align 4, !alias.scope [[META12:![0-9]+]], !noalias [[META15:![0-9]+]]
; CHECK-NEXT:    store i32 1, ptr [[TMP9]], align 4, !alias.scope [[META12]], !noalias [[META15]]
; CHECK-NEXT:    store i32 1, ptr [[TMP10]], align 4, !alias.scope [[META12]], !noalias [[META15]]
; CHECK-NEXT:    store i32 1, ptr [[TMP11]], align 4, !alias.scope [[META12]], !noalias [[META15]]
; CHECK-NEXT:    [[WIDE_VEC:%.*]] = load <8 x i32>, ptr [[TMP8]], align 4, !alias.scope [[META12]], !noalias [[META15]]
; CHECK-NEXT:    [[STRIDED_VEC:%.*]] = shufflevector <8 x i32> [[WIDE_VEC]], <8 x i32> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [2 x i32], ptr [[A]], i64 [[INDEX]], i64 1
; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [2 x i32], ptr [[A]], i64 [[TMP5]], i64 1
; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [2 x i32], ptr [[A]], i64 [[TMP6]], i64 1
; CHECK-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [2 x i32], ptr [[A]], i64 [[TMP7]], i64 1
; CHECK-NEXT:    store i32 3, ptr [[TMP12]], align 4, !alias.scope [[META18:![0-9]+]], !noalias [[META19:![0-9]+]]
; CHECK-NEXT:    store i32 3, ptr [[TMP13]], align 4, !alias.scope [[META18]], !noalias [[META19]]
; CHECK-NEXT:    store i32 3, ptr [[TMP14]], align 4, !alias.scope [[META18]], !noalias [[META19]]
; CHECK-NEXT:    store i32 3, ptr [[TMP15]], align 4, !alias.scope [[META18]], !noalias [[META19]]
; CHECK-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [2 x i32], ptr [[A_OFF]], i64 [[INDEX]]
; CHECK-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [2 x i32], ptr [[A_OFF]], i64 [[TMP5]]
; CHECK-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [2 x i32], ptr [[A_OFF]], i64 [[TMP6]]
; CHECK-NEXT:    [[TMP19:%.*]] = getelementptr inbounds [2 x i32], ptr [[A_OFF]], i64 [[TMP7]]
; CHECK-NEXT:    [[TMP20:%.*]] = extractelement <4 x i32> [[STRIDED_VEC]], i64 0
; CHECK-NEXT:    store i32 [[TMP20]], ptr [[TMP16]], align 4, !alias.scope [[META19]]
; CHECK-NEXT:    [[TMP21:%.*]] = extractelement <4 x i32> [[STRIDED_VEC]], i64 1
; CHECK-NEXT:    store i32 [[TMP21]], ptr [[TMP17]], align 4, !alias.scope [[META19]]
; CHECK-NEXT:    [[TMP22:%.*]] = extractelement <4 x i32> [[STRIDED_VEC]], i64 2
; CHECK-NEXT:    store i32 [[TMP22]], ptr [[TMP18]], align 4, !alias.scope [[META19]]
; CHECK-NEXT:    [[TMP23:%.*]] = extractelement <4 x i32> [[STRIDED_VEC]], i64 3
; CHECK-NEXT:    store i32 [[TMP23]], ptr [[TMP19]], align 4, !alias.scope [[META19]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT:    [[TMP24:%.*]] = icmp eq i64 [[INDEX_NEXT]], 96
; CHECK-NEXT:    br i1 [[TMP24]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP20:![0-9]+]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    br label %[[SCALAR_PH]]
; CHECK:       [[SCALAR_PH]]:
;
entry:
  %a.off = getelementptr i8, ptr %a, i64 %off
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %gep.a.0 = getelementptr inbounds [2 x i32], ptr %a, i64 %iv
  store i32 1, ptr %gep.a.0, align 4
  %l.a.0 = load i32, ptr %gep.a.0, align 4
  %gep.a.1 = getelementptr inbounds [2 x i32], ptr %a, i64 %iv, i64 1
  store i32 3, ptr %gep.a.1, align 4
  %gep.a.off = getelementptr inbounds [2 x i32], ptr %a.off, i64 %iv
  store i32 %l.a.0, ptr %gep.a.off, align 4
  %iv.next = add nuw nsw i64 %iv, 1
  %ec = icmp eq i64 %iv.next, 100
  br i1 %ec, label %exit, label %loop

exit:
  ret void
}

; The accesses to %B and %B.off have a non-constant distance, so LAA retries
; with runtime checks between them. The accesses to %A are only checked against
; the other objects, not against each other, so the forward dependence from
; the store to A[2 * i + 2] to the load of A[2 * (i + 1)] in the next iteration
; must still prevent the load from being hoisted above the store as part of an
; interleave group.
define void @interleave_group_with_forward_dep_after_retry(ptr noalias %A, ptr noalias %C, ptr %B, i64 %off, i64 %n) {
;
; CHECK-LABEL: define void @interleave_group_with_forward_dep_after_retry(
; CHECK-SAME: ptr noalias [[A:%.*]], ptr noalias [[C:%.*]], ptr [[B:%.*]], i64 [[OFF:%.*]], i64 [[N:%.*]]) {
; CHECK-NEXT:  [[ENTRY:.*:]]
; CHECK-NEXT:    [[B_OFF:%.*]] = getelementptr i8, ptr [[B]], i64 [[OFF]]
; CHECK-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ule i64 [[N]], 4
; CHECK-NEXT:    br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_MEMCHECK:.*]]
; CHECK:       [[VECTOR_MEMCHECK]]:
; CHECK-NEXT:    [[TMP0:%.*]] = sub i64 [[OFF]], 1
; CHECK-NEXT:    [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP0]], 15
; CHECK-NEXT:    br i1 [[DIFF_CHECK]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]]
; CHECK:       [[VECTOR_PH]]:
; CHECK-NEXT:    [[TMP1:%.*]] = and i64 [[N]], 3
; CHECK-NEXT:    [[TMP9:%.*]] = icmp eq i64 [[TMP1]], 0
; CHECK-NEXT:    [[TMP24:%.*]] = select i1 [[TMP9]], i64 4, i64 [[TMP1]]
; CHECK-NEXT:    [[N_VEC:%.*]] = sub i64 [[N]], [[TMP24]]
; 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 <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ]
; CHECK-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i32, ptr [[B]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i32>, ptr [[TMP2]], align 4
; CHECK-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[B_OFF]], i64 [[INDEX]]
; CHECK-NEXT:    store <4 x i32> [[WIDE_LOAD]], ptr [[TMP3]], align 4
; CHECK-NEXT:    [[TMP4:%.*]] = shl nuw nsw <4 x i64> [[VEC_IND]], splat (i64 1)
; CHECK-NEXT:    [[TMP5:%.*]] = extractelement <4 x i64> [[TMP4]], i64 0
; CHECK-NEXT:    [[TMP6:%.*]] = or disjoint i64 [[TMP5]], 1
; CHECK-NEXT:    [[TMP7:%.*]] = add nuw nsw <4 x i64> [[TMP4]], splat (i64 2)
; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP6]]
; CHECK-NEXT:    [[WIDE_VEC:%.*]] = load <8 x i32>, ptr [[TMP8]], align 4
; CHECK-NEXT:    [[STRIDED_VEC:%.*]] = shufflevector <8 x i32> [[WIDE_VEC]], <8 x i32> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT:    [[TMP10:%.*]] = extractelement <4 x i64> [[TMP7]], i64 0
; CHECK-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP10]]
; CHECK-NEXT:    [[TMP12:%.*]] = extractelement <4 x i64> [[TMP7]], i64 1
; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP12]]
; CHECK-NEXT:    [[TMP14:%.*]] = extractelement <4 x i64> [[TMP7]], i64 2
; CHECK-NEXT:    [[TMP15:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP14]]
; CHECK-NEXT:    [[TMP16:%.*]] = extractelement <4 x i64> [[TMP7]], i64 3
; CHECK-NEXT:    [[TMP17:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP16]]
; CHECK-NEXT:    [[TMP18:%.*]] = extractelement <4 x i32> [[STRIDED_VEC]], i64 0
; CHECK-NEXT:    store i32 [[TMP18]], ptr [[TMP11]], align 4
; CHECK-NEXT:    [[TMP19:%.*]] = extractelement <4 x i32> [[STRIDED_VEC]], i64 1
; CHECK-NEXT:    store i32 [[TMP19]], ptr [[TMP13]], align 4
; CHECK-NEXT:    [[TMP20:%.*]] = extractelement <4 x i32> [[STRIDED_VEC]], i64 2
; CHECK-NEXT:    store i32 [[TMP20]], ptr [[TMP15]], align 4
; CHECK-NEXT:    [[TMP21:%.*]] = extractelement <4 x i32> [[STRIDED_VEC]], i64 3
; CHECK-NEXT:    store i32 [[TMP21]], ptr [[TMP17]], align 4
; CHECK-NEXT:    [[TMP25:%.*]] = getelementptr inbounds i32, ptr [[A]], i64 [[TMP5]]
; CHECK-NEXT:    [[WIDE_VEC1:%.*]] = load <8 x i32>, ptr [[TMP25]], align 4
; CHECK-NEXT:    [[STRIDED_VEC2:%.*]] = shufflevector <8 x i32> [[WIDE_VEC1]], <8 x i32> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT:    [[TMP22:%.*]] = getelementptr inbounds i32, ptr [[C]], i64 [[INDEX]]
; CHECK-NEXT:    store <4 x i32> [[STRIDED_VEC2]], ptr [[TMP22]], align 4
; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; CHECK-NEXT:    [[VEC_IND_NEXT]] = add nuw nsw <4 x i64> [[VEC_IND]], splat (i64 4)
; CHECK-NEXT:    [[TMP23:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; CHECK-NEXT:    br i1 [[TMP23]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP22:![0-9]+]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    br label %[[SCALAR_PH]]
; CHECK:       [[SCALAR_PH]]:
;
entry:
  %B.off = getelementptr i8, ptr %B, i64 %off
  br label %loop

loop:
  %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ]
  %gep.B = getelementptr inbounds i32, ptr %B, i64 %iv
  %l.B = load i32, ptr %gep.B, align 4
  %gep.B.off = getelementptr inbounds i32, ptr %B.off, i64 %iv
  store i32 %l.B, ptr %gep.B.off, align 4
  %iv.2 = shl nuw nsw i64 %iv, 1
  %iv.2.1 = or disjoint i64 %iv.2, 1
  %iv.2.2 = add nuw nsw i64 %iv.2, 2
  %gep.A.1 = getelementptr inbounds i32, ptr %A, i64 %iv.2.1
  %l.A.1 = load i32, ptr %gep.A.1, align 4
  %gep.A.2 = getelementptr inbounds i32, ptr %A, i64 %iv.2.2
  store i32 %l.A.1, ptr %gep.A.2, align 4
  %gep.A.0 = getelementptr inbounds i32, ptr %A, i64 %iv.2
  %l.A.0 = load i32, ptr %gep.A.0, align 4
  %gep.C = getelementptr inbounds i32, ptr %C, i64 %iv
  store i32 %l.A.0, ptr %gep.C, align 4
  %iv.next = add nuw nsw i64 %iv, 1
  %ec = icmp eq i64 %iv.next, %n
  br i1 %ec, label %exit, label %loop

exit:
  ret void
}