; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 6
; RUN: opt < %s -passes="default<O2>" -S -mattr=+sse2 | FileCheck %s
; RUN: opt < %s -passes="default<O2>" -S -mattr=+avx | FileCheck %s
; RUN: opt < %s -passes="default<O2>" -S -mattr=+avx2 | FileCheck %s
; RUN: opt < %s -passes="default<O2>" -S -mattr=+avx512f | FileCheck %s
; RUN: opt < %s -passes="default<O2>" -S -mattr=+avx512bw | FileCheck %s

define i32 @_Z10test_shortPsS_i_128(ptr nocapture readonly %i0, ptr nocapture readonly %i1, i32 %i2) local_unnamed_addr #0 {
; CHECK-LABEL: define i32 @_Z10test_shortPsS_i_128(
; CHECK-SAME: ptr readonly captures(none) [[I0:%.*]], ptr readonly captures(none) [[I1:%.*]], i32 [[I2:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] {
; CHECK-NEXT:  [[ENTRY:.*]]:
; CHECK-NEXT:    [[TMP3:%.*]] = zext i32 [[I2]] to i64
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ [[TMP9:%.*]], %[[VECTOR_BODY]] ], [ zeroinitializer, %[[ENTRY]] ]
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [2 x i8], ptr [[I0]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i16>, ptr [[TMP4]], align 2
; CHECK-NEXT:    [[TMP5:%.*]] = sext <4 x i16> [[WIDE_LOAD]] to <4 x i32>
; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [2 x i8], ptr [[I1]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD14:%.*]] = load <4 x i16>, ptr [[TMP6]], align 2
; CHECK-NEXT:    [[TMP7:%.*]] = sext <4 x i16> [[WIDE_LOAD14]] to <4 x i32>
; CHECK-NEXT:    [[TMP8:%.*]] = mul nsw <4 x i32> [[TMP7]], [[TMP5]]
; CHECK-NEXT:    [[TMP9]] = add nsw <4 x i32> [[TMP8]], [[VEC_PHI]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 8
; CHECK-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[TMP3]]
; CHECK-NEXT:    br i1 [[TMP10]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    [[TMP11:%.*]] = tail call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP9]])
; CHECK-NEXT:    ret i32 [[TMP11]]
;
entry:
  %i3 = zext i32 %i2 to i64
  br label %vector.body

vector.body:
  %index = phi i64 [ %index.next, %vector.body ], [ 0, %entry ]
  %vec.phi = phi <4 x i32> [ %i11, %vector.body ], [ zeroinitializer, %entry ]
  %i4 = getelementptr inbounds i16, ptr %i0, i64 %index
  %i5 = bitcast ptr %i4 to ptr
  %wide.load = load <4 x i16>, ptr %i5, align 2
  %i6 = sext <4 x i16> %wide.load to <4 x i32>
  %i7 = getelementptr inbounds i16, ptr %i1, i64 %index
  %i8 = bitcast ptr %i7 to ptr
  %wide.load14 = load <4 x i16>, ptr %i8, align 2
  %i9 = sext <4 x i16> %wide.load14 to <4 x i32>
  %i10 = mul nsw <4 x i32> %i9, %i6
  %i11 = add nsw <4 x i32> %i10, %vec.phi
  %index.next = add i64 %index, 8
  %i12 = icmp eq i64 %index.next, %i3
  br i1 %i12, label %middle.block, label %vector.body

middle.block:
  %rdx.shuf15 = shufflevector <4 x i32> %i11, <4 x i32> undef, <4 x i32> <i32 2, i32 3, i32 undef, i32 undef>
  %bin.rdx16 = add <4 x i32> %i11, %rdx.shuf15
  %rdx.shuf17 = shufflevector <4 x i32> %bin.rdx16, <4 x i32> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
  %bin.rdx18 = add <4 x i32> %bin.rdx16, %rdx.shuf17
  %i13 = extractelement <4 x i32> %bin.rdx18, i32 0
  ret i32 %i13
}

define i32 @_Z10test_shortPsS_i_256(ptr nocapture readonly %i0, ptr nocapture readonly %i1, i32 %i2) local_unnamed_addr #0 {
; CHECK-LABEL: define i32 @_Z10test_shortPsS_i_256(
; CHECK-SAME: ptr readonly captures(none) [[I0:%.*]], ptr readonly captures(none) [[I1:%.*]], i32 [[I2:%.*]]) local_unnamed_addr #[[ATTR0]] {
; CHECK-NEXT:  [[ENTRY:.*]]:
; CHECK-NEXT:    [[TMP3:%.*]] = zext i32 [[I2]] to i64
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <8 x i32> [ [[TMP9:%.*]], %[[VECTOR_BODY]] ], [ zeroinitializer, %[[ENTRY]] ]
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [2 x i8], ptr [[I0]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <8 x i16>, ptr [[TMP4]], align 2
; CHECK-NEXT:    [[TMP5:%.*]] = sext <8 x i16> [[WIDE_LOAD]] to <8 x i32>
; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [2 x i8], ptr [[I1]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD14:%.*]] = load <8 x i16>, ptr [[TMP6]], align 2
; CHECK-NEXT:    [[TMP7:%.*]] = sext <8 x i16> [[WIDE_LOAD14]] to <8 x i32>
; CHECK-NEXT:    [[TMP8:%.*]] = mul nsw <8 x i32> [[TMP7]], [[TMP5]]
; CHECK-NEXT:    [[TMP9]] = add nsw <8 x i32> [[TMP8]], [[VEC_PHI]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 8
; CHECK-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[TMP3]]
; CHECK-NEXT:    br i1 [[TMP10]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    [[TMP11:%.*]] = tail call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> [[TMP9]])
; CHECK-NEXT:    ret i32 [[TMP11]]
;
entry:
  %i3 = zext i32 %i2 to i64
  br label %vector.body

vector.body:
  %index = phi i64 [ %index.next, %vector.body ], [ 0, %entry ]
  %vec.phi = phi <8 x i32> [ %i11, %vector.body ], [ zeroinitializer, %entry ]
  %i4 = getelementptr inbounds i16, ptr %i0, i64 %index
  %i5 = bitcast ptr %i4 to ptr
  %wide.load = load <8 x i16>, ptr %i5, align 2
  %i6 = sext <8 x i16> %wide.load to <8 x i32>
  %i7 = getelementptr inbounds i16, ptr %i1, i64 %index
  %i8 = bitcast ptr %i7 to ptr
  %wide.load14 = load <8 x i16>, ptr %i8, align 2
  %i9 = sext <8 x i16> %wide.load14 to <8 x i32>
  %i10 = mul nsw <8 x i32> %i9, %i6
  %i11 = add nsw <8 x i32> %i10, %vec.phi
  %index.next = add i64 %index, 8
  %i12 = icmp eq i64 %index.next, %i3
  br i1 %i12, label %middle.block, label %vector.body

middle.block:
  %rdx.shuf = shufflevector <8 x i32> %i11, <8 x i32> undef, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx = add <8 x i32> %i11, %rdx.shuf
  %rdx.shuf15 = shufflevector <8 x i32> %bin.rdx, <8 x i32> undef, <8 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx16 = add <8 x i32> %bin.rdx, %rdx.shuf15
  %rdx.shuf17 = shufflevector <8 x i32> %bin.rdx16, <8 x i32> undef, <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx18 = add <8 x i32> %bin.rdx16, %rdx.shuf17
  %i13 = extractelement <8 x i32> %bin.rdx18, i32 0
  ret i32 %i13
}

define i32 @_Z10test_shortPsS_i_512(ptr nocapture readonly %i0, ptr nocapture readonly %i1, i32 %i2) local_unnamed_addr #0 {
; CHECK-LABEL: define i32 @_Z10test_shortPsS_i_512(
; CHECK-SAME: ptr readonly captures(none) [[I0:%.*]], ptr readonly captures(none) [[I1:%.*]], i32 [[I2:%.*]]) local_unnamed_addr #[[ATTR0]] {
; CHECK-NEXT:  [[ENTRY:.*]]:
; CHECK-NEXT:    [[TMP3:%.*]] = zext i32 [[I2]] to i64
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <16 x i32> [ [[TMP9:%.*]], %[[VECTOR_BODY]] ], [ zeroinitializer, %[[ENTRY]] ]
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [2 x i8], ptr [[I0]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <16 x i16>, ptr [[TMP4]], align 2
; CHECK-NEXT:    [[TMP5:%.*]] = sext <16 x i16> [[WIDE_LOAD]] to <16 x i32>
; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [2 x i8], ptr [[I1]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD14:%.*]] = load <16 x i16>, ptr [[TMP6]], align 2
; CHECK-NEXT:    [[TMP7:%.*]] = sext <16 x i16> [[WIDE_LOAD14]] to <16 x i32>
; CHECK-NEXT:    [[TMP8:%.*]] = mul nsw <16 x i32> [[TMP7]], [[TMP5]]
; CHECK-NEXT:    [[TMP9]] = add nsw <16 x i32> [[TMP8]], [[VEC_PHI]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 16
; CHECK-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[TMP3]]
; CHECK-NEXT:    br i1 [[TMP10]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    [[TMP11:%.*]] = tail call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> [[TMP9]])
; CHECK-NEXT:    ret i32 [[TMP11]]
;
entry:
  %i3 = zext i32 %i2 to i64
  br label %vector.body

vector.body:
  %index = phi i64 [ %index.next, %vector.body ], [ 0, %entry ]
  %vec.phi = phi <16 x i32> [ %i11, %vector.body ], [ zeroinitializer, %entry ]
  %i4 = getelementptr inbounds i16, ptr %i0, i64 %index
  %i5 = bitcast ptr %i4 to ptr
  %wide.load = load <16 x i16>, ptr %i5, align 2
  %i6 = sext <16 x i16> %wide.load to <16 x i32>
  %i7 = getelementptr inbounds i16, ptr %i1, i64 %index
  %i8 = bitcast ptr %i7 to ptr
  %wide.load14 = load <16 x i16>, ptr %i8, align 2
  %i9 = sext <16 x i16> %wide.load14 to <16 x i32>
  %i10 = mul nsw <16 x i32> %i9, %i6
  %i11 = add nsw <16 x i32> %i10, %vec.phi
  %index.next = add i64 %index, 16
  %i12 = icmp eq i64 %index.next, %i3
  br i1 %i12, label %middle.block, label %vector.body

middle.block:
  %rdx.shuf1 = shufflevector <16 x i32> %i11, <16 x i32> undef, <16 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx1 = add <16 x i32> %i11, %rdx.shuf1
  %rdx.shuf = shufflevector <16 x i32> %bin.rdx1, <16 x i32> undef, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx = add <16 x i32> %bin.rdx1, %rdx.shuf
  %rdx.shuf15 = shufflevector <16 x i32> %bin.rdx, <16 x i32> undef, <16 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx16 = add <16 x i32> %bin.rdx, %rdx.shuf15
  %rdx.shuf17 = shufflevector <16 x i32> %bin.rdx16, <16 x i32> undef, <16 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx18 = add <16 x i32> %bin.rdx16, %rdx.shuf17
  %i13 = extractelement <16 x i32> %bin.rdx18, i32 0
  ret i32 %i13
}

define i32 @_Z10test_shortPsS_i_1024(ptr nocapture readonly %i0, ptr nocapture readonly %i1, i32 %i2) local_unnamed_addr #0 {
; CHECK-LABEL: define i32 @_Z10test_shortPsS_i_1024(
; CHECK-SAME: ptr readonly captures(none) [[I0:%.*]], ptr readonly captures(none) [[I1:%.*]], i32 [[I2:%.*]]) local_unnamed_addr #[[ATTR0]] {
; CHECK-NEXT:  [[ENTRY:.*]]:
; CHECK-NEXT:    [[TMP3:%.*]] = zext i32 [[I2]] to i64
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <32 x i32> [ [[TMP9:%.*]], %[[VECTOR_BODY]] ], [ zeroinitializer, %[[ENTRY]] ]
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [2 x i8], ptr [[I0]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <32 x i16>, ptr [[TMP4]], align 2
; CHECK-NEXT:    [[TMP5:%.*]] = sext <32 x i16> [[WIDE_LOAD]] to <32 x i32>
; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [2 x i8], ptr [[I1]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD14:%.*]] = load <32 x i16>, ptr [[TMP6]], align 2
; CHECK-NEXT:    [[TMP7:%.*]] = sext <32 x i16> [[WIDE_LOAD14]] to <32 x i32>
; CHECK-NEXT:    [[TMP8:%.*]] = mul nsw <32 x i32> [[TMP7]], [[TMP5]]
; CHECK-NEXT:    [[TMP9]] = add nsw <32 x i32> [[TMP8]], [[VEC_PHI]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 16
; CHECK-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[TMP3]]
; CHECK-NEXT:    br i1 [[TMP10]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    [[TMP11:%.*]] = tail call i32 @llvm.vector.reduce.add.v32i32(<32 x i32> [[TMP9]])
; CHECK-NEXT:    ret i32 [[TMP11]]
;
entry:
  %i3 = zext i32 %i2 to i64
  br label %vector.body

vector.body:
  %index = phi i64 [ %index.next, %vector.body ], [ 0, %entry ]
  %vec.phi = phi <32 x i32> [ %i11, %vector.body ], [ zeroinitializer, %entry ]
  %i4 = getelementptr inbounds i16, ptr %i0, i64 %index
  %i5 = bitcast ptr %i4 to ptr
  %wide.load = load <32 x i16>, ptr %i5, align 2
  %i6 = sext <32 x i16> %wide.load to <32 x i32>
  %i7 = getelementptr inbounds i16, ptr %i1, i64 %index
  %i8 = bitcast ptr %i7 to ptr
  %wide.load14 = load <32 x i16>, ptr %i8, align 2
  %i9 = sext <32 x i16> %wide.load14 to <32 x i32>
  %i10 = mul nsw <32 x i32> %i9, %i6
  %i11 = add nsw <32 x i32> %i10, %vec.phi
  %index.next = add i64 %index, 16
  %i12 = icmp eq i64 %index.next, %i3
  br i1 %i12, label %middle.block, label %vector.body

middle.block:
  %rdx.shuf2 = shufflevector <32 x i32> %i11, <32 x i32> undef, <32 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx2 = add <32 x i32> %i11, %rdx.shuf2
  %rdx.shuf1 = shufflevector <32 x i32> %bin.rdx2, <32 x i32> undef, <32 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx1 = add <32 x i32> %bin.rdx2, %rdx.shuf1
  %rdx.shuf = shufflevector <32 x i32> %bin.rdx1, <32 x i32> undef, <32 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx = add <32 x i32> %bin.rdx1, %rdx.shuf
  %rdx.shuf15 = shufflevector <32 x i32> %bin.rdx, <32 x i32> undef, <32 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx16 = add <32 x i32> %bin.rdx, %rdx.shuf15
  %rdx.shuf17 = shufflevector <32 x i32> %bin.rdx16, <32 x i32> undef, <32 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx18 = add <32 x i32> %bin.rdx16, %rdx.shuf17
  %i13 = extractelement <32 x i32> %bin.rdx18, i32 0
  ret i32 %i13
}

define i32 @_Z9test_charPcS_i_128(ptr nocapture readonly %i0, ptr nocapture readonly %i1, i32 %i2) local_unnamed_addr #0 {
; CHECK-LABEL: define i32 @_Z9test_charPcS_i_128(
; CHECK-SAME: ptr readonly captures(none) [[I0:%.*]], ptr readonly captures(none) [[I1:%.*]], i32 [[I2:%.*]]) local_unnamed_addr #[[ATTR0]] {
; CHECK-NEXT:  [[ENTRY:.*]]:
; CHECK-NEXT:    [[TMP3:%.*]] = zext i32 [[I2]] to i64
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ [[TMP9:%.*]], %[[VECTOR_BODY]] ], [ zeroinitializer, %[[ENTRY]] ]
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[I0]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i8>, ptr [[TMP4]], align 1
; CHECK-NEXT:    [[TMP5:%.*]] = sext <4 x i8> [[WIDE_LOAD]] to <4 x i32>
; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[I1]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD14:%.*]] = load <4 x i8>, ptr [[TMP6]], align 1
; CHECK-NEXT:    [[TMP7:%.*]] = sext <4 x i8> [[WIDE_LOAD14]] to <4 x i32>
; CHECK-NEXT:    [[TMP8:%.*]] = mul nsw <4 x i32> [[TMP7]], [[TMP5]]
; CHECK-NEXT:    [[TMP9]] = add nsw <4 x i32> [[TMP8]], [[VEC_PHI]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 16
; CHECK-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[TMP3]]
; CHECK-NEXT:    br i1 [[TMP10]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    [[TMP11:%.*]] = tail call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP9]])
; CHECK-NEXT:    ret i32 [[TMP11]]
;
entry:
  %i3 = zext i32 %i2 to i64
  br label %vector.body

vector.body:
  %index = phi i64 [ %index.next, %vector.body ], [ 0, %entry ]
  %vec.phi = phi <4 x i32> [ %i11, %vector.body ], [ zeroinitializer, %entry ]
  %i4 = getelementptr inbounds i8, ptr %i0, i64 %index
  %i5 = bitcast ptr %i4 to ptr
  %wide.load = load <4 x i8>, ptr %i5, align 1
  %i6 = sext <4 x i8> %wide.load to <4 x i32>
  %i7 = getelementptr inbounds i8, ptr %i1, i64 %index
  %i8 = bitcast ptr %i7 to ptr
  %wide.load14 = load <4 x i8>, ptr %i8, align 1
  %i9 = sext <4 x i8> %wide.load14 to <4 x i32>
  %i10 = mul nsw <4 x i32> %i9, %i6
  %i11 = add nsw <4 x i32> %i10, %vec.phi
  %index.next = add i64 %index, 16
  %i12 = icmp eq i64 %index.next, %i3
  br i1 %i12, label %middle.block, label %vector.body

middle.block:
  %rdx.shuf17 = shufflevector <4 x i32> %i11, <4 x i32> undef, <4 x i32> <i32 2, i32 3, i32 undef, i32 undef>
  %bin.rdx18 = add <4 x i32> %i11, %rdx.shuf17
  %rdx.shuf19 = shufflevector <4 x i32> %bin.rdx18, <4 x i32> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
  %bin.rdx20 = add <4 x i32> %bin.rdx18, %rdx.shuf19
  %i13 = extractelement <4 x i32> %bin.rdx20, i32 0
  ret i32 %i13
}

define i32 @_Z9test_charPcS_i_256(ptr nocapture readonly %i0, ptr nocapture readonly %i1, i32 %i2) local_unnamed_addr #0 {
; CHECK-LABEL: define i32 @_Z9test_charPcS_i_256(
; CHECK-SAME: ptr readonly captures(none) [[I0:%.*]], ptr readonly captures(none) [[I1:%.*]], i32 [[I2:%.*]]) local_unnamed_addr #[[ATTR0]] {
; CHECK-NEXT:  [[ENTRY:.*]]:
; CHECK-NEXT:    [[TMP3:%.*]] = zext i32 [[I2]] to i64
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <8 x i32> [ [[TMP9:%.*]], %[[VECTOR_BODY]] ], [ zeroinitializer, %[[ENTRY]] ]
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[I0]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <8 x i8>, ptr [[TMP4]], align 1
; CHECK-NEXT:    [[TMP5:%.*]] = sext <8 x i8> [[WIDE_LOAD]] to <8 x i32>
; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[I1]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD14:%.*]] = load <8 x i8>, ptr [[TMP6]], align 1
; CHECK-NEXT:    [[TMP7:%.*]] = sext <8 x i8> [[WIDE_LOAD14]] to <8 x i32>
; CHECK-NEXT:    [[TMP8:%.*]] = mul nsw <8 x i32> [[TMP7]], [[TMP5]]
; CHECK-NEXT:    [[TMP9]] = add nsw <8 x i32> [[TMP8]], [[VEC_PHI]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 16
; CHECK-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[TMP3]]
; CHECK-NEXT:    br i1 [[TMP10]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    [[TMP11:%.*]] = tail call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> [[TMP9]])
; CHECK-NEXT:    ret i32 [[TMP11]]
;
entry:
  %i3 = zext i32 %i2 to i64
  br label %vector.body

vector.body:
  %index = phi i64 [ %index.next, %vector.body ], [ 0, %entry ]
  %vec.phi = phi <8 x i32> [ %i11, %vector.body ], [ zeroinitializer, %entry ]
  %i4 = getelementptr inbounds i8, ptr %i0, i64 %index
  %i5 = bitcast ptr %i4 to ptr
  %wide.load = load <8 x i8>, ptr %i5, align 1
  %i6 = sext <8 x i8> %wide.load to <8 x i32>
  %i7 = getelementptr inbounds i8, ptr %i1, i64 %index
  %i8 = bitcast ptr %i7 to ptr
  %wide.load14 = load <8 x i8>, ptr %i8, align 1
  %i9 = sext <8 x i8> %wide.load14 to <8 x i32>
  %i10 = mul nsw <8 x i32> %i9, %i6
  %i11 = add nsw <8 x i32> %i10, %vec.phi
  %index.next = add i64 %index, 16
  %i12 = icmp eq i64 %index.next, %i3
  br i1 %i12, label %middle.block, label %vector.body

middle.block:
  %rdx.shuf15 = shufflevector <8 x i32> %i11, <8 x i32> undef, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx16 = add <8 x i32> %i11, %rdx.shuf15
  %rdx.shuf17 = shufflevector <8 x i32> %bin.rdx16, <8 x i32> undef, <8 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx18 = add <8 x i32> %bin.rdx16, %rdx.shuf17
  %rdx.shuf19 = shufflevector <8 x i32> %bin.rdx18, <8 x i32> undef, <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx20 = add <8 x i32> %bin.rdx18, %rdx.shuf19
  %i13 = extractelement <8 x i32> %bin.rdx20, i32 0
  ret i32 %i13
}

define i32 @_Z9test_charPcS_i_512(ptr nocapture readonly %i0, ptr nocapture readonly %i1, i32 %i2) local_unnamed_addr #0 {
; CHECK-LABEL: define i32 @_Z9test_charPcS_i_512(
; CHECK-SAME: ptr readonly captures(none) [[I0:%.*]], ptr readonly captures(none) [[I1:%.*]], i32 [[I2:%.*]]) local_unnamed_addr #[[ATTR0]] {
; CHECK-NEXT:  [[ENTRY:.*]]:
; CHECK-NEXT:    [[TMP3:%.*]] = zext i32 [[I2]] to i64
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <16 x i32> [ [[TMP9:%.*]], %[[VECTOR_BODY]] ], [ zeroinitializer, %[[ENTRY]] ]
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[I0]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <16 x i8>, ptr [[TMP4]], align 1
; CHECK-NEXT:    [[TMP5:%.*]] = sext <16 x i8> [[WIDE_LOAD]] to <16 x i32>
; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[I1]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD14:%.*]] = load <16 x i8>, ptr [[TMP6]], align 1
; CHECK-NEXT:    [[TMP7:%.*]] = sext <16 x i8> [[WIDE_LOAD14]] to <16 x i32>
; CHECK-NEXT:    [[TMP8:%.*]] = mul nsw <16 x i32> [[TMP7]], [[TMP5]]
; CHECK-NEXT:    [[TMP9]] = add nsw <16 x i32> [[TMP8]], [[VEC_PHI]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 16
; CHECK-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[TMP3]]
; CHECK-NEXT:    br i1 [[TMP10]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    [[TMP11:%.*]] = tail call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> [[TMP9]])
; CHECK-NEXT:    ret i32 [[TMP11]]
;
entry:
  %i3 = zext i32 %i2 to i64
  br label %vector.body

vector.body:
  %index = phi i64 [ %index.next, %vector.body ], [ 0, %entry ]
  %vec.phi = phi <16 x i32> [ %i11, %vector.body ], [ zeroinitializer, %entry ]
  %i4 = getelementptr inbounds i8, ptr %i0, i64 %index
  %i5 = bitcast ptr %i4 to ptr
  %wide.load = load <16 x i8>, ptr %i5, align 1
  %i6 = sext <16 x i8> %wide.load to <16 x i32>
  %i7 = getelementptr inbounds i8, ptr %i1, i64 %index
  %i8 = bitcast ptr %i7 to ptr
  %wide.load14 = load <16 x i8>, ptr %i8, align 1
  %i9 = sext <16 x i8> %wide.load14 to <16 x i32>
  %i10 = mul nsw <16 x i32> %i9, %i6
  %i11 = add nsw <16 x i32> %i10, %vec.phi
  %index.next = add i64 %index, 16
  %i12 = icmp eq i64 %index.next, %i3
  br i1 %i12, label %middle.block, label %vector.body

middle.block:
  %rdx.shuf = shufflevector <16 x i32> %i11, <16 x i32> undef, <16 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx = add <16 x i32> %i11, %rdx.shuf
  %rdx.shuf15 = shufflevector <16 x i32> %bin.rdx, <16 x i32> undef, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx16 = add <16 x i32> %bin.rdx, %rdx.shuf15
  %rdx.shuf17 = shufflevector <16 x i32> %bin.rdx16, <16 x i32> undef, <16 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx18 = add <16 x i32> %bin.rdx16, %rdx.shuf17
  %rdx.shuf19 = shufflevector <16 x i32> %bin.rdx18, <16 x i32> undef, <16 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx20 = add <16 x i32> %bin.rdx18, %rdx.shuf19
  %i13 = extractelement <16 x i32> %bin.rdx20, i32 0
  ret i32 %i13
}

define i32 @_Z9test_charPcS_i_1024(ptr nocapture readonly %i0, ptr nocapture readonly %i1, i32 %i2) local_unnamed_addr #0 {
; CHECK-LABEL: define i32 @_Z9test_charPcS_i_1024(
; CHECK-SAME: ptr readonly captures(none) [[I0:%.*]], ptr readonly captures(none) [[I1:%.*]], i32 [[I2:%.*]]) local_unnamed_addr #[[ATTR0]] {
; CHECK-NEXT:  [[ENTRY:.*]]:
; CHECK-NEXT:    [[TMP3:%.*]] = zext i32 [[I2]] to i64
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <32 x i32> [ [[TMP9:%.*]], %[[VECTOR_BODY]] ], [ zeroinitializer, %[[ENTRY]] ]
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[I0]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <32 x i8>, ptr [[TMP4]], align 1
; CHECK-NEXT:    [[TMP5:%.*]] = sext <32 x i8> [[WIDE_LOAD]] to <32 x i32>
; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds i8, ptr [[I1]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD14:%.*]] = load <32 x i8>, ptr [[TMP6]], align 1
; CHECK-NEXT:    [[TMP7:%.*]] = sext <32 x i8> [[WIDE_LOAD14]] to <32 x i32>
; CHECK-NEXT:    [[TMP8:%.*]] = mul nsw <32 x i32> [[TMP7]], [[TMP5]]
; CHECK-NEXT:    [[TMP9]] = add nsw <32 x i32> [[TMP8]], [[VEC_PHI]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 32
; CHECK-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[TMP3]]
; CHECK-NEXT:    br i1 [[TMP10]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    [[TMP11:%.*]] = tail call i32 @llvm.vector.reduce.add.v32i32(<32 x i32> [[TMP9]])
; CHECK-NEXT:    ret i32 [[TMP11]]
;
entry:
  %i3 = zext i32 %i2 to i64
  br label %vector.body

vector.body:
  %index = phi i64 [ %index.next, %vector.body ], [ 0, %entry ]
  %vec.phi = phi <32 x i32> [ %i11, %vector.body ], [ zeroinitializer, %entry ]
  %i4 = getelementptr inbounds i8, ptr %i0, i64 %index
  %i5 = bitcast ptr %i4 to ptr
  %wide.load = load <32 x i8>, ptr %i5, align 1
  %i6 = sext <32 x i8> %wide.load to <32 x i32>
  %i7 = getelementptr inbounds i8, ptr %i1, i64 %index
  %i8 = bitcast ptr %i7 to ptr
  %wide.load14 = load <32 x i8>, ptr %i8, align 1
  %i9 = sext <32 x i8> %wide.load14 to <32 x i32>
  %i10 = mul nsw <32 x i32> %i9, %i6
  %i11 = add nsw <32 x i32> %i10, %vec.phi
  %index.next = add i64 %index, 32
  %i12 = icmp eq i64 %index.next, %i3
  br i1 %i12, label %middle.block, label %vector.body

middle.block:
  %rdx.shuf1 = shufflevector <32 x i32> %i11, <32 x i32> undef, <32 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx1 = add <32 x i32> %i11, %rdx.shuf1
  %rdx.shuf = shufflevector <32 x i32> %bin.rdx1, <32 x i32> undef, <32 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx = add <32 x i32> %bin.rdx1, %rdx.shuf
  %rdx.shuf15 = shufflevector <32 x i32> %bin.rdx, <32 x i32> undef, <32 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx32 = add <32 x i32> %bin.rdx, %rdx.shuf15
  %rdx.shuf17 = shufflevector <32 x i32> %bin.rdx32, <32 x i32> undef, <32 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx18 = add <32 x i32> %bin.rdx32, %rdx.shuf17
  %rdx.shuf19 = shufflevector <32 x i32> %bin.rdx18, <32 x i32> undef, <32 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx20 = add <32 x i32> %bin.rdx18, %rdx.shuf19
  %i13 = extractelement <32 x i32> %bin.rdx20, i32 0
  ret i32 %i13
}

define i32 @test_unsigned_short_128(ptr nocapture readonly %i0, ptr nocapture readonly %i1, i32 %i2) local_unnamed_addr #0 {
; CHECK-LABEL: define i32 @test_unsigned_short_128(
; CHECK-SAME: ptr readonly captures(none) [[I0:%.*]], ptr readonly captures(none) [[I1:%.*]], i32 [[I2:%.*]]) local_unnamed_addr #[[ATTR0]] {
; CHECK-NEXT:  [[ENTRY:.*]]:
; CHECK-NEXT:    [[TMP3:%.*]] = zext i32 [[I2]] to i64
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <4 x i32> [ [[TMP9:%.*]], %[[VECTOR_BODY]] ], [ zeroinitializer, %[[ENTRY]] ]
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [2 x i8], ptr [[I0]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <4 x i16>, ptr [[TMP4]], align 2
; CHECK-NEXT:    [[TMP5:%.*]] = zext <4 x i16> [[WIDE_LOAD]] to <4 x i32>
; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [2 x i8], ptr [[I1]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD14:%.*]] = load <4 x i16>, ptr [[TMP6]], align 2
; CHECK-NEXT:    [[TMP7:%.*]] = zext <4 x i16> [[WIDE_LOAD14]] to <4 x i32>
; CHECK-NEXT:    [[TMP8:%.*]] = mul nuw nsw <4 x i32> [[TMP7]], [[TMP5]]
; CHECK-NEXT:    [[TMP9]] = add nuw nsw <4 x i32> [[TMP8]], [[VEC_PHI]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 16
; CHECK-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[TMP3]]
; CHECK-NEXT:    br i1 [[TMP10]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    [[TMP11:%.*]] = tail call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP9]])
; CHECK-NEXT:    ret i32 [[TMP11]]
;
entry:
  %i3 = zext i32 %i2 to i64
  br label %vector.body

vector.body:
  %index = phi i64 [ %index.next, %vector.body ], [ 0, %entry ]
  %vec.phi = phi <4 x i32> [ %i11, %vector.body ], [ zeroinitializer, %entry ]
  %i4 = getelementptr inbounds i16, ptr %i0, i64 %index
  %i5 = bitcast ptr %i4 to ptr
  %wide.load = load <4 x i16>, ptr %i5, align 2
  %i6 = zext <4 x i16> %wide.load to <4 x i32>
  %i7 = getelementptr inbounds i16, ptr %i1, i64 %index
  %i8 = bitcast ptr %i7 to ptr
  %wide.load14 = load <4 x i16>, ptr %i8, align 2
  %i9 = zext <4 x i16> %wide.load14 to <4 x i32>
  %i10 = mul nsw <4 x i32> %i9, %i6
  %i11 = add nsw <4 x i32> %i10, %vec.phi
  %index.next = add i64 %index, 16
  %i12 = icmp eq i64 %index.next, %i3
  br i1 %i12, label %middle.block, label %vector.body

middle.block:
  %rdx.shuf15 = shufflevector <4 x i32> %i11, <4 x i32> undef, <4 x i32> <i32 2, i32 3, i32 undef, i32 undef>
  %bin.rdx16 = add <4 x i32> %i11, %rdx.shuf15
  %rdx.shuf17 = shufflevector <4 x i32> %bin.rdx16, <4 x i32> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
  %bin.rdx18 = add <4 x i32> %bin.rdx16, %rdx.shuf17
  %i13 = extractelement <4 x i32> %bin.rdx18, i32 0
  ret i32 %i13
}

define i32 @test_unsigned_short_256(ptr nocapture readonly %i0, ptr nocapture readonly %i1, i32 %i2) local_unnamed_addr #0 {
; CHECK-LABEL: define i32 @test_unsigned_short_256(
; CHECK-SAME: ptr readonly captures(none) [[I0:%.*]], ptr readonly captures(none) [[I1:%.*]], i32 [[I2:%.*]]) local_unnamed_addr #[[ATTR0]] {
; CHECK-NEXT:  [[ENTRY:.*]]:
; CHECK-NEXT:    [[TMP3:%.*]] = zext i32 [[I2]] to i64
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <8 x i32> [ [[TMP9:%.*]], %[[VECTOR_BODY]] ], [ zeroinitializer, %[[ENTRY]] ]
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [2 x i8], ptr [[I0]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <8 x i16>, ptr [[TMP4]], align 2
; CHECK-NEXT:    [[TMP5:%.*]] = zext <8 x i16> [[WIDE_LOAD]] to <8 x i32>
; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [2 x i8], ptr [[I1]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD14:%.*]] = load <8 x i16>, ptr [[TMP6]], align 2
; CHECK-NEXT:    [[TMP7:%.*]] = zext <8 x i16> [[WIDE_LOAD14]] to <8 x i32>
; CHECK-NEXT:    [[TMP8:%.*]] = mul nuw nsw <8 x i32> [[TMP7]], [[TMP5]]
; CHECK-NEXT:    [[TMP9]] = add nuw nsw <8 x i32> [[TMP8]], [[VEC_PHI]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 16
; CHECK-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[TMP3]]
; CHECK-NEXT:    br i1 [[TMP10]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    [[TMP11:%.*]] = tail call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> [[TMP9]])
; CHECK-NEXT:    ret i32 [[TMP11]]
;
entry:
  %i3 = zext i32 %i2 to i64
  br label %vector.body

vector.body:
  %index = phi i64 [ %index.next, %vector.body ], [ 0, %entry ]
  %vec.phi = phi <8 x i32> [ %i11, %vector.body ], [ zeroinitializer, %entry ]
  %i4 = getelementptr inbounds i16, ptr %i0, i64 %index
  %i5 = bitcast ptr %i4 to ptr
  %wide.load = load <8 x i16>, ptr %i5, align 2
  %i6 = zext <8 x i16> %wide.load to <8 x i32>
  %i7 = getelementptr inbounds i16, ptr %i1, i64 %index
  %i8 = bitcast ptr %i7 to ptr
  %wide.load14 = load <8 x i16>, ptr %i8, align 2
  %i9 = zext <8 x i16> %wide.load14 to <8 x i32>
  %i10 = mul nsw <8 x i32> %i9, %i6
  %i11 = add nsw <8 x i32> %i10, %vec.phi
  %index.next = add i64 %index, 16
  %i12 = icmp eq i64 %index.next, %i3
  br i1 %i12, label %middle.block, label %vector.body

middle.block:
  %rdx.shuf = shufflevector <8 x i32> %i11, <8 x i32> undef, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx = add <8 x i32> %i11, %rdx.shuf
  %rdx.shuf15 = shufflevector <8 x i32> %bin.rdx, <8 x i32> undef, <8 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx16 = add <8 x i32> %bin.rdx, %rdx.shuf15
  %rdx.shuf17 = shufflevector <8 x i32> %bin.rdx16, <8 x i32> undef, <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx18 = add <8 x i32> %bin.rdx16, %rdx.shuf17
  %i13 = extractelement <8 x i32> %bin.rdx18, i32 0
  ret i32 %i13
}

define i32 @test_unsigned_short_512(ptr nocapture readonly %i0, ptr nocapture readonly %i1, i32 %i2) local_unnamed_addr #0 {
; CHECK-LABEL: define i32 @test_unsigned_short_512(
; CHECK-SAME: ptr readonly captures(none) [[I0:%.*]], ptr readonly captures(none) [[I1:%.*]], i32 [[I2:%.*]]) local_unnamed_addr #[[ATTR0]] {
; CHECK-NEXT:  [[ENTRY:.*]]:
; CHECK-NEXT:    [[TMP3:%.*]] = zext i32 [[I2]] to i64
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <16 x i32> [ [[TMP9:%.*]], %[[VECTOR_BODY]] ], [ zeroinitializer, %[[ENTRY]] ]
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [2 x i8], ptr [[I0]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <16 x i16>, ptr [[TMP4]], align 2
; CHECK-NEXT:    [[TMP5:%.*]] = zext <16 x i16> [[WIDE_LOAD]] to <16 x i32>
; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [2 x i8], ptr [[I1]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD14:%.*]] = load <16 x i16>, ptr [[TMP6]], align 2
; CHECK-NEXT:    [[TMP7:%.*]] = zext <16 x i16> [[WIDE_LOAD14]] to <16 x i32>
; CHECK-NEXT:    [[TMP8:%.*]] = mul nuw nsw <16 x i32> [[TMP7]], [[TMP5]]
; CHECK-NEXT:    [[TMP9]] = add nuw nsw <16 x i32> [[TMP8]], [[VEC_PHI]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 16
; CHECK-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[TMP3]]
; CHECK-NEXT:    br i1 [[TMP10]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    [[TMP11:%.*]] = tail call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> [[TMP9]])
; CHECK-NEXT:    ret i32 [[TMP11]]
;
entry:
  %i3 = zext i32 %i2 to i64
  br label %vector.body

vector.body:
  %index = phi i64 [ %index.next, %vector.body ], [ 0, %entry ]
  %vec.phi = phi <16 x i32> [ %i11, %vector.body ], [ zeroinitializer, %entry ]
  %i4 = getelementptr inbounds i16, ptr %i0, i64 %index
  %i5 = bitcast ptr %i4 to ptr
  %wide.load = load <16 x i16>, ptr %i5, align 2
  %i6 = zext <16 x i16> %wide.load to <16 x i32>
  %i7 = getelementptr inbounds i16, ptr %i1, i64 %index
  %i8 = bitcast ptr %i7 to ptr
  %wide.load14 = load <16 x i16>, ptr %i8, align 2
  %i9 = zext <16 x i16> %wide.load14 to <16 x i32>
  %i10 = mul nsw <16 x i32> %i9, %i6
  %i11 = add nsw <16 x i32> %i10, %vec.phi
  %index.next = add i64 %index, 16
  %i12 = icmp eq i64 %index.next, %i3
  br i1 %i12, label %middle.block, label %vector.body

middle.block:
  %rdx.shuf1 = shufflevector <16 x i32> %i11, <16 x i32> undef, <16 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx1 = add <16 x i32> %i11, %rdx.shuf1
  %rdx.shuf = shufflevector <16 x i32> %bin.rdx1, <16 x i32> undef, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx = add <16 x i32> %bin.rdx1, %rdx.shuf
  %rdx.shuf15 = shufflevector <16 x i32> %bin.rdx, <16 x i32> undef, <16 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx16 = add <16 x i32> %bin.rdx, %rdx.shuf15
  %rdx.shuf17 = shufflevector <16 x i32> %bin.rdx16, <16 x i32> undef, <16 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx18 = add <16 x i32> %bin.rdx16, %rdx.shuf17
  %i13 = extractelement <16 x i32> %bin.rdx18, i32 0
  ret i32 %i13
}

define i32 @test_unsigned_short_1024(ptr nocapture readonly %i0, ptr nocapture readonly %i1, i32 %i2) local_unnamed_addr #0 {
; CHECK-LABEL: define i32 @test_unsigned_short_1024(
; CHECK-SAME: ptr readonly captures(none) [[I0:%.*]], ptr readonly captures(none) [[I1:%.*]], i32 [[I2:%.*]]) local_unnamed_addr #[[ATTR0]] {
; CHECK-NEXT:  [[ENTRY:.*]]:
; CHECK-NEXT:    [[TMP3:%.*]] = zext i32 [[I2]] to i64
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <32 x i32> [ [[TMP9:%.*]], %[[VECTOR_BODY]] ], [ zeroinitializer, %[[ENTRY]] ]
; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [2 x i8], ptr [[I0]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <32 x i16>, ptr [[TMP4]], align 2
; CHECK-NEXT:    [[TMP5:%.*]] = zext <32 x i16> [[WIDE_LOAD]] to <32 x i32>
; CHECK-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [2 x i8], ptr [[I1]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD14:%.*]] = load <32 x i16>, ptr [[TMP6]], align 2
; CHECK-NEXT:    [[TMP7:%.*]] = zext <32 x i16> [[WIDE_LOAD14]] to <32 x i32>
; CHECK-NEXT:    [[TMP8:%.*]] = mul nuw nsw <32 x i32> [[TMP7]], [[TMP5]]
; CHECK-NEXT:    [[TMP9]] = add nuw nsw <32 x i32> [[TMP8]], [[VEC_PHI]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 16
; CHECK-NEXT:    [[TMP10:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[TMP3]]
; CHECK-NEXT:    br i1 [[TMP10]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    [[TMP11:%.*]] = tail call i32 @llvm.vector.reduce.add.v32i32(<32 x i32> [[TMP9]])
; CHECK-NEXT:    ret i32 [[TMP11]]
;
entry:
  %i3 = zext i32 %i2 to i64
  br label %vector.body

vector.body:
  %index = phi i64 [ %index.next, %vector.body ], [ 0, %entry ]
  %vec.phi = phi <32 x i32> [ %i11, %vector.body ], [ zeroinitializer, %entry ]
  %i4 = getelementptr inbounds i16, ptr %i0, i64 %index
  %i5 = bitcast ptr %i4 to ptr
  %wide.load = load <32 x i16>, ptr %i5, align 2
  %i6 = zext <32 x i16> %wide.load to <32 x i32>
  %i7 = getelementptr inbounds i16, ptr %i1, i64 %index
  %i8 = bitcast ptr %i7 to ptr
  %wide.load14 = load <32 x i16>, ptr %i8, align 2
  %i9 = zext <32 x i16> %wide.load14 to <32 x i32>
  %i10 = mul nsw <32 x i32> %i9, %i6
  %i11 = add nsw <32 x i32> %i10, %vec.phi
  %index.next = add i64 %index, 16
  %i12 = icmp eq i64 %index.next, %i3
  br i1 %i12, label %middle.block, label %vector.body

middle.block:
  %rdx.shuf2 = shufflevector <32 x i32> %i11, <32 x i32> undef, <32 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx2 = add <32 x i32> %i11, %rdx.shuf2
  %rdx.shuf1 = shufflevector <32 x i32> %bin.rdx2, <32 x i32> undef, <32 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx1 = add <32 x i32> %bin.rdx2, %rdx.shuf1
  %rdx.shuf = shufflevector <32 x i32> %bin.rdx1, <32 x i32> undef, <32 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx = add <32 x i32> %bin.rdx1, %rdx.shuf
  %rdx.shuf15 = shufflevector <32 x i32> %bin.rdx, <32 x i32> undef, <32 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx16 = add <32 x i32> %bin.rdx, %rdx.shuf15
  %rdx.shuf17 = shufflevector <32 x i32> %bin.rdx16, <32 x i32> undef, <32 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx18 = add <32 x i32> %bin.rdx16, %rdx.shuf17
  %i13 = extractelement <32 x i32> %bin.rdx18, i32 0
  ret i32 %i13
}

define <4 x i32> @pmaddwd_8(<8 x i16> %a0, <8 x i16> %a1) {
; CHECK-LABEL: define <4 x i32> @pmaddwd_8(
; CHECK-SAME: <8 x i16> [[A0:%.*]], <8 x i16> [[A1:%.*]]) local_unnamed_addr #[[ATTR1:[0-9]+]] {
; CHECK-NEXT:    [[A:%.*]] = sext <8 x i16> [[A0]] to <8 x i32>
; CHECK-NEXT:    [[B:%.*]] = sext <8 x i16> [[A1]] to <8 x i32>
; CHECK-NEXT:    [[M:%.*]] = mul nsw <8 x i32> [[B]], [[A]]
; CHECK-NEXT:    [[ODD:%.*]] = shufflevector <8 x i32> [[M]], <8 x i32> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT:    [[EVEN:%.*]] = shufflevector <8 x i32> [[M]], <8 x i32> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; CHECK-NEXT:    [[RET:%.*]] = add <4 x i32> [[ODD]], [[EVEN]]
; CHECK-NEXT:    ret <4 x i32> [[RET]]
;
  %a = sext <8 x i16> %a0 to <8 x i32>
  %b = sext <8 x i16> %a1 to <8 x i32>
  %m = mul nsw <8 x i32> %a, %b
  %odd = shufflevector <8 x i32> %m, <8 x i32> undef, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
  %even = shufflevector <8 x i32> %m, <8 x i32> undef, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
  %ret = add <4 x i32> %odd, %even
  ret <4 x i32> %ret
}

define <4 x i32> @pmaddwd_8_swapped(<8 x i16> %a0, <8 x i16> %b0) {
; CHECK-LABEL: define <4 x i32> @pmaddwd_8_swapped(
; CHECK-SAME: <8 x i16> [[A0:%.*]], <8 x i16> [[B0:%.*]]) local_unnamed_addr #[[ATTR1]] {
; CHECK-NEXT:    [[A:%.*]] = sext <8 x i16> [[A0]] to <8 x i32>
; CHECK-NEXT:    [[B:%.*]] = sext <8 x i16> [[B0]] to <8 x i32>
; CHECK-NEXT:    [[M:%.*]] = mul nsw <8 x i32> [[B]], [[A]]
; CHECK-NEXT:    [[ODD:%.*]] = shufflevector <8 x i32> [[M]], <8 x i32> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT:    [[EVEN:%.*]] = shufflevector <8 x i32> [[M]], <8 x i32> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; CHECK-NEXT:    [[RET:%.*]] = add <4 x i32> [[EVEN]], [[ODD]]
; CHECK-NEXT:    ret <4 x i32> [[RET]]
;
  %a = sext <8 x i16> %a0 to <8 x i32>
  %b = sext <8 x i16> %b0 to <8 x i32>
  %m = mul nsw <8 x i32> %a, %b
  %odd = shufflevector <8 x i32> %m, <8 x i32> undef, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
  %even = shufflevector <8 x i32> %m, <8 x i32> undef, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
  %ret = add <4 x i32> %even, %odd
  ret <4 x i32> %ret
}

; FIXME: SSE fails to match PMADDWD
define <4 x i32> @larger_mul(<16 x i16> %a0, <16 x i16> %b0) {
; CHECK-LABEL: define <4 x i32> @larger_mul(
; CHECK-SAME: <16 x i16> [[A0:%.*]], <16 x i16> [[B0:%.*]]) local_unnamed_addr #[[ATTR1]] {
; CHECK-NEXT:    [[A:%.*]] = sext <16 x i16> [[A0]] to <16 x i32>
; CHECK-NEXT:    [[B:%.*]] = sext <16 x i16> [[B0]] to <16 x i32>
; CHECK-NEXT:    [[M:%.*]] = mul nsw <16 x i32> [[B]], [[A]]
; CHECK-NEXT:    [[ODD:%.*]] = shufflevector <16 x i32> [[M]], <16 x i32> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT:    [[EVEN:%.*]] = shufflevector <16 x i32> [[M]], <16 x i32> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; CHECK-NEXT:    [[RET:%.*]] = add <4 x i32> [[ODD]], [[EVEN]]
; CHECK-NEXT:    ret <4 x i32> [[RET]]
;
  %a = sext <16 x i16> %a0 to <16 x i32>
  %b = sext <16 x i16> %b0 to <16 x i32>
  %m = mul nsw <16 x i32> %a, %b
  %odd = shufflevector <16 x i32> %m, <16 x i32> undef, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
  %even = shufflevector <16 x i32> %m, <16 x i32> undef, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
  %ret = add <4 x i32> %odd, %even
  ret <4 x i32> %ret
}

define <8 x i32> @pmaddwd_16(<16 x i16> %a0, <16 x i16> %b0) {
; CHECK-LABEL: define <8 x i32> @pmaddwd_16(
; CHECK-SAME: <16 x i16> [[A0:%.*]], <16 x i16> [[B0:%.*]]) local_unnamed_addr #[[ATTR1]] {
; CHECK-NEXT:    [[A:%.*]] = sext <16 x i16> [[A0]] to <16 x i32>
; CHECK-NEXT:    [[B:%.*]] = sext <16 x i16> [[B0]] to <16 x i32>
; CHECK-NEXT:    [[M:%.*]] = mul nsw <16 x i32> [[B]], [[A]]
; CHECK-NEXT:    [[ODD:%.*]] = shufflevector <16 x i32> [[M]], <16 x i32> poison, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
; CHECK-NEXT:    [[EVEN:%.*]] = shufflevector <16 x i32> [[M]], <16 x i32> poison, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
; CHECK-NEXT:    [[RET:%.*]] = add <8 x i32> [[ODD]], [[EVEN]]
; CHECK-NEXT:    ret <8 x i32> [[RET]]
;
  %a = sext <16 x i16> %a0 to <16 x i32>
  %b = sext <16 x i16> %b0 to <16 x i32>
  %m = mul nsw <16 x i32> %a, %b
  %odd = shufflevector <16 x i32> %m, <16 x i32> undef, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
  %even = shufflevector <16 x i32> %m, <16 x i32> undef, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
  %ret = add <8 x i32> %odd, %even
  ret <8 x i32> %ret
}

define <16 x i32> @pmaddwd_32(<32 x i16> %a0, <32 x i16> %b0) {
; CHECK-LABEL: define <16 x i32> @pmaddwd_32(
; CHECK-SAME: <32 x i16> [[A0:%.*]], <32 x i16> [[B0:%.*]]) local_unnamed_addr #[[ATTR1]] {
; CHECK-NEXT:    [[A:%.*]] = sext <32 x i16> [[A0]] to <32 x i32>
; CHECK-NEXT:    [[B:%.*]] = sext <32 x i16> [[B0]] to <32 x i32>
; CHECK-NEXT:    [[M:%.*]] = mul nsw <32 x i32> [[B]], [[A]]
; CHECK-NEXT:    [[ODD:%.*]] = shufflevector <32 x i32> [[M]], <32 x i32> poison, <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30>
; CHECK-NEXT:    [[EVEN:%.*]] = shufflevector <32 x i32> [[M]], <32 x i32> poison, <16 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31>
; CHECK-NEXT:    [[RET:%.*]] = add <16 x i32> [[ODD]], [[EVEN]]
; CHECK-NEXT:    ret <16 x i32> [[RET]]
;
  %a = sext <32 x i16> %a0 to <32 x i32>
  %b = sext <32 x i16> %b0 to <32 x i32>
  %m = mul nsw <32 x i32> %a, %b
  %odd = shufflevector <32 x i32> %m, <32 x i32> undef, <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30>
  %even = shufflevector <32 x i32> %m, <32 x i32> undef, <16 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31>
  %ret = add <16 x i32> %odd, %even
  ret <16 x i32> %ret
}

define <4 x i32> @pmaddwd_const(<8 x i16> %a0) {
; CHECK-LABEL: define <4 x i32> @pmaddwd_const(
; CHECK-SAME: <8 x i16> [[A0:%.*]]) local_unnamed_addr #[[ATTR1]] {
; CHECK-NEXT:    [[A:%.*]] = sext <8 x i16> [[A0]] to <8 x i32>
; CHECK-NEXT:    [[M:%.*]] = mul nsw <8 x i32> [[A]], <i32 32767, i32 -32768, i32 0, i32 0, i32 1, i32 7, i32 42, i32 32>
; CHECK-NEXT:    [[ODD:%.*]] = shufflevector <8 x i32> [[M]], <8 x i32> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT:    [[EVEN:%.*]] = shufflevector <8 x i32> [[M]], <8 x i32> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; CHECK-NEXT:    [[RET:%.*]] = add <4 x i32> [[ODD]], [[EVEN]]
; CHECK-NEXT:    ret <4 x i32> [[RET]]
;
  %a = sext <8 x i16> %a0 to <8 x i32>
  %m = mul nsw <8 x i32> %a, <i32 32767, i32 -32768, i32 0, i32 0, i32 1, i32 7, i32 42, i32 32>
  %odd = shufflevector <8 x i32> %m, <8 x i32> undef, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
  %even = shufflevector <8 x i32> %m, <8 x i32> undef, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
  %ret = add <4 x i32> %odd, %even
  ret <4 x i32> %ret
}

; Do not select unsigned i16 multiplication
define <4 x i32> @pmaddwd_negative1(<8 x i16> %a0, <8 x i16> %b0) {
; CHECK-LABEL: define <4 x i32> @pmaddwd_negative1(
; CHECK-SAME: <8 x i16> [[A0:%.*]], <8 x i16> [[B0:%.*]]) local_unnamed_addr #[[ATTR1]] {
; CHECK-NEXT:    [[A:%.*]] = zext <8 x i16> [[A0]] to <8 x i32>
; CHECK-NEXT:    [[B:%.*]] = zext <8 x i16> [[B0]] to <8 x i32>
; CHECK-NEXT:    [[M:%.*]] = mul nuw <8 x i32> [[B]], [[A]]
; CHECK-NEXT:    [[ODD:%.*]] = shufflevector <8 x i32> [[M]], <8 x i32> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT:    [[EVEN:%.*]] = shufflevector <8 x i32> [[M]], <8 x i32> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; CHECK-NEXT:    [[RET:%.*]] = add <4 x i32> [[ODD]], [[EVEN]]
; CHECK-NEXT:    ret <4 x i32> [[RET]]
;
  %a = zext <8 x i16> %a0 to <8 x i32>
  %b = zext <8 x i16> %b0 to <8 x i32>
  %m = mul nuw <8 x i32> %a, %b
  %odd = shufflevector <8 x i32> %m, <8 x i32> undef, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
  %even = shufflevector <8 x i32> %m, <8 x i32> undef, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
  %ret = add <4 x i32> %odd, %even
  ret <4 x i32> %ret
}

; Do not select if constant is too large
; Lower half is too large, upper half is in range.
define <4 x i32> @pmaddwd_negative2(<8 x i16> %a0) {
; CHECK-LABEL: define <4 x i32> @pmaddwd_negative2(
; CHECK-SAME: <8 x i16> [[A0:%.*]]) local_unnamed_addr #[[ATTR1]] {
; CHECK-NEXT:    [[A:%.*]] = sext <8 x i16> [[A0]] to <8 x i32>
; CHECK-NEXT:    [[M:%.*]] = mul nsw <8 x i32> [[A]], <i32 32768, i32 -32768, i32 0, i32 0, i32 1, i32 7, i32 42, i32 32>
; CHECK-NEXT:    [[ODD:%.*]] = shufflevector <8 x i32> [[M]], <8 x i32> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT:    [[EVEN:%.*]] = shufflevector <8 x i32> [[M]], <8 x i32> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; CHECK-NEXT:    [[RET:%.*]] = add <4 x i32> [[ODD]], [[EVEN]]
; CHECK-NEXT:    ret <4 x i32> [[RET]]
;
  %a = sext <8 x i16> %a0 to <8 x i32>
  %m = mul nsw <8 x i32> %a, <i32 32768, i32 -32768, i32 0, i32 0, i32 1, i32 7, i32 42, i32 32>
  %odd = shufflevector <8 x i32> %m, <8 x i32> undef, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
  %even = shufflevector <8 x i32> %m, <8 x i32> undef, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
  %ret = add <4 x i32> %odd, %even
  ret <4 x i32> %ret
}

define <4 x i32> @jumbled_indices4(<8 x i16> %a0, <8 x i16> %b0) {
; CHECK-LABEL: define <4 x i32> @jumbled_indices4(
; CHECK-SAME: <8 x i16> [[A0:%.*]], <8 x i16> [[B0:%.*]]) local_unnamed_addr #[[ATTR1]] {
; CHECK-NEXT:    [[EXTA:%.*]] = sext <8 x i16> [[A0]] to <8 x i32>
; CHECK-NEXT:    [[EXTB:%.*]] = sext <8 x i16> [[B0]] to <8 x i32>
; CHECK-NEXT:    [[M:%.*]] = mul nsw <8 x i32> [[EXTB]], [[EXTA]]
; CHECK-NEXT:    [[SA:%.*]] = shufflevector <8 x i32> [[M]], <8 x i32> poison, <4 x i32> <i32 3, i32 1, i32 5, i32 6>
; CHECK-NEXT:    [[SB:%.*]] = shufflevector <8 x i32> [[M]], <8 x i32> poison, <4 x i32> <i32 2, i32 0, i32 4, i32 7>
; CHECK-NEXT:    [[A:%.*]] = add <4 x i32> [[SA]], [[SB]]
; CHECK-NEXT:    ret <4 x i32> [[A]]
;
  %exta = sext <8 x i16> %a0 to <8 x i32>
  %extb = sext <8 x i16> %b0 to <8 x i32>
  %m = mul <8 x i32> %exta, %extb
  %sa = shufflevector <8 x i32> %m, <8 x i32> undef, <4 x i32> <i32 3, i32 1, i32 5, i32 6>
  %sb = shufflevector <8 x i32> %m, <8 x i32> undef, <4 x i32> <i32 2, i32 0, i32 4, i32 7>
  %a = add <4 x i32> %sa, %sb
  ret <4 x i32> %a
}

define <8 x i32> @jumbled_indices8(<16 x i16> %a0, <16 x i16> %b0) {
; CHECK-LABEL: define <8 x i32> @jumbled_indices8(
; CHECK-SAME: <16 x i16> [[A0:%.*]], <16 x i16> [[B0:%.*]]) local_unnamed_addr #[[ATTR1]] {
; CHECK-NEXT:    [[EXTA:%.*]] = sext <16 x i16> [[A0]] to <16 x i32>
; CHECK-NEXT:    [[EXTB:%.*]] = sext <16 x i16> [[B0]] to <16 x i32>
; CHECK-NEXT:    [[M:%.*]] = mul nsw <16 x i32> [[EXTB]], [[EXTA]]
; CHECK-NEXT:    [[SA:%.*]] = shufflevector <16 x i32> [[M]], <16 x i32> poison, <8 x i32> <i32 0, i32 2, i32 7, i32 4, i32 11, i32 8, i32 15, i32 12>
; CHECK-NEXT:    [[SB:%.*]] = shufflevector <16 x i32> [[M]], <16 x i32> poison, <8 x i32> <i32 1, i32 3, i32 6, i32 5, i32 10, i32 9, i32 14, i32 13>
; CHECK-NEXT:    [[A:%.*]] = add <8 x i32> [[SA]], [[SB]]
; CHECK-NEXT:    ret <8 x i32> [[A]]
;
  %exta = sext <16 x i16> %a0 to <16 x i32>
  %extb = sext <16 x i16> %b0 to <16 x i32>
  %m = mul <16 x i32> %exta, %extb
  %sa = shufflevector <16 x i32> %m, <16 x i32> undef, <8 x i32> <i32 0, i32 2, i32 7, i32 4, i32 11, i32 8, i32 15, i32 12>
  %sb = shufflevector <16 x i32> %m, <16 x i32> undef, <8 x i32> <i32 1, i32 3, i32 6, i32 5, i32 10, i32 9, i32 14, i32 13>
  %a = add <8 x i32> %sa, %sb
  ret <8 x i32> %a
}

define <16 x i32> @jumbled_indices16(<32 x i16> %a0, <32 x i16> %b0) {
; CHECK-LABEL: define <16 x i32> @jumbled_indices16(
; CHECK-SAME: <32 x i16> [[A0:%.*]], <32 x i16> [[B0:%.*]]) local_unnamed_addr #[[ATTR1]] {
; CHECK-NEXT:    [[EXTA:%.*]] = sext <32 x i16> [[A0]] to <32 x i32>
; CHECK-NEXT:    [[EXTB:%.*]] = sext <32 x i16> [[B0]] to <32 x i32>
; CHECK-NEXT:    [[M:%.*]] = mul nsw <32 x i32> [[EXTB]], [[EXTA]]
; CHECK-NEXT:    [[SA:%.*]] = shufflevector <32 x i32> [[M]], <32 x i32> poison, <16 x i32> <i32 2, i32 0, i32 5, i32 6, i32 11, i32 9, i32 15, i32 12, i32 17, i32 18, i32 20, i32 23, i32 27, i32 24, i32 31, i32 29>
; CHECK-NEXT:    [[SB:%.*]] = shufflevector <32 x i32> [[M]], <32 x i32> poison, <16 x i32> <i32 3, i32 1, i32 4, i32 7, i32 10, i32 8, i32 14, i32 13, i32 16, i32 19, i32 21, i32 22, i32 26, i32 25, i32 30, i32 28>
; CHECK-NEXT:    [[A:%.*]] = add <16 x i32> [[SA]], [[SB]]
; CHECK-NEXT:    ret <16 x i32> [[A]]
;
  %exta = sext <32 x i16> %a0 to <32 x i32>
  %extb = sext <32 x i16> %b0 to <32 x i32>
  %m = mul <32 x i32> %exta, %extb
  %sa = shufflevector <32 x i32> %m, <32 x i32> undef, <16 x i32> <i32 2, i32 0, i32 5, i32 6, i32 11, i32 9, i32 15, i32 12, i32 17, i32 18, i32 20, i32 23, i32 27, i32 24, i32 31, i32 29>
  %sb = shufflevector <32 x i32> %m, <32 x i32> undef, <16 x i32> <i32 3, i32 1, i32 4, i32 7, i32 10, i32 8, i32 14, i32 13, i32 16, i32 19, i32 21, i32 22, i32 26, i32 25, i32 30, i32 28>
  %a = add <16 x i32> %sa, %sb
  ret <16 x i32> %a
}

define <32 x i32> @jumbled_indices32(<64 x i16> %a0, <64 x i16> %b0) {
; CHECK-LABEL: define <32 x i32> @jumbled_indices32(
; CHECK-SAME: <64 x i16> [[A0:%.*]], <64 x i16> [[B0:%.*]]) local_unnamed_addr #[[ATTR1]] {
; CHECK-NEXT:    [[EXTA:%.*]] = sext <64 x i16> [[A0]] to <64 x i32>
; CHECK-NEXT:    [[EXTB:%.*]] = sext <64 x i16> [[B0]] to <64 x i32>
; CHECK-NEXT:    [[M:%.*]] = mul nsw <64 x i32> [[EXTB]], [[EXTA]]
; CHECK-NEXT:    [[SA:%.*]] = shufflevector <64 x i32> [[M]], <64 x i32> poison, <32 x i32> <i32 1, i32 2, i32 6, i32 5, i32 10, i32 8, i32 14, i32 12, i32 19, i32 17, i32 22, i32 20, i32 25, i32 27, i32 30, i32 28, i32 32, i32 34, i32 37, i32 38, i32 41, i32 43, i32 45, i32 47, i32 50, i32 48, i32 52, i32 54, i32 59, i32 56, i32 61, i32 63>
; CHECK-NEXT:    [[SB:%.*]] = shufflevector <64 x i32> [[M]], <64 x i32> poison, <32 x i32> <i32 0, i32 3, i32 7, i32 4, i32 11, i32 9, i32 15, i32 13, i32 18, i32 16, i32 23, i32 21, i32 24, i32 26, i32 31, i32 29, i32 33, i32 35, i32 36, i32 39, i32 40, i32 42, i32 44, i32 46, i32 51, i32 49, i32 53, i32 55, i32 58, i32 57, i32 60, i32 62>
; CHECK-NEXT:    [[A:%.*]] = add <32 x i32> [[SA]], [[SB]]
; CHECK-NEXT:    ret <32 x i32> [[A]]
;
  %exta = sext <64 x i16> %a0 to <64 x i32>
  %extb = sext <64 x i16> %b0 to <64 x i32>
  %m = mul <64 x i32> %exta, %extb
  %sa = shufflevector <64 x i32> %m, <64 x i32> undef, <32 x i32> <i32 1, i32 2, i32 6, i32 5, i32 10, i32 8, i32 14, i32 12, i32 19, i32 17, i32 22, i32 20, i32 25, i32 27, i32 30, i32 28, i32 32, i32 34, i32 37, i32 38, i32 41, i32 43, i32 45, i32 47, i32 50, i32 48, i32 52, i32 54, i32 59, i32 56, i32 61, i32 63>
  %sb = shufflevector <64 x i32> %m, <64 x i32> undef, <32 x i32> <i32 0, i32 3, i32 7, i32 4, i32 11, i32 9, i32 15, i32 13, i32 18, i32 16, i32 23, i32 21, i32 24, i32 26, i32 31, i32 29, i32 33, i32 35, i32 36, i32 39, i32 40, i32 42, i32 44, i32 46, i32 51, i32 49, i32 53, i32 55, i32 58, i32 57, i32 60, i32 62>
  %a = add <32 x i32> %sa, %sb
  ret <32 x i32> %a
}

; NOTE: We're testing with loads because ABI lowering creates a concat_vectors that extract_vector_elt creation can see through.
; This would require the combine to recreate the concat_vectors.
define <4 x i32> @pmaddwd_128(ptr %Aptr, ptr %Bptr) {
; CHECK-LABEL: define range(i32 -2147418112, -2147483647) <4 x i32> @pmaddwd_128(
; CHECK-SAME: ptr readonly captures(none) [[APTR:%.*]], ptr readonly captures(none) [[BPTR:%.*]]) local_unnamed_addr #[[ATTR2:[0-9]+]] {
; CHECK-NEXT:    [[A:%.*]] = load <8 x i16>, ptr [[APTR]], align 16
; CHECK-NEXT:    [[B:%.*]] = load <8 x i16>, ptr [[BPTR]], align 16
; CHECK-NEXT:    [[A_EVEN:%.*]] = shufflevector <8 x i16> [[A]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT:    [[A_ODD:%.*]] = shufflevector <8 x i16> [[A]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; CHECK-NEXT:    [[B_EVEN:%.*]] = shufflevector <8 x i16> [[B]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT:    [[B_ODD:%.*]] = shufflevector <8 x i16> [[B]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; CHECK-NEXT:    [[A_EVEN_EXT:%.*]] = sext <4 x i16> [[A_EVEN]] to <4 x i32>
; CHECK-NEXT:    [[B_EVEN_EXT:%.*]] = sext <4 x i16> [[B_EVEN]] to <4 x i32>
; CHECK-NEXT:    [[A_ODD_EXT:%.*]] = sext <4 x i16> [[A_ODD]] to <4 x i32>
; CHECK-NEXT:    [[B_ODD_EXT:%.*]] = sext <4 x i16> [[B_ODD]] to <4 x i32>
; CHECK-NEXT:    [[EVEN_MUL:%.*]] = mul nsw <4 x i32> [[B_EVEN_EXT]], [[A_EVEN_EXT]]
; CHECK-NEXT:    [[ODD_MUL:%.*]] = mul nsw <4 x i32> [[B_ODD_EXT]], [[A_ODD_EXT]]
; CHECK-NEXT:    [[ADD:%.*]] = add <4 x i32> [[EVEN_MUL]], [[ODD_MUL]]
; CHECK-NEXT:    ret <4 x i32> [[ADD]]
;
  %A = load <8 x i16>, ptr %Aptr
  %B = load <8 x i16>, ptr %Bptr
  %A_even = shufflevector <8 x i16> %A, <8 x i16> undef, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
  %A_odd = shufflevector <8 x i16> %A, <8 x i16> undef, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
  %B_even = shufflevector <8 x i16> %B, <8 x i16> undef, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
  %B_odd = shufflevector <8 x i16> %B, <8 x i16> undef, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
  %A_even_ext = sext <4 x i16> %A_even to <4 x i32>
  %B_even_ext = sext <4 x i16> %B_even to <4 x i32>
  %A_odd_ext = sext <4 x i16> %A_odd to <4 x i32>
  %B_odd_ext = sext <4 x i16> %B_odd to <4 x i32>
  %even_mul = mul <4 x i32> %A_even_ext, %B_even_ext
  %odd_mul = mul <4 x i32> %A_odd_ext, %B_odd_ext
  %add = add <4 x i32> %even_mul, %odd_mul
  ret <4 x i32> %add
}

define <8 x i32> @pmaddwd_256(ptr %Aptr, ptr %Bptr) {
; CHECK-LABEL: define range(i32 -2147418112, -2147483647) <8 x i32> @pmaddwd_256(
; CHECK-SAME: ptr readonly captures(none) [[APTR:%.*]], ptr readonly captures(none) [[BPTR:%.*]]) local_unnamed_addr #[[ATTR2]] {
; CHECK-NEXT:    [[A:%.*]] = load <16 x i16>, ptr [[APTR]], align 32
; CHECK-NEXT:    [[B:%.*]] = load <16 x i16>, ptr [[BPTR]], align 32
; CHECK-NEXT:    [[A_EVEN:%.*]] = shufflevector <16 x i16> [[A]], <16 x i16> poison, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
; CHECK-NEXT:    [[A_ODD:%.*]] = shufflevector <16 x i16> [[A]], <16 x i16> poison, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
; CHECK-NEXT:    [[B_EVEN:%.*]] = shufflevector <16 x i16> [[B]], <16 x i16> poison, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
; CHECK-NEXT:    [[B_ODD:%.*]] = shufflevector <16 x i16> [[B]], <16 x i16> poison, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
; CHECK-NEXT:    [[A_EVEN_EXT:%.*]] = sext <8 x i16> [[A_EVEN]] to <8 x i32>
; CHECK-NEXT:    [[B_EVEN_EXT:%.*]] = sext <8 x i16> [[B_EVEN]] to <8 x i32>
; CHECK-NEXT:    [[A_ODD_EXT:%.*]] = sext <8 x i16> [[A_ODD]] to <8 x i32>
; CHECK-NEXT:    [[B_ODD_EXT:%.*]] = sext <8 x i16> [[B_ODD]] to <8 x i32>
; CHECK-NEXT:    [[EVEN_MUL:%.*]] = mul nsw <8 x i32> [[B_EVEN_EXT]], [[A_EVEN_EXT]]
; CHECK-NEXT:    [[ODD_MUL:%.*]] = mul nsw <8 x i32> [[B_ODD_EXT]], [[A_ODD_EXT]]
; CHECK-NEXT:    [[ADD:%.*]] = add <8 x i32> [[EVEN_MUL]], [[ODD_MUL]]
; CHECK-NEXT:    ret <8 x i32> [[ADD]]
;
  %A = load <16 x i16>, ptr %Aptr
  %B = load <16 x i16>, ptr %Bptr
  %A_even = shufflevector <16 x i16> %A, <16 x i16> undef, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
  %A_odd = shufflevector <16 x i16> %A, <16 x i16> undef, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
  %B_even = shufflevector <16 x i16> %B, <16 x i16> undef, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
  %B_odd = shufflevector <16 x i16> %B, <16 x i16> undef, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
  %A_even_ext = sext <8 x i16> %A_even to <8 x i32>
  %B_even_ext = sext <8 x i16> %B_even to <8 x i32>
  %A_odd_ext = sext <8 x i16> %A_odd to <8 x i32>
  %B_odd_ext = sext <8 x i16> %B_odd to <8 x i32>
  %even_mul = mul <8 x i32> %A_even_ext, %B_even_ext
  %odd_mul = mul <8 x i32> %A_odd_ext, %B_odd_ext
  %add = add <8 x i32> %even_mul, %odd_mul
  ret <8 x i32> %add
}

define <16 x i32> @pmaddwd_512(ptr %Aptr, ptr %Bptr) {
; CHECK-LABEL: define range(i32 -2147418112, -2147483647) <16 x i32> @pmaddwd_512(
; CHECK-SAME: ptr readonly captures(none) [[APTR:%.*]], ptr readonly captures(none) [[BPTR:%.*]]) local_unnamed_addr #[[ATTR2]] {
; CHECK-NEXT:    [[A:%.*]] = load <32 x i16>, ptr [[APTR]], align 64
; CHECK-NEXT:    [[B:%.*]] = load <32 x i16>, ptr [[BPTR]], align 64
; CHECK-NEXT:    [[A_EVEN:%.*]] = shufflevector <32 x i16> [[A]], <32 x i16> poison, <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30>
; CHECK-NEXT:    [[A_ODD:%.*]] = shufflevector <32 x i16> [[A]], <32 x i16> poison, <16 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31>
; CHECK-NEXT:    [[B_EVEN:%.*]] = shufflevector <32 x i16> [[B]], <32 x i16> poison, <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30>
; CHECK-NEXT:    [[B_ODD:%.*]] = shufflevector <32 x i16> [[B]], <32 x i16> poison, <16 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31>
; CHECK-NEXT:    [[A_EVEN_EXT:%.*]] = sext <16 x i16> [[A_EVEN]] to <16 x i32>
; CHECK-NEXT:    [[B_EVEN_EXT:%.*]] = sext <16 x i16> [[B_EVEN]] to <16 x i32>
; CHECK-NEXT:    [[A_ODD_EXT:%.*]] = sext <16 x i16> [[A_ODD]] to <16 x i32>
; CHECK-NEXT:    [[B_ODD_EXT:%.*]] = sext <16 x i16> [[B_ODD]] to <16 x i32>
; CHECK-NEXT:    [[EVEN_MUL:%.*]] = mul nsw <16 x i32> [[B_EVEN_EXT]], [[A_EVEN_EXT]]
; CHECK-NEXT:    [[ODD_MUL:%.*]] = mul nsw <16 x i32> [[B_ODD_EXT]], [[A_ODD_EXT]]
; CHECK-NEXT:    [[ADD:%.*]] = add <16 x i32> [[EVEN_MUL]], [[ODD_MUL]]
; CHECK-NEXT:    ret <16 x i32> [[ADD]]
;
  %A = load <32 x i16>, ptr %Aptr
  %B = load <32 x i16>, ptr %Bptr
  %A_even = shufflevector <32 x i16> %A, <32 x i16> undef, <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30>
  %A_odd = shufflevector <32 x i16> %A, <32 x i16> undef, <16 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31>
  %B_even = shufflevector <32 x i16> %B, <32 x i16> undef, <16 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30>
  %B_odd = shufflevector <32 x i16> %B, <32 x i16> undef, <16 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31>
  %A_even_ext = sext <16 x i16> %A_even to <16 x i32>
  %B_even_ext = sext <16 x i16> %B_even to <16 x i32>
  %A_odd_ext = sext <16 x i16> %A_odd to <16 x i32>
  %B_odd_ext = sext <16 x i16> %B_odd to <16 x i32>
  %even_mul = mul <16 x i32> %A_even_ext, %B_even_ext
  %odd_mul = mul <16 x i32> %A_odd_ext, %B_odd_ext
  %add = add <16 x i32> %even_mul, %odd_mul
  ret <16 x i32> %add
}

define <32 x i32> @pmaddwd_1024(ptr %Aptr, ptr %Bptr) {
; CHECK-LABEL: define range(i32 -2147418112, -2147483647) <32 x i32> @pmaddwd_1024(
; CHECK-SAME: ptr readonly captures(none) [[APTR:%.*]], ptr readonly captures(none) [[BPTR:%.*]]) local_unnamed_addr #[[ATTR2]] {
; CHECK-NEXT:    [[A:%.*]] = load <64 x i16>, ptr [[APTR]], align 128
; CHECK-NEXT:    [[B:%.*]] = load <64 x i16>, ptr [[BPTR]], align 128
; CHECK-NEXT:    [[A_EVEN:%.*]] = shufflevector <64 x i16> [[A]], <64 x i16> poison, <32 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30, i32 32, i32 34, i32 36, i32 38, i32 40, i32 42, i32 44, i32 46, i32 48, i32 50, i32 52, i32 54, i32 56, i32 58, i32 60, i32 62>
; CHECK-NEXT:    [[A_ODD:%.*]] = shufflevector <64 x i16> [[A]], <64 x i16> poison, <32 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31, i32 33, i32 35, i32 37, i32 39, i32 41, i32 43, i32 45, i32 47, i32 49, i32 51, i32 53, i32 55, i32 57, i32 59, i32 61, i32 63>
; CHECK-NEXT:    [[B_EVEN:%.*]] = shufflevector <64 x i16> [[B]], <64 x i16> poison, <32 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30, i32 32, i32 34, i32 36, i32 38, i32 40, i32 42, i32 44, i32 46, i32 48, i32 50, i32 52, i32 54, i32 56, i32 58, i32 60, i32 62>
; CHECK-NEXT:    [[B_ODD:%.*]] = shufflevector <64 x i16> [[B]], <64 x i16> poison, <32 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31, i32 33, i32 35, i32 37, i32 39, i32 41, i32 43, i32 45, i32 47, i32 49, i32 51, i32 53, i32 55, i32 57, i32 59, i32 61, i32 63>
; CHECK-NEXT:    [[A_EVEN_EXT:%.*]] = sext <32 x i16> [[A_EVEN]] to <32 x i32>
; CHECK-NEXT:    [[B_EVEN_EXT:%.*]] = sext <32 x i16> [[B_EVEN]] to <32 x i32>
; CHECK-NEXT:    [[A_ODD_EXT:%.*]] = sext <32 x i16> [[A_ODD]] to <32 x i32>
; CHECK-NEXT:    [[B_ODD_EXT:%.*]] = sext <32 x i16> [[B_ODD]] to <32 x i32>
; CHECK-NEXT:    [[EVEN_MUL:%.*]] = mul nsw <32 x i32> [[B_EVEN_EXT]], [[A_EVEN_EXT]]
; CHECK-NEXT:    [[ODD_MUL:%.*]] = mul nsw <32 x i32> [[B_ODD_EXT]], [[A_ODD_EXT]]
; CHECK-NEXT:    [[ADD:%.*]] = add <32 x i32> [[EVEN_MUL]], [[ODD_MUL]]
; CHECK-NEXT:    ret <32 x i32> [[ADD]]
;
  %A = load <64 x i16>, ptr %Aptr
  %B = load <64 x i16>, ptr %Bptr
  %A_even = shufflevector <64 x i16> %A, <64 x i16> undef, <32 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30, i32 32, i32 34, i32 36, i32 38, i32 40, i32 42, i32 44, i32 46, i32 48, i32 50, i32 52, i32 54, i32 56, i32 58, i32 60, i32 62>
  %A_odd = shufflevector <64 x i16> %A, <64 x i16> undef, <32 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31, i32 33, i32 35, i32 37, i32 39, i32 41, i32 43, i32 45, i32 47, i32 49, i32 51, i32 53, i32 55, i32 57, i32 59, i32 61, i32 63>
  %B_even = shufflevector <64 x i16> %B, <64 x i16> undef, <32 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14, i32 16, i32 18, i32 20, i32 22, i32 24, i32 26, i32 28, i32 30, i32 32, i32 34, i32 36, i32 38, i32 40, i32 42, i32 44, i32 46, i32 48, i32 50, i32 52, i32 54, i32 56, i32 58, i32 60, i32 62>
  %B_odd = shufflevector <64 x i16> %B, <64 x i16> undef, <32 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15, i32 17, i32 19, i32 21, i32 23, i32 25, i32 27, i32 29, i32 31, i32 33, i32 35, i32 37, i32 39, i32 41, i32 43, i32 45, i32 47, i32 49, i32 51, i32 53, i32 55, i32 57, i32 59, i32 61, i32 63>
  %A_even_ext = sext <32 x i16> %A_even to <32 x i32>
  %B_even_ext = sext <32 x i16> %B_even to <32 x i32>
  %A_odd_ext = sext <32 x i16> %A_odd to <32 x i32>
  %B_odd_ext = sext <32 x i16> %B_odd to <32 x i32>
  %even_mul = mul <32 x i32> %A_even_ext, %B_even_ext
  %odd_mul = mul <32 x i32> %A_odd_ext, %B_odd_ext
  %add = add <32 x i32> %even_mul, %odd_mul
  ret <32 x i32> %add
}

define <4 x i32> @pmaddwd_commuted_mul(ptr %Aptr, ptr %Bptr) {
; CHECK-LABEL: define range(i32 -2147418112, -2147483647) <4 x i32> @pmaddwd_commuted_mul(
; CHECK-SAME: ptr readonly captures(none) [[APTR:%.*]], ptr readonly captures(none) [[BPTR:%.*]]) local_unnamed_addr #[[ATTR2]] {
; CHECK-NEXT:    [[A:%.*]] = load <8 x i16>, ptr [[APTR]], align 16
; CHECK-NEXT:    [[B:%.*]] = load <8 x i16>, ptr [[BPTR]], align 16
; CHECK-NEXT:    [[A_EVEN:%.*]] = shufflevector <8 x i16> [[A]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT:    [[A_ODD:%.*]] = shufflevector <8 x i16> [[A]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; CHECK-NEXT:    [[B_EVEN:%.*]] = shufflevector <8 x i16> [[B]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT:    [[B_ODD:%.*]] = shufflevector <8 x i16> [[B]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; CHECK-NEXT:    [[A_EVEN_EXT:%.*]] = sext <4 x i16> [[A_EVEN]] to <4 x i32>
; CHECK-NEXT:    [[B_EVEN_EXT:%.*]] = sext <4 x i16> [[B_EVEN]] to <4 x i32>
; CHECK-NEXT:    [[A_ODD_EXT:%.*]] = sext <4 x i16> [[A_ODD]] to <4 x i32>
; CHECK-NEXT:    [[B_ODD_EXT:%.*]] = sext <4 x i16> [[B_ODD]] to <4 x i32>
; CHECK-NEXT:    [[EVEN_MUL:%.*]] = mul nsw <4 x i32> [[B_EVEN_EXT]], [[A_EVEN_EXT]]
; CHECK-NEXT:    [[ODD_MUL:%.*]] = mul nsw <4 x i32> [[B_ODD_EXT]], [[A_ODD_EXT]]
; CHECK-NEXT:    [[ADD:%.*]] = add <4 x i32> [[EVEN_MUL]], [[ODD_MUL]]
; CHECK-NEXT:    ret <4 x i32> [[ADD]]
;
  %A = load <8 x i16>, ptr %Aptr
  %B = load <8 x i16>, ptr %Bptr
  %A_even = shufflevector <8 x i16> %A, <8 x i16> undef, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
  %A_odd = shufflevector <8 x i16> %A, <8 x i16> undef, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
  %B_even = shufflevector <8 x i16> %B, <8 x i16> undef, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
  %B_odd = shufflevector <8 x i16> %B, <8 x i16> undef, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
  %A_even_ext = sext <4 x i16> %A_even to <4 x i32>
  %B_even_ext = sext <4 x i16> %B_even to <4 x i32>
  %A_odd_ext = sext <4 x i16> %A_odd to <4 x i32>
  %B_odd_ext = sext <4 x i16> %B_odd to <4 x i32>
  %even_mul = mul <4 x i32> %A_even_ext, %B_even_ext
  %odd_mul = mul <4 x i32> %B_odd_ext, %A_odd_ext ; Different order than previous mul
  %add = add <4 x i32> %even_mul, %odd_mul
  ret <4 x i32> %add
}

define <4 x i32> @pmaddwd_swapped_indices(ptr %Aptr, ptr %Bptr) {
; CHECK-LABEL: define range(i32 -2147418112, -2147483647) <4 x i32> @pmaddwd_swapped_indices(
; CHECK-SAME: ptr readonly captures(none) [[APTR:%.*]], ptr readonly captures(none) [[BPTR:%.*]]) local_unnamed_addr #[[ATTR2]] {
; CHECK-NEXT:    [[A:%.*]] = load <8 x i16>, ptr [[APTR]], align 16
; CHECK-NEXT:    [[B:%.*]] = load <8 x i16>, ptr [[BPTR]], align 16
; CHECK-NEXT:    [[A_EVEN:%.*]] = shufflevector <8 x i16> [[A]], <8 x i16> poison, <4 x i32> <i32 1, i32 2, i32 5, i32 6>
; CHECK-NEXT:    [[A_ODD:%.*]] = shufflevector <8 x i16> [[A]], <8 x i16> poison, <4 x i32> <i32 0, i32 3, i32 4, i32 7>
; CHECK-NEXT:    [[B_EVEN:%.*]] = shufflevector <8 x i16> [[B]], <8 x i16> poison, <4 x i32> <i32 1, i32 2, i32 5, i32 6>
; CHECK-NEXT:    [[B_ODD:%.*]] = shufflevector <8 x i16> [[B]], <8 x i16> poison, <4 x i32> <i32 0, i32 3, i32 4, i32 7>
; CHECK-NEXT:    [[A_EVEN_EXT:%.*]] = sext <4 x i16> [[A_EVEN]] to <4 x i32>
; CHECK-NEXT:    [[B_EVEN_EXT:%.*]] = sext <4 x i16> [[B_EVEN]] to <4 x i32>
; CHECK-NEXT:    [[A_ODD_EXT:%.*]] = sext <4 x i16> [[A_ODD]] to <4 x i32>
; CHECK-NEXT:    [[B_ODD_EXT:%.*]] = sext <4 x i16> [[B_ODD]] to <4 x i32>
; CHECK-NEXT:    [[EVEN_MUL:%.*]] = mul nsw <4 x i32> [[B_EVEN_EXT]], [[A_EVEN_EXT]]
; CHECK-NEXT:    [[ODD_MUL:%.*]] = mul nsw <4 x i32> [[B_ODD_EXT]], [[A_ODD_EXT]]
; CHECK-NEXT:    [[ADD:%.*]] = add <4 x i32> [[EVEN_MUL]], [[ODD_MUL]]
; CHECK-NEXT:    ret <4 x i32> [[ADD]]
;
  %A = load <8 x i16>, ptr %Aptr
  %B = load <8 x i16>, ptr %Bptr
  %A_even = shufflevector <8 x i16> %A, <8 x i16> undef, <4 x i32> <i32 1, i32 2, i32 5, i32 6> ; indices aren't all even
  %A_odd = shufflevector <8 x i16> %A, <8 x i16> undef, <4 x i32> <i32 0, i32 3, i32 4, i32 7> ; indices aren't all odd
  %B_even = shufflevector <8 x i16> %B, <8 x i16> undef, <4 x i32> <i32 1, i32 2, i32 5, i32 6> ; same indices as A
  %B_odd = shufflevector <8 x i16> %B, <8 x i16> undef, <4 x i32> <i32 0, i32 3, i32 4, i32 7> ; same indices as A
  %A_even_ext = sext <4 x i16> %A_even to <4 x i32>
  %B_even_ext = sext <4 x i16> %B_even to <4 x i32>
  %A_odd_ext = sext <4 x i16> %A_odd to <4 x i32>
  %B_odd_ext = sext <4 x i16> %B_odd to <4 x i32>
  %even_mul = mul <4 x i32> %A_even_ext, %B_even_ext
  %odd_mul = mul <4 x i32> %A_odd_ext, %B_odd_ext
  %add = add <4 x i32> %even_mul, %odd_mul
  ret <4 x i32> %add
}

; Negative test where indices aren't paired properly
define <4 x i32> @pmaddwd_bad_indices(ptr %Aptr, ptr %Bptr) {
; CHECK-LABEL: define range(i32 -2147418112, -2147483647) <4 x i32> @pmaddwd_bad_indices(
; CHECK-SAME: ptr readonly captures(none) [[APTR:%.*]], ptr readonly captures(none) [[BPTR:%.*]]) local_unnamed_addr #[[ATTR2]] {
; CHECK-NEXT:    [[A:%.*]] = load <8 x i16>, ptr [[APTR]], align 16
; CHECK-NEXT:    [[B:%.*]] = load <8 x i16>, ptr [[BPTR]], align 16
; CHECK-NEXT:    [[A_EVEN:%.*]] = shufflevector <8 x i16> [[A]], <8 x i16> poison, <4 x i32> <i32 1, i32 2, i32 5, i32 6>
; CHECK-NEXT:    [[A_ODD:%.*]] = shufflevector <8 x i16> [[A]], <8 x i16> poison, <4 x i32> <i32 0, i32 3, i32 4, i32 7>
; CHECK-NEXT:    [[B_EVEN:%.*]] = shufflevector <8 x i16> [[B]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT:    [[B_ODD:%.*]] = shufflevector <8 x i16> [[B]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; CHECK-NEXT:    [[A_EVEN_EXT:%.*]] = sext <4 x i16> [[A_EVEN]] to <4 x i32>
; CHECK-NEXT:    [[B_EVEN_EXT:%.*]] = sext <4 x i16> [[B_EVEN]] to <4 x i32>
; CHECK-NEXT:    [[A_ODD_EXT:%.*]] = sext <4 x i16> [[A_ODD]] to <4 x i32>
; CHECK-NEXT:    [[B_ODD_EXT:%.*]] = sext <4 x i16> [[B_ODD]] to <4 x i32>
; CHECK-NEXT:    [[EVEN_MUL:%.*]] = mul nsw <4 x i32> [[B_EVEN_EXT]], [[A_EVEN_EXT]]
; CHECK-NEXT:    [[ODD_MUL:%.*]] = mul nsw <4 x i32> [[B_ODD_EXT]], [[A_ODD_EXT]]
; CHECK-NEXT:    [[ADD:%.*]] = add <4 x i32> [[EVEN_MUL]], [[ODD_MUL]]
; CHECK-NEXT:    ret <4 x i32> [[ADD]]
;
  %A = load <8 x i16>, ptr %Aptr
  %B = load <8 x i16>, ptr %Bptr
  %A_even = shufflevector <8 x i16> %A, <8 x i16> undef, <4 x i32> <i32 1, i32 2, i32 5, i32 6>
  %A_odd = shufflevector <8 x i16> %A, <8 x i16> undef, <4 x i32> <i32 0, i32 3, i32 4, i32 7>
  %B_even = shufflevector <8 x i16> %B, <8 x i16> undef, <4 x i32> <i32 0, i32 2, i32 4, i32 6> ; different indices than A
  %B_odd = shufflevector <8 x i16> %B, <8 x i16> undef, <4 x i32> <i32 1, i32 3, i32 5, i32 7> ; different indices than A
  %A_even_ext = sext <4 x i16> %A_even to <4 x i32>
  %B_even_ext = sext <4 x i16> %B_even to <4 x i32>
  %A_odd_ext = sext <4 x i16> %A_odd to <4 x i32>
  %B_odd_ext = sext <4 x i16> %B_odd to <4 x i32>
  %even_mul = mul <4 x i32> %A_even_ext, %B_even_ext
  %odd_mul = mul <4 x i32> %A_odd_ext, %B_odd_ext
  %add = add <4 x i32> %even_mul, %odd_mul
  ret <4 x i32> %add
}

; This test contains two multiplies joined by an add. The result of that add is then reduced to a single element.
; SelectionDAGBuilder should tag the joining add as a vector reduction. We need to recognize that both sides can use pmaddwd
define i32 @madd_double_reduction(ptr %arg, ptr %arg1, ptr %arg2, ptr %arg3) {
; CHECK-LABEL: define i32 @madd_double_reduction(
; CHECK-SAME: ptr readonly captures(none) [[ARG:%.*]], ptr readonly captures(none) [[ARG1:%.*]], ptr readonly captures(none) [[ARG2:%.*]], ptr readonly captures(none) [[ARG3:%.*]]) local_unnamed_addr #[[ATTR2]] {
; CHECK-NEXT:    [[TMP:%.*]] = load <8 x i16>, ptr [[ARG]], align 1
; CHECK-NEXT:    [[TMP6:%.*]] = load <8 x i16>, ptr [[ARG1]], align 1
; CHECK-NEXT:    [[TMP7:%.*]] = sext <8 x i16> [[TMP]] to <8 x i32>
; CHECK-NEXT:    [[TMP17:%.*]] = sext <8 x i16> [[TMP6]] to <8 x i32>
; CHECK-NEXT:    [[TMP19:%.*]] = mul nsw <8 x i32> [[TMP17]], [[TMP7]]
; CHECK-NEXT:    [[TMP20:%.*]] = load <8 x i16>, ptr [[ARG2]], align 1
; CHECK-NEXT:    [[TMP21:%.*]] = load <8 x i16>, ptr [[ARG3]], align 1
; CHECK-NEXT:    [[TMP22:%.*]] = sext <8 x i16> [[TMP20]] to <8 x i32>
; CHECK-NEXT:    [[TMP23:%.*]] = sext <8 x i16> [[TMP21]] to <8 x i32>
; CHECK-NEXT:    [[TMP25:%.*]] = mul nsw <8 x i32> [[TMP23]], [[TMP22]]
; CHECK-NEXT:    [[TMP26:%.*]] = add nuw nsw <8 x i32> [[TMP25]], [[TMP19]]
; CHECK-NEXT:    [[TMP35:%.*]] = tail call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> [[TMP26]])
; CHECK-NEXT:    ret i32 [[TMP35]]
;
  %i = load <8 x i16>, ptr %arg, align 1
  %i6 = load <8 x i16>, ptr %arg1, align 1
  %i7 = sext <8 x i16> %i to <8 x i32>
  %i17 = sext <8 x i16> %i6 to <8 x i32>
  %i19 = mul nsw <8 x i32> %i7, %i17
  %i20 = load <8 x i16>, ptr %arg2, align 1
  %i21 = load <8 x i16>, ptr %arg3, align 1
  %i22 = sext <8 x i16> %i20 to <8 x i32>
  %i23 = sext <8 x i16> %i21 to <8 x i32>
  %i25 = mul nsw <8 x i32> %i22, %i23
  %i26 = add nuw nsw <8 x i32> %i25, %i19
  %i29 = shufflevector <8 x i32> %i26, <8 x i32> undef, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef>
  %i30 = add <8 x i32> %i26, %i29
  %i31 = shufflevector <8 x i32> %i30, <8 x i32> undef, <8 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %i32 = add <8 x i32> %i30, %i31
  %i33 = shufflevector <8 x i32> %i32, <8 x i32> undef, <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %i34 = add <8 x i32> %i32, %i33
  %i35 = extractelement <8 x i32> %i34, i64 0
  ret i32 %i35
}

define i32 @madd_quad_reduction(ptr %arg, ptr %arg1, ptr %arg2, ptr %arg3, ptr %arg4, ptr %arg5, ptr %arg6, ptr %arg7) {
; CHECK-LABEL: define i32 @madd_quad_reduction(
; CHECK-SAME: ptr readonly captures(none) [[ARG:%.*]], ptr readonly captures(none) [[ARG1:%.*]], ptr readonly captures(none) [[ARG2:%.*]], ptr readonly captures(none) [[ARG3:%.*]], ptr readonly captures(none) [[ARG4:%.*]], ptr readonly captures(none) [[ARG5:%.*]], ptr readonly captures(none) [[ARG6:%.*]], ptr readonly captures(none) [[ARG7:%.*]]) local_unnamed_addr #[[ATTR2]] {
; CHECK-NEXT:    [[TMP:%.*]] = load <8 x i16>, ptr [[ARG]], align 1
; CHECK-NEXT:    [[TMP6:%.*]] = load <8 x i16>, ptr [[ARG1]], align 1
; CHECK-NEXT:    [[TMP7:%.*]] = sext <8 x i16> [[TMP]] to <8 x i32>
; CHECK-NEXT:    [[TMP17:%.*]] = sext <8 x i16> [[TMP6]] to <8 x i32>
; CHECK-NEXT:    [[TMP19:%.*]] = mul nsw <8 x i32> [[TMP17]], [[TMP7]]
; CHECK-NEXT:    [[TMP20:%.*]] = load <8 x i16>, ptr [[ARG2]], align 1
; CHECK-NEXT:    [[TMP21:%.*]] = load <8 x i16>, ptr [[ARG3]], align 1
; CHECK-NEXT:    [[TMP22:%.*]] = sext <8 x i16> [[TMP20]] to <8 x i32>
; CHECK-NEXT:    [[TMP23:%.*]] = sext <8 x i16> [[TMP21]] to <8 x i32>
; CHECK-NEXT:    [[TMP25:%.*]] = mul nsw <8 x i32> [[TMP23]], [[TMP22]]
; CHECK-NEXT:    [[TMP26:%.*]] = add nuw nsw <8 x i32> [[TMP25]], [[TMP19]]
; CHECK-NEXT:    [[TMP40:%.*]] = load <8 x i16>, ptr [[ARG4]], align 1
; CHECK-NEXT:    [[TMP41:%.*]] = load <8 x i16>, ptr [[ARG5]], align 1
; CHECK-NEXT:    [[TMP42:%.*]] = sext <8 x i16> [[TMP40]] to <8 x i32>
; CHECK-NEXT:    [[TMP43:%.*]] = sext <8 x i16> [[TMP41]] to <8 x i32>
; CHECK-NEXT:    [[TMP45:%.*]] = mul nsw <8 x i32> [[TMP43]], [[TMP42]]
; CHECK-NEXT:    [[TMP56:%.*]] = add nuw nsw <8 x i32> [[TMP26]], [[TMP45]]
; CHECK-NEXT:    [[TMP50:%.*]] = load <8 x i16>, ptr [[ARG6]], align 1
; CHECK-NEXT:    [[TMP51:%.*]] = load <8 x i16>, ptr [[ARG7]], align 1
; CHECK-NEXT:    [[TMP52:%.*]] = sext <8 x i16> [[TMP50]] to <8 x i32>
; CHECK-NEXT:    [[TMP53:%.*]] = sext <8 x i16> [[TMP51]] to <8 x i32>
; CHECK-NEXT:    [[TMP55:%.*]] = mul nsw <8 x i32> [[TMP53]], [[TMP52]]
; CHECK-NEXT:    [[TMP57:%.*]] = add nuw nsw <8 x i32> [[TMP56]], [[TMP55]]
; CHECK-NEXT:    [[TMP35:%.*]] = tail call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> [[TMP57]])
; CHECK-NEXT:    ret i32 [[TMP35]]
;
  %i = load <8 x i16>, ptr %arg, align 1
  %i6 = load <8 x i16>, ptr %arg1, align 1
  %i7 = sext <8 x i16> %i to <8 x i32>
  %i17 = sext <8 x i16> %i6 to <8 x i32>
  %i19 = mul nsw <8 x i32> %i7, %i17
  %i20 = load <8 x i16>, ptr %arg2, align 1
  %i21 = load <8 x i16>, ptr %arg3, align 1
  %i22 = sext <8 x i16> %i20 to <8 x i32>
  %i23 = sext <8 x i16> %i21 to <8 x i32>
  %i25 = mul nsw <8 x i32> %i22, %i23
  %i26 = add nuw nsw <8 x i32> %i25, %i19

  %i40 = load <8 x i16>, ptr %arg4, align 1
  %i41 = load <8 x i16>, ptr %arg5, align 1
  %i42 = sext <8 x i16> %i40 to <8 x i32>
  %i43 = sext <8 x i16> %i41 to <8 x i32>
  %i45 = mul nsw <8 x i32> %i42, %i43
  %i56 = add nuw nsw <8 x i32> %i26, %i45

  %i50 = load <8 x i16>, ptr %arg6, align 1
  %i51 = load <8 x i16>, ptr %arg7, align 1
  %i52 = sext <8 x i16> %i50 to <8 x i32>
  %i53 = sext <8 x i16> %i51 to <8 x i32>
  %i55 = mul nsw <8 x i32> %i52, %i53
  %i57 = add nuw nsw <8 x i32> %i55, %i56

  %i29 = shufflevector <8 x i32> %i57, <8 x i32> undef, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef>
  %i30 = add <8 x i32> %i57, %i29
  %i31 = shufflevector <8 x i32> %i30, <8 x i32> undef, <8 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %i32 = add <8 x i32> %i30, %i31
  %i33 = shufflevector <8 x i32> %i32, <8 x i32> undef, <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %i34 = add <8 x i32> %i32, %i33
  %i35 = extractelement <8 x i32> %i34, i64 0
  ret i32 %i35
}

define i64 @sum_and_sum_of_squares(ptr %a, i32 %n) {
; CHECK-LABEL: define i64 @sum_and_sum_of_squares(
; CHECK-SAME: ptr readonly captures(none) [[A:%.*]], i32 [[N:%.*]]) local_unnamed_addr #[[ATTR0]] {
; CHECK-NEXT:  [[ENTRY:.*]]:
; CHECK-NEXT:    [[TMP0:%.*]] = zext i32 [[N]] to i64
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <8 x i32> [ [[TMP5:%.*]], %[[VECTOR_BODY]] ], [ zeroinitializer, %[[ENTRY]] ]
; CHECK-NEXT:    [[SUM_PHI:%.*]] = phi <8 x i32> [ [[TMP3:%.*]], %[[VECTOR_BODY]] ], [ zeroinitializer, %[[ENTRY]] ]
; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <8 x i8>, ptr [[TMP1]], align 1
; CHECK-NEXT:    [[TMP2:%.*]] = zext <8 x i8> [[WIDE_LOAD]] to <8 x i32>
; CHECK-NEXT:    [[TMP3]] = add nuw nsw <8 x i32> [[SUM_PHI]], [[TMP2]]
; CHECK-NEXT:    [[TMP4:%.*]] = mul nuw nsw <8 x i32> [[TMP2]], [[TMP2]]
; CHECK-NEXT:    [[TMP5]] = add nuw nsw <8 x i32> [[TMP4]], [[VEC_PHI]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 8
; CHECK-NEXT:    [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[TMP0]]
; CHECK-NEXT:    br i1 [[TMP6]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    [[TMP7:%.*]] = tail call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> [[TMP3]])
; CHECK-NEXT:    [[TMP8:%.*]] = tail call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> [[TMP5]])
; CHECK-NEXT:    [[TMP:%.*]] = zext i32 [[TMP7]] to i64
; CHECK-NEXT:    [[TMP28:%.*]] = shl nuw i64 [[TMP]], 32
; CHECK-NEXT:    [[TMP29:%.*]] = zext i32 [[TMP8]] to i64
; CHECK-NEXT:    [[TMP30:%.*]] = or disjoint i64 [[TMP28]], [[TMP29]]
; CHECK-NEXT:    ret i64 [[TMP30]]
;
entry:
  %i0 = zext i32 %n to i64
  br label %vector.body

vector.body:
  %index = phi i64 [ %index.next, %vector.body ], [ 0, %entry ]
  %vec.phi = phi <8 x i32> [ %i6, %vector.body ], [ zeroinitializer, %entry ]
  %sum.phi = phi <8 x i32> [ %i4, %vector.body ], [ zeroinitializer, %entry ]
  %i1 = getelementptr inbounds i8, ptr %a, i64 %index
  %i2 = bitcast ptr %i1 to ptr
  %wide.load = load <8 x i8>, ptr %i2, align 1
  %i3 = zext <8 x i8> %wide.load to <8 x i32>
  %i4 = add nsw <8 x i32> %i3, %sum.phi
  %i5 = mul nsw <8 x i32> %i3, %i3
  %i6 = add nsw <8 x i32> %i5, %vec.phi
  %index.next = add i64 %index, 8
  %i7 = icmp eq i64 %index.next, %i0
  br i1 %i7, label %middle.block, label %vector.body

middle.block:
  %rdx.shuf35 = shufflevector <8 x i32> %i4, <8 x i32> undef, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx36 = add <8 x i32> %i4, %rdx.shuf35
  %rdx.shuf37 = shufflevector <8 x i32> %bin.rdx36, <8 x i32> undef, <8 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx38 = add <8 x i32> %bin.rdx36, %rdx.shuf37
  %rdx.shuf39 = shufflevector <8 x i32> %bin.rdx38, <8 x i32> undef, <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx40 = add <8 x i32> %bin.rdx38, %rdx.shuf39
  %i8 = extractelement <8 x i32> %bin.rdx40, i32 0
  %rdx.shuf = shufflevector <8 x i32> %i6, <8 x i32> undef, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx = add <8 x i32> %i6, %rdx.shuf
  %rdx.shuf31 = shufflevector <8 x i32> %bin.rdx, <8 x i32> undef, <8 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx32 = add <8 x i32> %bin.rdx, %rdx.shuf31
  %rdx.shuf33 = shufflevector <8 x i32> %bin.rdx32, <8 x i32> undef, <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx34 = add <8 x i32> %bin.rdx32, %rdx.shuf33
  %i9 = extractelement <8 x i32> %bin.rdx34, i32 0
  %i = zext i32 %i8 to i64
  %i28 = shl nuw i64 %i, 32
  %i29 = zext i32 %i9 to i64
  %i30 = or i64 %i28, %i29
  ret i64 %i30
}

define i32 @sum_of_square_differences(ptr %a, ptr %b, i32 %n) {
; CHECK-LABEL: define i32 @sum_of_square_differences(
; CHECK-SAME: ptr readonly captures(none) [[A:%.*]], ptr readonly captures(none) [[B:%.*]], i32 [[N:%.*]]) local_unnamed_addr #[[ATTR0]] {
; CHECK-NEXT:  [[ENTRY:.*]]:
; CHECK-NEXT:    [[TMP0:%.*]] = zext i32 [[N]] to i64
; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
; CHECK:       [[VECTOR_BODY]]:
; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT:    [[VEC_PHI:%.*]] = phi <8 x i32> [ [[TMP7:%.*]], %[[VECTOR_BODY]] ], [ zeroinitializer, %[[ENTRY]] ]
; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD:%.*]] = load <8 x i8>, ptr [[TMP1]], align 1
; CHECK-NEXT:    [[TMP2:%.*]] = zext <8 x i8> [[WIDE_LOAD]] to <8 x i32>
; CHECK-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[INDEX]]
; CHECK-NEXT:    [[WIDE_LOAD2:%.*]] = load <8 x i8>, ptr [[TMP3]], align 1
; CHECK-NEXT:    [[TMP4:%.*]] = zext <8 x i8> [[WIDE_LOAD2]] to <8 x i32>
; CHECK-NEXT:    [[TMP5:%.*]] = sub nsw <8 x i32> [[TMP4]], [[TMP2]]
; CHECK-NEXT:    [[TMP6:%.*]] = mul nsw <8 x i32> [[TMP5]], [[TMP5]]
; CHECK-NEXT:    [[TMP7]] = add nuw nsw <8 x i32> [[TMP6]], [[VEC_PHI]]
; CHECK-NEXT:    [[INDEX_NEXT]] = add i64 [[INDEX]], 8
; CHECK-NEXT:    [[TMP8:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[TMP0]]
; CHECK-NEXT:    br i1 [[TMP8]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]]
; CHECK:       [[MIDDLE_BLOCK]]:
; CHECK-NEXT:    [[TMP9:%.*]] = tail call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> [[TMP7]])
; CHECK-NEXT:    ret i32 [[TMP9]]
;
entry:
  %i0 = zext i32 %n to i64
  br label %vector.body

vector.body:
  %index = phi i64 [ %index.next, %vector.body ], [ 0, %entry ]
  %vec.phi = phi <8 x i32> [ %i9, %vector.body ], [ zeroinitializer, %entry ]
  %i1 = getelementptr inbounds i8, ptr %a, i64 %index
  %i2 = bitcast ptr %i1 to ptr
  %wide.load = load <8 x i8>, ptr %i2, align 1
  %i3 = zext <8 x i8> %wide.load to <8 x i32>
  %i4 = getelementptr inbounds i8, ptr %b, i64 %index
  %i5 = bitcast ptr %i4 to ptr
  %wide.load2 = load <8 x i8>, ptr %i5, align 1
  %i6 = zext <8 x i8> %wide.load2 to <8 x i32>
  %i7 = sub <8 x i32> %i6, %i3
  %i8 = mul <8 x i32> %i7, %i7
  %i9 = add nsw <8 x i32> %i8, %vec.phi
  %index.next = add i64 %index, 8
  %i10 = icmp eq i64 %index.next, %i0
  br i1 %i10, label %middle.block, label %vector.body

middle.block:
  %rdx.shuf = shufflevector <8 x i32> %i9, <8 x i32> undef, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx = add <8 x i32> %i9, %rdx.shuf
  %rdx.shuf31 = shufflevector <8 x i32> %bin.rdx, <8 x i32> undef, <8 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx32 = add <8 x i32> %bin.rdx, %rdx.shuf31
  %rdx.shuf33 = shufflevector <8 x i32> %bin.rdx32, <8 x i32> undef, <8 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx34 = add <8 x i32> %bin.rdx32, %rdx.shuf33
  %i11 = extractelement <8 x i32> %bin.rdx34, i32 0
  ret i32 %i11
}

; PR49716 - https://llvm.org/PR49716

define <4 x i32> @input_size_mismatch(<16 x i16> %x, ptr %p) {
; CHECK-LABEL: define range(i32 -2147418112, -2147483647) <4 x i32> @input_size_mismatch(
; CHECK-SAME: <16 x i16> [[X:%.*]], ptr readonly captures(none) [[P:%.*]]) local_unnamed_addr #[[ATTR2]] {
; CHECK-NEXT:    [[TMP1:%.*]] = load <8 x i16>, ptr [[P]], align 32
; CHECK-NEXT:    [[X0:%.*]] = shufflevector <16 x i16> [[X]], <16 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT:    [[X1:%.*]] = shufflevector <16 x i16> [[X]], <16 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; CHECK-NEXT:    [[Y0:%.*]] = shufflevector <8 x i16> [[TMP1]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT:    [[Y1:%.*]] = shufflevector <8 x i16> [[TMP1]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; CHECK-NEXT:    [[SX0:%.*]] = sext <4 x i16> [[X0]] to <4 x i32>
; CHECK-NEXT:    [[SX1:%.*]] = sext <4 x i16> [[X1]] to <4 x i32>
; CHECK-NEXT:    [[SY0:%.*]] = sext <4 x i16> [[Y0]] to <4 x i32>
; CHECK-NEXT:    [[SY1:%.*]] = sext <4 x i16> [[Y1]] to <4 x i32>
; CHECK-NEXT:    [[M0:%.*]] = mul nsw <4 x i32> [[SY0]], [[SX0]]
; CHECK-NEXT:    [[M1:%.*]] = mul nsw <4 x i32> [[SY1]], [[SX1]]
; CHECK-NEXT:    [[R:%.*]] = add <4 x i32> [[M0]], [[M1]]
; CHECK-NEXT:    ret <4 x i32> [[R]]
;
  %y = load <16 x i16>, ptr %p, align 32
  %x0 = shufflevector <16 x i16> %x, <16 x i16> undef, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
  %x1 = shufflevector <16 x i16> %x, <16 x i16> undef, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
  %y0 = shufflevector <16 x i16> %y, <16 x i16> undef, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
  %y1 = shufflevector <16 x i16> %y, <16 x i16> undef, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
  %sx0 = sext <4 x i16> %x0 to <4 x i32>
  %sx1 = sext <4 x i16> %x1 to <4 x i32>
  %sy0 = sext <4 x i16> %y0 to <4 x i32>
  %sy1 = sext <4 x i16> %y1 to <4 x i32>
  %m0 = mul <4 x i32> %sx0, %sy0
  %m1 = mul <4 x i32> %sx1, %sy1
  %r = add <4 x i32> %m0, %m1
  ret <4 x i32> %r
}

define <4 x i32> @output_size_mismatch(<16 x i16> %x, <16 x i16> %y) {
; CHECK-LABEL: define range(i32 -2147418112, -2147483647) <4 x i32> @output_size_mismatch(
; CHECK-SAME: <16 x i16> [[X:%.*]], <16 x i16> [[Y:%.*]]) local_unnamed_addr #[[ATTR1]] {
; CHECK-NEXT:    [[X0:%.*]] = shufflevector <16 x i16> [[X]], <16 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT:    [[X1:%.*]] = shufflevector <16 x i16> [[X]], <16 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; CHECK-NEXT:    [[Y0:%.*]] = shufflevector <16 x i16> [[Y]], <16 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT:    [[Y1:%.*]] = shufflevector <16 x i16> [[Y]], <16 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; CHECK-NEXT:    [[SX0:%.*]] = sext <4 x i16> [[X0]] to <4 x i32>
; CHECK-NEXT:    [[SX1:%.*]] = sext <4 x i16> [[X1]] to <4 x i32>
; CHECK-NEXT:    [[SY0:%.*]] = sext <4 x i16> [[Y0]] to <4 x i32>
; CHECK-NEXT:    [[SY1:%.*]] = sext <4 x i16> [[Y1]] to <4 x i32>
; CHECK-NEXT:    [[M0:%.*]] = mul nsw <4 x i32> [[SY0]], [[SX0]]
; CHECK-NEXT:    [[M1:%.*]] = mul nsw <4 x i32> [[SY1]], [[SX1]]
; CHECK-NEXT:    [[R:%.*]] = add <4 x i32> [[M0]], [[M1]]
; CHECK-NEXT:    ret <4 x i32> [[R]]
;
  %x0 = shufflevector <16 x i16> %x, <16 x i16> undef, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
  %x1 = shufflevector <16 x i16> %x, <16 x i16> undef, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
  %y0 = shufflevector <16 x i16> %y, <16 x i16> undef, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
  %y1 = shufflevector <16 x i16> %y, <16 x i16> undef, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
  %sx0 = sext <4 x i16> %x0 to <4 x i32>
  %sx1 = sext <4 x i16> %x1 to <4 x i32>
  %sy0 = sext <4 x i16> %y0 to <4 x i32>
  %sy1 = sext <4 x i16> %y1 to <4 x i32>
  %m0 = mul <4 x i32> %sx0, %sy0
  %m1 = mul <4 x i32> %sx1, %sy1
  %r = add <4 x i32> %m0, %m1
  ret <4 x i32> %r
}

define <4 x i32> @output_size_mismatch_high_subvector(<16 x i16> %x, <16 x i16> %y) {
; CHECK-LABEL: define range(i32 -2147418112, -2147483647) <4 x i32> @output_size_mismatch_high_subvector(
; CHECK-SAME: <16 x i16> [[X:%.*]], <16 x i16> [[Y:%.*]]) local_unnamed_addr #[[ATTR1]] {
; CHECK-NEXT:    [[X0:%.*]] = shufflevector <16 x i16> [[X]], <16 x i16> poison, <4 x i32> <i32 8, i32 10, i32 12, i32 14>
; CHECK-NEXT:    [[X1:%.*]] = shufflevector <16 x i16> [[X]], <16 x i16> poison, <4 x i32> <i32 9, i32 11, i32 13, i32 15>
; CHECK-NEXT:    [[Y0:%.*]] = shufflevector <16 x i16> [[Y]], <16 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; CHECK-NEXT:    [[Y1:%.*]] = shufflevector <16 x i16> [[Y]], <16 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; CHECK-NEXT:    [[SX0:%.*]] = sext <4 x i16> [[X0]] to <4 x i32>
; CHECK-NEXT:    [[SX1:%.*]] = sext <4 x i16> [[X1]] to <4 x i32>
; CHECK-NEXT:    [[SY0:%.*]] = sext <4 x i16> [[Y0]] to <4 x i32>
; CHECK-NEXT:    [[SY1:%.*]] = sext <4 x i16> [[Y1]] to <4 x i32>
; CHECK-NEXT:    [[M0:%.*]] = mul nsw <4 x i32> [[SY0]], [[SX0]]
; CHECK-NEXT:    [[M1:%.*]] = mul nsw <4 x i32> [[SY1]], [[SX1]]
; CHECK-NEXT:    [[R:%.*]] = add <4 x i32> [[M0]], [[M1]]
; CHECK-NEXT:    ret <4 x i32> [[R]]
;
  %x0 = shufflevector <16 x i16> %x, <16 x i16> undef, <4 x i32> <i32 8, i32 10, i32 12, i32 14>
  %x1 = shufflevector <16 x i16> %x, <16 x i16> undef, <4 x i32> <i32 9, i32 11, i32 13, i32 15>
  %y0 = shufflevector <16 x i16> %y, <16 x i16> undef, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
  %y1 = shufflevector <16 x i16> %y, <16 x i16> undef, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
  %sx0 = sext <4 x i16> %x0 to <4 x i32>
  %sx1 = sext <4 x i16> %x1 to <4 x i32>
  %sy0 = sext <4 x i16> %y0 to <4 x i32>
  %sy1 = sext <4 x i16> %y1 to <4 x i32>
  %m0 = mul <4 x i32> %sx0, %sy0
  %m1 = mul <4 x i32> %sx1, %sy1
  %r = add <4 x i32> %m0, %m1
  ret <4 x i32> %r
}

define i32 @add_used_by_loop_phi(ptr %a, ptr %b, i64 %offset_a, i64 %offset_b, i64 %k) {
; CHECK-LABEL: define i32 @add_used_by_loop_phi(
; CHECK-SAME: ptr readonly captures(none) [[A:%.*]], ptr readonly captures(none) [[B:%.*]], i64 [[OFFSET_A:%.*]], i64 [[OFFSET_B:%.*]], i64 [[K:%.*]]) local_unnamed_addr #[[ATTR0]] {
; CHECK-NEXT:  [[ENTRY:.*]]:
; CHECK-NEXT:    [[SCEVGEP_A:%.*]] = getelementptr i8, ptr [[A]], i64 [[OFFSET_A]]
; CHECK-NEXT:    [[SCEVGEP_B:%.*]] = getelementptr i8, ptr [[B]], i64 [[OFFSET_B]]
; CHECK-NEXT:    br label %[[LOOP:.*]]
; CHECK:       [[LOOP]]:
; CHECK-NEXT:    [[T0:%.*]] = phi <16 x i32> [ [[TMP3:%.*]], %[[LOOP]] ], [ zeroinitializer, %[[ENTRY]] ]
; CHECK-NEXT:    [[IVLOOP:%.*]] = phi i64 [ [[NEXTIVLOOP:%.*]], %[[LOOP]] ], [ 0, %[[ENTRY]] ]
; CHECK-NEXT:    [[SCEVGEP_A1:%.*]] = getelementptr i8, ptr [[SCEVGEP_A]], i64 [[IVLOOP]]
; CHECK-NEXT:    [[GEPLOAD_A:%.*]] = load <16 x i8>, ptr [[SCEVGEP_A1]], align 1
; CHECK-NEXT:    [[SCEVGEP_B1:%.*]] = getelementptr i8, ptr [[SCEVGEP_B]], i64 [[IVLOOP]]
; CHECK-NEXT:    [[GEPLOAD_B:%.*]] = load <16 x i8>, ptr [[SCEVGEP_B1]], align 1
; CHECK-NEXT:    [[TMP0:%.*]] = sext <16 x i8> [[GEPLOAD_A]] to <16 x i32>
; CHECK-NEXT:    [[TMP1:%.*]] = sext <16 x i8> [[GEPLOAD_B]] to <16 x i32>
; CHECK-NEXT:    [[TMP2:%.*]] = mul nsw <16 x i32> [[TMP1]], [[TMP0]]
; CHECK-NEXT:    [[TMP3]] = add <16 x i32> [[TMP2]], [[T0]]
; CHECK-NEXT:    [[NEXTIVLOOP]] = add nuw nsw i64 [[IVLOOP]], 16
; CHECK-NEXT:    [[CONDLOOP:%.*]] = icmp ult i64 [[NEXTIVLOOP]], [[K]]
; CHECK-NEXT:    br i1 [[CONDLOOP]], label %[[LOOP]], label %[[AFTERLOOP:.*]]
; CHECK:       [[AFTERLOOP]]:
; CHECK-NEXT:    [[SUM:%.*]] = tail call i32 @llvm.vector.reduce.add.v16i32(<16 x i32> [[TMP3]])
; CHECK-NEXT:    ret i32 [[SUM]]
;
entry:
  %scevgep_a = getelementptr i8, ptr %a, i64 %offset_a
  %scevgep_b = getelementptr i8, ptr %b, i64 %offset_b
  br label %loop

loop:
  %t0 = phi <16 x i32> [ %i3, %loop ], [ zeroinitializer, %entry ]
  %ivloop = phi i64 [ %nextivloop, %loop ], [ 0, %entry ]
  %scevgep_a1 = getelementptr i8, ptr %scevgep_a, i64 %ivloop
  %scevgep_a2 = bitcast ptr %scevgep_a1 to ptr
  %gepload_a = load <16 x i8>, ptr %scevgep_a2, align 1
  %scevgep_b1 = getelementptr i8, ptr %scevgep_b, i64 %ivloop
  %scevgep_b2 = bitcast ptr %scevgep_b1 to ptr
  %gepload_b = load <16 x i8>, ptr %scevgep_b2, align 1
  %i0 = sext <16 x i8> %gepload_a to <16 x i32>
  %i1 = sext <16 x i8> %gepload_b to <16 x i32>
  %i2 = mul nsw <16 x i32> %i0, %i1
  %i3 = add <16 x i32> %i2, %t0
  %nextivloop = add nuw nsw i64 %ivloop, 16
  %condloop = icmp ult i64 %nextivloop, %k
  br i1 %condloop, label %loop, label %afterloop

afterloop:
  %.lcssa = phi <16 x i32> [ %i3, %loop ]
  %rdx.shuf = shufflevector <16 x i32> %.lcssa, <16 x i32> poison, <16 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx = add <16 x i32> %.lcssa, %rdx.shuf
  %rdx.shuf90 = shufflevector <16 x i32> %bin.rdx, <16 x i32> poison, <16 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx91 = add <16 x i32> %bin.rdx, %rdx.shuf90
  %rdx.shuf92 = shufflevector <16 x i32> %bin.rdx91, <16 x i32> poison, <16 x i32> <i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx93 = add <16 x i32> %bin.rdx91, %rdx.shuf92
  %rdx.shuf94 = shufflevector <16 x i32> %bin.rdx93, <16 x i32> poison, <16 x i32> <i32 1, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef>
  %bin.rdx95 = add <16 x i32> %bin.rdx93, %rdx.shuf94
  %sum = extractelement <16 x i32> %bin.rdx95, i32 0
  ret i32 %sum
}