; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt < %s -S -passes=loop-vectorize -mtriple=x86_64-- -mattr=+sse2   | FileCheck %s --check-prefix=SSE2
; RUN: opt < %s -S -passes=loop-vectorize -mtriple=x86_64-- -mattr=+sse4.1 | FileCheck %s --check-prefix=SSE41
; RUN: opt < %s -S -passes=loop-vectorize -mtriple=x86_64-- -mattr=+avx    | FileCheck %s --check-prefix=AVX1
; RUN: opt < %s -S -passes=loop-vectorize -mtriple=x86_64-- -mattr=+avx2   | FileCheck %s --check-prefix=AVX2
; RUN: opt < %s -S -passes=loop-vectorize -mtriple=x86_64-- -mcpu=slm      | FileCheck %s --check-prefix=SSE41

define void @test_muladd(ptr noalias nocapture %d1, ptr noalias nocapture readonly %s1, ptr noalias nocapture readonly %s2, i32 %n) {
; SSE2-LABEL: @test_muladd(
; SSE2-NEXT:  entry:
; SSE2-NEXT:    [[CMP30:%.*]] = icmp sgt i32 [[N:%.*]], 0
; SSE2-NEXT:    br i1 [[CMP30]], label [[FOR_BODY_PREHEADER:%.*]], label [[FOR_END:%.*]]
; SSE2:       for.body.preheader:
; SSE2-NEXT:    [[WIDE_TRIP_COUNT:%.*]] = zext i32 [[N]] to i64
; SSE2-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[WIDE_TRIP_COUNT]], 4
; SSE2-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; SSE2:       vector.ph:
; SSE2-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[WIDE_TRIP_COUNT]], 4
; SSE2-NEXT:    [[N_VEC:%.*]] = sub i64 [[WIDE_TRIP_COUNT]], [[N_MOD_VF]]
; SSE2-NEXT:    br label [[VECTOR_BODY:%.*]]
; SSE2:       vector.body:
; SSE2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; SSE2-NEXT:    [[TMP1:%.*]] = shl nuw nsw i64 [[INDEX]], 1
; SSE2-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i16, ptr [[S1:%.*]], i64 [[TMP1]]
; SSE2-NEXT:    [[WIDE_VEC:%.*]] = load <8 x i16>, ptr [[TMP2]], align 2
; SSE2-NEXT:    [[STRIDED_VEC:%.*]] = shufflevector <8 x i16> [[WIDE_VEC]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; SSE2-NEXT:    [[STRIDED_VEC1:%.*]] = shufflevector <8 x i16> [[WIDE_VEC]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; SSE2-NEXT:    [[TMP4:%.*]] = sext <4 x i16> [[STRIDED_VEC]] to <4 x i32>
; SSE2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i16, ptr [[S2:%.*]], i64 [[TMP1]]
; SSE2-NEXT:    [[WIDE_VEC2:%.*]] = load <8 x i16>, ptr [[TMP5]], align 2
; SSE2-NEXT:    [[STRIDED_VEC3:%.*]] = shufflevector <8 x i16> [[WIDE_VEC2]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; SSE2-NEXT:    [[STRIDED_VEC4:%.*]] = shufflevector <8 x i16> [[WIDE_VEC2]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; SSE2-NEXT:    [[TMP7:%.*]] = sext <4 x i16> [[STRIDED_VEC3]] to <4 x i32>
; SSE2-NEXT:    [[TMP8:%.*]] = mul nsw <4 x i32> [[TMP7]], [[TMP4]]
; SSE2-NEXT:    [[TMP9:%.*]] = sext <4 x i16> [[STRIDED_VEC1]] to <4 x i32>
; SSE2-NEXT:    [[TMP10:%.*]] = sext <4 x i16> [[STRIDED_VEC4]] to <4 x i32>
; SSE2-NEXT:    [[TMP11:%.*]] = mul nsw <4 x i32> [[TMP10]], [[TMP9]]
; SSE2-NEXT:    [[TMP12:%.*]] = add nsw <4 x i32> [[TMP11]], [[TMP8]]
; SSE2-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[D1:%.*]], i64 [[INDEX]]
; SSE2-NEXT:    store <4 x i32> [[TMP12]], ptr [[TMP13]], align 4
; SSE2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
; SSE2-NEXT:    [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; SSE2-NEXT:    br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; SSE2:       middle.block:
; SSE2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[WIDE_TRIP_COUNT]], [[N_VEC]]
; SSE2-NEXT:    br i1 [[CMP_N]], label [[FOR_END_LOOPEXIT:%.*]], label [[SCALAR_PH]]
; SSE2:       scalar.ph:
; SSE2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
; SSE2-NEXT:    br label [[FOR_BODY:%.*]]
; SSE2:       for.body:
; SSE2-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
; SSE2-NEXT:    [[TMP16:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 1
; SSE2-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i16, ptr [[S1]], i64 [[TMP16]]
; SSE2-NEXT:    [[TMP17:%.*]] = load i16, ptr [[ARRAYIDX]], align 2
; SSE2-NEXT:    [[CONV:%.*]] = sext i16 [[TMP17]] to i32
; SSE2-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds i16, ptr [[S2]], i64 [[TMP16]]
; SSE2-NEXT:    [[TMP18:%.*]] = load i16, ptr [[ARRAYIDX4]], align 2
; SSE2-NEXT:    [[CONV5:%.*]] = sext i16 [[TMP18]] to i32
; SSE2-NEXT:    [[MUL6:%.*]] = mul nsw i32 [[CONV5]], [[CONV]]
; SSE2-NEXT:    [[TMP19:%.*]] = or disjoint i64 [[TMP16]], 1
; SSE2-NEXT:    [[ARRAYIDX10:%.*]] = getelementptr inbounds i16, ptr [[S1]], i64 [[TMP19]]
; SSE2-NEXT:    [[TMP20:%.*]] = load i16, ptr [[ARRAYIDX10]], align 2
; SSE2-NEXT:    [[CONV11:%.*]] = sext i16 [[TMP20]] to i32
; SSE2-NEXT:    [[ARRAYIDX15:%.*]] = getelementptr inbounds i16, ptr [[S2]], i64 [[TMP19]]
; SSE2-NEXT:    [[TMP21:%.*]] = load i16, ptr [[ARRAYIDX15]], align 2
; SSE2-NEXT:    [[CONV16:%.*]] = sext i16 [[TMP21]] to i32
; SSE2-NEXT:    [[MUL17:%.*]] = mul nsw i32 [[CONV16]], [[CONV11]]
; SSE2-NEXT:    [[ADD18:%.*]] = add nsw i32 [[MUL17]], [[MUL6]]
; SSE2-NEXT:    [[ARRAYIDX20:%.*]] = getelementptr inbounds i32, ptr [[D1]], i64 [[INDVARS_IV]]
; SSE2-NEXT:    store i32 [[ADD18]], ptr [[ARRAYIDX20]], align 4
; SSE2-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; SSE2-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[WIDE_TRIP_COUNT]]
; SSE2-NEXT:    br i1 [[EXITCOND_NOT]], label [[FOR_END_LOOPEXIT]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; SSE2:       for.end.loopexit:
; SSE2-NEXT:    br label [[FOR_END]]
; SSE2:       for.end:
; SSE2-NEXT:    ret void
;
; SSE41-LABEL: @test_muladd(
; SSE41-NEXT:  entry:
; SSE41-NEXT:    [[CMP30:%.*]] = icmp sgt i32 [[N:%.*]], 0
; SSE41-NEXT:    br i1 [[CMP30]], label [[FOR_BODY_PREHEADER:%.*]], label [[FOR_END:%.*]]
; SSE41:       for.body.preheader:
; SSE41-NEXT:    [[WIDE_TRIP_COUNT:%.*]] = zext i32 [[N]] to i64
; SSE41-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[WIDE_TRIP_COUNT]], 8
; SSE41-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; SSE41:       vector.ph:
; SSE41-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[WIDE_TRIP_COUNT]], 8
; SSE41-NEXT:    [[N_VEC:%.*]] = sub i64 [[WIDE_TRIP_COUNT]], [[N_MOD_VF]]
; SSE41-NEXT:    br label [[VECTOR_BODY:%.*]]
; SSE41:       vector.body:
; SSE41-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; SSE41-NEXT:    [[TMP1:%.*]] = add i64 [[INDEX]], 4
; SSE41-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[INDEX]], 1
; SSE41-NEXT:    [[TMP3:%.*]] = shl nuw nsw i64 [[TMP1]], 1
; SSE41-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i16, ptr [[S1:%.*]], i64 [[TMP2]]
; SSE41-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i16, ptr [[S1]], i64 [[TMP3]]
; SSE41-NEXT:    [[WIDE_VEC:%.*]] = load <8 x i16>, ptr [[TMP4]], align 2
; SSE41-NEXT:    [[STRIDED_VEC:%.*]] = shufflevector <8 x i16> [[WIDE_VEC]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; SSE41-NEXT:    [[STRIDED_VEC3:%.*]] = shufflevector <8 x i16> [[WIDE_VEC]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; SSE41-NEXT:    [[WIDE_VEC1:%.*]] = load <8 x i16>, ptr [[TMP5]], align 2
; SSE41-NEXT:    [[STRIDED_VEC2:%.*]] = shufflevector <8 x i16> [[WIDE_VEC1]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; SSE41-NEXT:    [[STRIDED_VEC4:%.*]] = shufflevector <8 x i16> [[WIDE_VEC1]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; SSE41-NEXT:    [[TMP8:%.*]] = sext <4 x i16> [[STRIDED_VEC]] to <4 x i32>
; SSE41-NEXT:    [[TMP9:%.*]] = sext <4 x i16> [[STRIDED_VEC2]] to <4 x i32>
; SSE41-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i16, ptr [[S2:%.*]], i64 [[TMP2]]
; SSE41-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i16, ptr [[S2]], i64 [[TMP3]]
; SSE41-NEXT:    [[WIDE_VEC5:%.*]] = load <8 x i16>, ptr [[TMP10]], align 2
; SSE41-NEXT:    [[STRIDED_VEC7:%.*]] = shufflevector <8 x i16> [[WIDE_VEC5]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; SSE41-NEXT:    [[STRIDED_VEC9:%.*]] = shufflevector <8 x i16> [[WIDE_VEC5]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; SSE41-NEXT:    [[WIDE_VEC6:%.*]] = load <8 x i16>, ptr [[TMP11]], align 2
; SSE41-NEXT:    [[STRIDED_VEC8:%.*]] = shufflevector <8 x i16> [[WIDE_VEC6]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; SSE41-NEXT:    [[STRIDED_VEC10:%.*]] = shufflevector <8 x i16> [[WIDE_VEC6]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; SSE41-NEXT:    [[TMP14:%.*]] = sext <4 x i16> [[STRIDED_VEC7]] to <4 x i32>
; SSE41-NEXT:    [[TMP15:%.*]] = sext <4 x i16> [[STRIDED_VEC8]] to <4 x i32>
; SSE41-NEXT:    [[TMP16:%.*]] = mul nsw <4 x i32> [[TMP14]], [[TMP8]]
; SSE41-NEXT:    [[TMP17:%.*]] = mul nsw <4 x i32> [[TMP15]], [[TMP9]]
; SSE41-NEXT:    [[TMP18:%.*]] = sext <4 x i16> [[STRIDED_VEC3]] to <4 x i32>
; SSE41-NEXT:    [[TMP19:%.*]] = sext <4 x i16> [[STRIDED_VEC4]] to <4 x i32>
; SSE41-NEXT:    [[TMP20:%.*]] = sext <4 x i16> [[STRIDED_VEC9]] to <4 x i32>
; SSE41-NEXT:    [[TMP21:%.*]] = sext <4 x i16> [[STRIDED_VEC10]] to <4 x i32>
; SSE41-NEXT:    [[TMP22:%.*]] = mul nsw <4 x i32> [[TMP20]], [[TMP18]]
; SSE41-NEXT:    [[TMP23:%.*]] = mul nsw <4 x i32> [[TMP21]], [[TMP19]]
; SSE41-NEXT:    [[TMP24:%.*]] = add nsw <4 x i32> [[TMP22]], [[TMP16]]
; SSE41-NEXT:    [[TMP25:%.*]] = add nsw <4 x i32> [[TMP23]], [[TMP17]]
; SSE41-NEXT:    [[TMP26:%.*]] = getelementptr inbounds i32, ptr [[D1:%.*]], i64 [[INDEX]]
; SSE41-NEXT:    [[TMP29:%.*]] = getelementptr inbounds i32, ptr [[TMP26]], i32 4
; SSE41-NEXT:    store <4 x i32> [[TMP24]], ptr [[TMP26]], align 4
; SSE41-NEXT:    store <4 x i32> [[TMP25]], ptr [[TMP29]], align 4
; SSE41-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
; SSE41-NEXT:    [[TMP30:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; SSE41-NEXT:    br i1 [[TMP30]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; SSE41:       middle.block:
; SSE41-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[WIDE_TRIP_COUNT]], [[N_VEC]]
; SSE41-NEXT:    br i1 [[CMP_N]], label [[FOR_END_LOOPEXIT:%.*]], label [[SCALAR_PH]]
; SSE41:       scalar.ph:
; SSE41-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
; SSE41-NEXT:    br label [[FOR_BODY:%.*]]
; SSE41:       for.body:
; SSE41-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
; SSE41-NEXT:    [[TMP31:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 1
; SSE41-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i16, ptr [[S1]], i64 [[TMP31]]
; SSE41-NEXT:    [[TMP32:%.*]] = load i16, ptr [[ARRAYIDX]], align 2
; SSE41-NEXT:    [[CONV:%.*]] = sext i16 [[TMP32]] to i32
; SSE41-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds i16, ptr [[S2]], i64 [[TMP31]]
; SSE41-NEXT:    [[TMP33:%.*]] = load i16, ptr [[ARRAYIDX4]], align 2
; SSE41-NEXT:    [[CONV5:%.*]] = sext i16 [[TMP33]] to i32
; SSE41-NEXT:    [[MUL6:%.*]] = mul nsw i32 [[CONV5]], [[CONV]]
; SSE41-NEXT:    [[TMP34:%.*]] = or disjoint i64 [[TMP31]], 1
; SSE41-NEXT:    [[ARRAYIDX10:%.*]] = getelementptr inbounds i16, ptr [[S1]], i64 [[TMP34]]
; SSE41-NEXT:    [[TMP35:%.*]] = load i16, ptr [[ARRAYIDX10]], align 2
; SSE41-NEXT:    [[CONV11:%.*]] = sext i16 [[TMP35]] to i32
; SSE41-NEXT:    [[ARRAYIDX15:%.*]] = getelementptr inbounds i16, ptr [[S2]], i64 [[TMP34]]
; SSE41-NEXT:    [[TMP36:%.*]] = load i16, ptr [[ARRAYIDX15]], align 2
; SSE41-NEXT:    [[CONV16:%.*]] = sext i16 [[TMP36]] to i32
; SSE41-NEXT:    [[MUL17:%.*]] = mul nsw i32 [[CONV16]], [[CONV11]]
; SSE41-NEXT:    [[ADD18:%.*]] = add nsw i32 [[MUL17]], [[MUL6]]
; SSE41-NEXT:    [[ARRAYIDX20:%.*]] = getelementptr inbounds i32, ptr [[D1]], i64 [[INDVARS_IV]]
; SSE41-NEXT:    store i32 [[ADD18]], ptr [[ARRAYIDX20]], align 4
; SSE41-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; SSE41-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[WIDE_TRIP_COUNT]]
; SSE41-NEXT:    br i1 [[EXITCOND_NOT]], label [[FOR_END_LOOPEXIT]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; SSE41:       for.end.loopexit:
; SSE41-NEXT:    br label [[FOR_END]]
; SSE41:       for.end:
; SSE41-NEXT:    ret void
;
; AVX1-LABEL: @test_muladd(
; AVX1-NEXT:  entry:
; AVX1-NEXT:    [[CMP30:%.*]] = icmp sgt i32 [[N:%.*]], 0
; AVX1-NEXT:    br i1 [[CMP30]], label [[FOR_BODY_PREHEADER:%.*]], label [[FOR_END:%.*]]
; AVX1:       iter.check:
; AVX1-NEXT:    [[WIDE_TRIP_COUNT:%.*]] = zext i32 [[N]] to i64
; AVX1-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[WIDE_TRIP_COUNT]], 4
; AVX1-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; AVX1:       vector.main.loop.iter.check:
; AVX1-NEXT:    [[MIN_ITERS_CHECK1:%.*]] = icmp ult i64 [[WIDE_TRIP_COUNT]], 16
; AVX1-NEXT:    br i1 [[MIN_ITERS_CHECK1]], label [[VEC_EPILOG_PH:%.*]], label [[VECTOR_PH1:%.*]]
; AVX1:       vector.ph:
; AVX1-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[WIDE_TRIP_COUNT]], 16
; AVX1-NEXT:    [[N_VEC:%.*]] = sub i64 [[WIDE_TRIP_COUNT]], [[N_MOD_VF]]
; AVX1-NEXT:    br label [[VECTOR_BODY:%.*]]
; AVX1:       vector.body:
; AVX1-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH1]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; AVX1-NEXT:    [[TMP1:%.*]] = add i64 [[INDEX]], 4
; AVX1-NEXT:    [[TMP3:%.*]] = add i64 [[INDEX]], 8
; AVX1-NEXT:    [[TMP8:%.*]] = add i64 [[INDEX]], 12
; AVX1-NEXT:    [[TMP7:%.*]] = shl nuw nsw i64 [[INDEX]], 1
; AVX1-NEXT:    [[TMP2:%.*]] = shl nuw nsw i64 [[TMP1]], 1
; AVX1-NEXT:    [[TMP5:%.*]] = shl nuw nsw i64 [[TMP3]], 1
; AVX1-NEXT:    [[TMP6:%.*]] = shl nuw nsw i64 [[TMP8]], 1
; AVX1-NEXT:    [[TMP11:%.*]] = getelementptr inbounds i16, ptr [[S1:%.*]], i64 [[TMP7]]
; AVX1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds i16, ptr [[S1]], i64 [[TMP2]]
; AVX1-NEXT:    [[TMP9:%.*]] = getelementptr inbounds i16, ptr [[S1]], i64 [[TMP5]]
; AVX1-NEXT:    [[TMP10:%.*]] = getelementptr inbounds i16, ptr [[S1]], i64 [[TMP6]]
; AVX1-NEXT:    [[WIDE_VEC2:%.*]] = load <8 x i16>, ptr [[TMP11]], align 2
; AVX1-NEXT:    [[STRIDED_VEC5:%.*]] = shufflevector <8 x i16> [[WIDE_VEC2]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; AVX1-NEXT:    [[STRIDED_VEC9:%.*]] = shufflevector <8 x i16> [[WIDE_VEC2]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; AVX1-NEXT:    [[WIDE_VEC3:%.*]] = load <8 x i16>, ptr [[TMP4]], align 2
; AVX1-NEXT:    [[STRIDED_VEC6:%.*]] = shufflevector <8 x i16> [[WIDE_VEC3]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; AVX1-NEXT:    [[STRIDED_VEC10:%.*]] = shufflevector <8 x i16> [[WIDE_VEC3]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; AVX1-NEXT:    [[WIDE_VEC6:%.*]] = load <8 x i16>, ptr [[TMP9]], align 2
; AVX1-NEXT:    [[STRIDED_VEC7:%.*]] = shufflevector <8 x i16> [[WIDE_VEC6]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; AVX1-NEXT:    [[STRIDED_VEC8:%.*]] = shufflevector <8 x i16> [[WIDE_VEC6]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; AVX1-NEXT:    [[WIDE_VEC9:%.*]] = load <8 x i16>, ptr [[TMP10]], align 2
; AVX1-NEXT:    [[STRIDED_VEC12:%.*]] = shufflevector <8 x i16> [[WIDE_VEC9]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; AVX1-NEXT:    [[STRIDED_VEC11:%.*]] = shufflevector <8 x i16> [[WIDE_VEC9]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; AVX1-NEXT:    [[TMP36:%.*]] = sext <4 x i16> [[STRIDED_VEC5]] to <4 x i32>
; AVX1-NEXT:    [[TMP37:%.*]] = sext <4 x i16> [[STRIDED_VEC6]] to <4 x i32>
; AVX1-NEXT:    [[TMP38:%.*]] = sext <4 x i16> [[STRIDED_VEC7]] to <4 x i32>
; AVX1-NEXT:    [[TMP39:%.*]] = sext <4 x i16> [[STRIDED_VEC12]] to <4 x i32>
; AVX1-NEXT:    [[TMP22:%.*]] = getelementptr inbounds i16, ptr [[S2:%.*]], i64 [[TMP7]]
; AVX1-NEXT:    [[TMP23:%.*]] = getelementptr inbounds i16, ptr [[S2]], i64 [[TMP2]]
; AVX1-NEXT:    [[TMP17:%.*]] = getelementptr inbounds i16, ptr [[S2]], i64 [[TMP5]]
; AVX1-NEXT:    [[TMP18:%.*]] = getelementptr inbounds i16, ptr [[S2]], i64 [[TMP6]]
; AVX1-NEXT:    [[WIDE_VEC13:%.*]] = load <8 x i16>, ptr [[TMP22]], align 2
; AVX1-NEXT:    [[STRIDED_VEC17:%.*]] = shufflevector <8 x i16> [[WIDE_VEC13]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; AVX1-NEXT:    [[STRIDED_VEC21:%.*]] = shufflevector <8 x i16> [[WIDE_VEC13]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; AVX1-NEXT:    [[WIDE_VEC14:%.*]] = load <8 x i16>, ptr [[TMP23]], align 2
; AVX1-NEXT:    [[STRIDED_VEC18:%.*]] = shufflevector <8 x i16> [[WIDE_VEC14]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; AVX1-NEXT:    [[STRIDED_VEC22:%.*]] = shufflevector <8 x i16> [[WIDE_VEC14]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; AVX1-NEXT:    [[WIDE_VEC18:%.*]] = load <8 x i16>, ptr [[TMP17]], align 2
; AVX1-NEXT:    [[STRIDED_VEC19:%.*]] = shufflevector <8 x i16> [[WIDE_VEC18]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; AVX1-NEXT:    [[STRIDED_VEC20:%.*]] = shufflevector <8 x i16> [[WIDE_VEC18]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; AVX1-NEXT:    [[WIDE_VEC21:%.*]] = load <8 x i16>, ptr [[TMP18]], align 2
; AVX1-NEXT:    [[STRIDED_VEC24:%.*]] = shufflevector <8 x i16> [[WIDE_VEC21]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; AVX1-NEXT:    [[STRIDED_VEC23:%.*]] = shufflevector <8 x i16> [[WIDE_VEC21]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; AVX1-NEXT:    [[TMP40:%.*]] = sext <4 x i16> [[STRIDED_VEC17]] to <4 x i32>
; AVX1-NEXT:    [[TMP41:%.*]] = sext <4 x i16> [[STRIDED_VEC18]] to <4 x i32>
; AVX1-NEXT:    [[TMP42:%.*]] = sext <4 x i16> [[STRIDED_VEC19]] to <4 x i32>
; AVX1-NEXT:    [[TMP43:%.*]] = sext <4 x i16> [[STRIDED_VEC24]] to <4 x i32>
; AVX1-NEXT:    [[TMP44:%.*]] = mul nsw <4 x i32> [[TMP40]], [[TMP36]]
; AVX1-NEXT:    [[TMP45:%.*]] = mul nsw <4 x i32> [[TMP41]], [[TMP37]]
; AVX1-NEXT:    [[TMP46:%.*]] = mul nsw <4 x i32> [[TMP42]], [[TMP38]]
; AVX1-NEXT:    [[TMP47:%.*]] = mul nsw <4 x i32> [[TMP43]], [[TMP39]]
; AVX1-NEXT:    [[TMP27:%.*]] = sext <4 x i16> [[STRIDED_VEC9]] to <4 x i32>
; AVX1-NEXT:    [[TMP28:%.*]] = sext <4 x i16> [[STRIDED_VEC10]] to <4 x i32>
; AVX1-NEXT:    [[TMP29:%.*]] = sext <4 x i16> [[STRIDED_VEC8]] to <4 x i32>
; AVX1-NEXT:    [[TMP30:%.*]] = sext <4 x i16> [[STRIDED_VEC11]] to <4 x i32>
; AVX1-NEXT:    [[TMP31:%.*]] = sext <4 x i16> [[STRIDED_VEC21]] to <4 x i32>
; AVX1-NEXT:    [[TMP32:%.*]] = sext <4 x i16> [[STRIDED_VEC22]] to <4 x i32>
; AVX1-NEXT:    [[TMP33:%.*]] = sext <4 x i16> [[STRIDED_VEC20]] to <4 x i32>
; AVX1-NEXT:    [[TMP34:%.*]] = sext <4 x i16> [[STRIDED_VEC23]] to <4 x i32>
; AVX1-NEXT:    [[TMP35:%.*]] = mul nsw <4 x i32> [[TMP31]], [[TMP27]]
; AVX1-NEXT:    [[TMP60:%.*]] = mul nsw <4 x i32> [[TMP32]], [[TMP28]]
; AVX1-NEXT:    [[TMP67:%.*]] = mul nsw <4 x i32> [[TMP33]], [[TMP29]]
; AVX1-NEXT:    [[TMP68:%.*]] = mul nsw <4 x i32> [[TMP34]], [[TMP30]]
; AVX1-NEXT:    [[TMP19:%.*]] = add nsw <4 x i32> [[TMP35]], [[TMP44]]
; AVX1-NEXT:    [[TMP20:%.*]] = add nsw <4 x i32> [[TMP60]], [[TMP45]]
; AVX1-NEXT:    [[TMP69:%.*]] = add nsw <4 x i32> [[TMP67]], [[TMP46]]
; AVX1-NEXT:    [[TMP70:%.*]] = add nsw <4 x i32> [[TMP68]], [[TMP47]]
; AVX1-NEXT:    [[TMP21:%.*]] = getelementptr inbounds i32, ptr [[D1:%.*]], i64 [[INDEX]]
; AVX1-NEXT:    [[TMP26:%.*]] = getelementptr inbounds i32, ptr [[TMP21]], i32 4
; AVX1-NEXT:    [[TMP71:%.*]] = getelementptr inbounds i32, ptr [[TMP21]], i32 8
; AVX1-NEXT:    [[TMP72:%.*]] = getelementptr inbounds i32, ptr [[TMP21]], i32 12
; AVX1-NEXT:    store <4 x i32> [[TMP19]], ptr [[TMP21]], align 4
; AVX1-NEXT:    store <4 x i32> [[TMP20]], ptr [[TMP26]], align 4
; AVX1-NEXT:    store <4 x i32> [[TMP69]], ptr [[TMP71]], align 4
; AVX1-NEXT:    store <4 x i32> [[TMP70]], ptr [[TMP72]], align 4
; AVX1-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
; AVX1-NEXT:    [[TMP24:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; AVX1-NEXT:    br i1 [[TMP24]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; AVX1:       middle.block:
; AVX1-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[WIDE_TRIP_COUNT]], [[N_VEC]]
; AVX1-NEXT:    br i1 [[CMP_N]], label [[FOR_END_LOOPEXIT:%.*]], label [[VEC_EPILOG_ITER_CHECK:%.*]]
; AVX1:       vec.epilog.iter.check:
; AVX1-NEXT:    [[MIN_EPILOG_ITERS_CHECK:%.*]] = icmp ult i64 [[N_MOD_VF]], 4
; AVX1-NEXT:    br i1 [[MIN_EPILOG_ITERS_CHECK]], label [[SCALAR_PH]], label [[VEC_EPILOG_PH]], !prof [[PROF3:![0-9]+]]
; AVX1:       vec.epilog.ph:
; AVX1-NEXT:    [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_PH]] ]
; AVX1-NEXT:    [[N_MOD_VF24:%.*]] = urem i64 [[WIDE_TRIP_COUNT]], 4
; AVX1-NEXT:    [[N_VEC25:%.*]] = sub i64 [[WIDE_TRIP_COUNT]], [[N_MOD_VF24]]
; AVX1-NEXT:    br label [[FOR_BODY1:%.*]]
; AVX1:       vec.epilog.vector.body:
; AVX1-NEXT:    [[INDEX26:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT33:%.*]], [[FOR_BODY1]] ]
; AVX1-NEXT:    [[TMP48:%.*]] = shl nuw nsw i64 [[INDEX26]], 1
; AVX1-NEXT:    [[TMP49:%.*]] = getelementptr inbounds i16, ptr [[S1]], i64 [[TMP48]]
; AVX1-NEXT:    [[WIDE_VEC27:%.*]] = load <8 x i16>, ptr [[TMP49]], align 2
; AVX1-NEXT:    [[STRIDED_VEC28:%.*]] = shufflevector <8 x i16> [[WIDE_VEC27]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; AVX1-NEXT:    [[STRIDED_VEC29:%.*]] = shufflevector <8 x i16> [[WIDE_VEC27]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; AVX1-NEXT:    [[TMP50:%.*]] = sext <4 x i16> [[STRIDED_VEC28]] to <4 x i32>
; AVX1-NEXT:    [[TMP51:%.*]] = getelementptr inbounds i16, ptr [[S2]], i64 [[TMP48]]
; AVX1-NEXT:    [[WIDE_VEC30:%.*]] = load <8 x i16>, ptr [[TMP51]], align 2
; AVX1-NEXT:    [[STRIDED_VEC31:%.*]] = shufflevector <8 x i16> [[WIDE_VEC30]], <8 x i16> poison, <4 x i32> <i32 0, i32 2, i32 4, i32 6>
; AVX1-NEXT:    [[STRIDED_VEC32:%.*]] = shufflevector <8 x i16> [[WIDE_VEC30]], <8 x i16> poison, <4 x i32> <i32 1, i32 3, i32 5, i32 7>
; AVX1-NEXT:    [[TMP52:%.*]] = sext <4 x i16> [[STRIDED_VEC31]] to <4 x i32>
; AVX1-NEXT:    [[TMP53:%.*]] = mul nsw <4 x i32> [[TMP52]], [[TMP50]]
; AVX1-NEXT:    [[TMP54:%.*]] = sext <4 x i16> [[STRIDED_VEC29]] to <4 x i32>
; AVX1-NEXT:    [[TMP55:%.*]] = sext <4 x i16> [[STRIDED_VEC32]] to <4 x i32>
; AVX1-NEXT:    [[TMP56:%.*]] = mul nsw <4 x i32> [[TMP55]], [[TMP54]]
; AVX1-NEXT:    [[TMP57:%.*]] = add nsw <4 x i32> [[TMP56]], [[TMP53]]
; AVX1-NEXT:    [[TMP58:%.*]] = getelementptr inbounds i32, ptr [[D1]], i64 [[INDEX26]]
; AVX1-NEXT:    store <4 x i32> [[TMP57]], ptr [[TMP58]], align 4
; AVX1-NEXT:    [[INDEX_NEXT33]] = add nuw i64 [[INDEX26]], 4
; AVX1-NEXT:    [[TMP59:%.*]] = icmp eq i64 [[INDEX_NEXT33]], [[N_VEC25]]
; AVX1-NEXT:    br i1 [[TMP59]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[FOR_BODY1]], !llvm.loop [[LOOP4:![0-9]+]]
; AVX1:       vec.epilog.middle.block:
; AVX1-NEXT:    [[CMP_N34:%.*]] = icmp eq i64 [[WIDE_TRIP_COUNT]], [[N_VEC25]]
; AVX1-NEXT:    br i1 [[CMP_N34]], label [[FOR_END_LOOPEXIT]], label [[SCALAR_PH]]
; AVX1:       vec.epilog.scalar.ph:
; AVX1-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC25]], [[VEC_EPILOG_MIDDLE_BLOCK]] ], [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
; AVX1-NEXT:    br label [[FOR_BODY:%.*]]
; AVX1:       for.body:
; AVX1-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
; AVX1-NEXT:    [[TMP61:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 1
; AVX1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i16, ptr [[S1]], i64 [[TMP61]]
; AVX1-NEXT:    [[TMP62:%.*]] = load i16, ptr [[ARRAYIDX]], align 2
; AVX1-NEXT:    [[CONV:%.*]] = sext i16 [[TMP62]] to i32
; AVX1-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds i16, ptr [[S2]], i64 [[TMP61]]
; AVX1-NEXT:    [[TMP63:%.*]] = load i16, ptr [[ARRAYIDX4]], align 2
; AVX1-NEXT:    [[CONV5:%.*]] = sext i16 [[TMP63]] to i32
; AVX1-NEXT:    [[MUL6:%.*]] = mul nsw i32 [[CONV5]], [[CONV]]
; AVX1-NEXT:    [[TMP64:%.*]] = or disjoint i64 [[TMP61]], 1
; AVX1-NEXT:    [[ARRAYIDX10:%.*]] = getelementptr inbounds i16, ptr [[S1]], i64 [[TMP64]]
; AVX1-NEXT:    [[TMP65:%.*]] = load i16, ptr [[ARRAYIDX10]], align 2
; AVX1-NEXT:    [[CONV11:%.*]] = sext i16 [[TMP65]] to i32
; AVX1-NEXT:    [[ARRAYIDX15:%.*]] = getelementptr inbounds i16, ptr [[S2]], i64 [[TMP64]]
; AVX1-NEXT:    [[TMP66:%.*]] = load i16, ptr [[ARRAYIDX15]], align 2
; AVX1-NEXT:    [[CONV16:%.*]] = sext i16 [[TMP66]] to i32
; AVX1-NEXT:    [[MUL17:%.*]] = mul nsw i32 [[CONV16]], [[CONV11]]
; AVX1-NEXT:    [[ADD18:%.*]] = add nsw i32 [[MUL17]], [[MUL6]]
; AVX1-NEXT:    [[ARRAYIDX20:%.*]] = getelementptr inbounds i32, ptr [[D1]], i64 [[INDVARS_IV]]
; AVX1-NEXT:    store i32 [[ADD18]], ptr [[ARRAYIDX20]], align 4
; AVX1-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; AVX1-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[WIDE_TRIP_COUNT]]
; AVX1-NEXT:    br i1 [[EXITCOND_NOT]], label [[FOR_END_LOOPEXIT]], label [[FOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]]
; AVX1:       for.end.loopexit:
; AVX1-NEXT:    br label [[FOR_END]]
; AVX1:       for.end:
; AVX1-NEXT:    ret void
;
; AVX2-LABEL: @test_muladd(
; AVX2-NEXT:  entry:
; AVX2-NEXT:    [[CMP30:%.*]] = icmp sgt i32 [[N:%.*]], 0
; AVX2-NEXT:    br i1 [[CMP30]], label [[FOR_BODY_PREHEADER:%.*]], label [[FOR_END:%.*]]
; AVX2:       for.body.preheader:
; AVX2-NEXT:    [[WIDE_TRIP_COUNT:%.*]] = zext i32 [[N]] to i64
; AVX2-NEXT:    [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[WIDE_TRIP_COUNT]], 8
; AVX2-NEXT:    br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; AVX2:       vector.ph:
; AVX2-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[WIDE_TRIP_COUNT]], 8
; AVX2-NEXT:    [[N_VEC:%.*]] = sub i64 [[WIDE_TRIP_COUNT]], [[N_MOD_VF]]
; AVX2-NEXT:    br label [[VECTOR_BODY:%.*]]
; AVX2:       vector.body:
; AVX2-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
; AVX2-NEXT:    [[TMP1:%.*]] = shl nuw nsw i64 [[INDEX]], 1
; AVX2-NEXT:    [[TMP2:%.*]] = getelementptr inbounds i16, ptr [[S1:%.*]], i64 [[TMP1]]
; AVX2-NEXT:    [[WIDE_VEC:%.*]] = load <16 x i16>, ptr [[TMP2]], align 2
; AVX2-NEXT:    [[STRIDED_VEC:%.*]] = shufflevector <16 x i16> [[WIDE_VEC]], <16 x i16> poison, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
; AVX2-NEXT:    [[STRIDED_VEC1:%.*]] = shufflevector <16 x i16> [[WIDE_VEC]], <16 x i16> poison, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
; AVX2-NEXT:    [[TMP4:%.*]] = sext <8 x i16> [[STRIDED_VEC]] to <8 x i32>
; AVX2-NEXT:    [[TMP5:%.*]] = getelementptr inbounds i16, ptr [[S2:%.*]], i64 [[TMP1]]
; AVX2-NEXT:    [[WIDE_VEC2:%.*]] = load <16 x i16>, ptr [[TMP5]], align 2
; AVX2-NEXT:    [[STRIDED_VEC3:%.*]] = shufflevector <16 x i16> [[WIDE_VEC2]], <16 x i16> poison, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 8, i32 10, i32 12, i32 14>
; AVX2-NEXT:    [[STRIDED_VEC4:%.*]] = shufflevector <16 x i16> [[WIDE_VEC2]], <16 x i16> poison, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 9, i32 11, i32 13, i32 15>
; AVX2-NEXT:    [[TMP7:%.*]] = sext <8 x i16> [[STRIDED_VEC3]] to <8 x i32>
; AVX2-NEXT:    [[TMP8:%.*]] = mul nsw <8 x i32> [[TMP7]], [[TMP4]]
; AVX2-NEXT:    [[TMP9:%.*]] = sext <8 x i16> [[STRIDED_VEC1]] to <8 x i32>
; AVX2-NEXT:    [[TMP10:%.*]] = sext <8 x i16> [[STRIDED_VEC4]] to <8 x i32>
; AVX2-NEXT:    [[TMP11:%.*]] = mul nsw <8 x i32> [[TMP10]], [[TMP9]]
; AVX2-NEXT:    [[TMP12:%.*]] = add nsw <8 x i32> [[TMP11]], [[TMP8]]
; AVX2-NEXT:    [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[D1:%.*]], i64 [[INDEX]]
; AVX2-NEXT:    store <8 x i32> [[TMP12]], ptr [[TMP13]], align 4
; AVX2-NEXT:    [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 8
; AVX2-NEXT:    [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
; AVX2-NEXT:    br i1 [[TMP15]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
; AVX2:       middle.block:
; AVX2-NEXT:    [[CMP_N:%.*]] = icmp eq i64 [[WIDE_TRIP_COUNT]], [[N_VEC]]
; AVX2-NEXT:    br i1 [[CMP_N]], label [[FOR_END_LOOPEXIT:%.*]], label [[SCALAR_PH]]
; AVX2:       scalar.ph:
; AVX2-NEXT:    [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[FOR_BODY_PREHEADER]] ]
; AVX2-NEXT:    br label [[FOR_BODY:%.*]]
; AVX2:       for.body:
; AVX2-NEXT:    [[INDVARS_IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INDVARS_IV_NEXT:%.*]], [[FOR_BODY]] ]
; AVX2-NEXT:    [[TMP16:%.*]] = shl nuw nsw i64 [[INDVARS_IV]], 1
; AVX2-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i16, ptr [[S1]], i64 [[TMP16]]
; AVX2-NEXT:    [[TMP17:%.*]] = load i16, ptr [[ARRAYIDX]], align 2
; AVX2-NEXT:    [[CONV:%.*]] = sext i16 [[TMP17]] to i32
; AVX2-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds i16, ptr [[S2]], i64 [[TMP16]]
; AVX2-NEXT:    [[TMP18:%.*]] = load i16, ptr [[ARRAYIDX4]], align 2
; AVX2-NEXT:    [[CONV5:%.*]] = sext i16 [[TMP18]] to i32
; AVX2-NEXT:    [[MUL6:%.*]] = mul nsw i32 [[CONV5]], [[CONV]]
; AVX2-NEXT:    [[TMP19:%.*]] = or disjoint i64 [[TMP16]], 1
; AVX2-NEXT:    [[ARRAYIDX10:%.*]] = getelementptr inbounds i16, ptr [[S1]], i64 [[TMP19]]
; AVX2-NEXT:    [[TMP20:%.*]] = load i16, ptr [[ARRAYIDX10]], align 2
; AVX2-NEXT:    [[CONV11:%.*]] = sext i16 [[TMP20]] to i32
; AVX2-NEXT:    [[ARRAYIDX15:%.*]] = getelementptr inbounds i16, ptr [[S2]], i64 [[TMP19]]
; AVX2-NEXT:    [[TMP21:%.*]] = load i16, ptr [[ARRAYIDX15]], align 2
; AVX2-NEXT:    [[CONV16:%.*]] = sext i16 [[TMP21]] to i32
; AVX2-NEXT:    [[MUL17:%.*]] = mul nsw i32 [[CONV16]], [[CONV11]]
; AVX2-NEXT:    [[ADD18:%.*]] = add nsw i32 [[MUL17]], [[MUL6]]
; AVX2-NEXT:    [[ARRAYIDX20:%.*]] = getelementptr inbounds i32, ptr [[D1]], i64 [[INDVARS_IV]]
; AVX2-NEXT:    store i32 [[ADD18]], ptr [[ARRAYIDX20]], align 4
; AVX2-NEXT:    [[INDVARS_IV_NEXT]] = add nuw nsw i64 [[INDVARS_IV]], 1
; AVX2-NEXT:    [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INDVARS_IV_NEXT]], [[WIDE_TRIP_COUNT]]
; AVX2-NEXT:    br i1 [[EXITCOND_NOT]], label [[FOR_END_LOOPEXIT]], label [[FOR_BODY]], !llvm.loop [[LOOP3:![0-9]+]]
; AVX2:       for.end.loopexit:
; AVX2-NEXT:    br label [[FOR_END]]
; AVX2:       for.end:
; AVX2-NEXT:    ret void
;
entry:
  %cmp30 = icmp sgt i32 %n, 0
  br i1 %cmp30, label %for.body.preheader, label %for.end

for.body.preheader:
  %wide.trip.count = zext i32 %n to i64
  br label %for.body

for.body:
  %indvars.iv = phi i64 [ 0, %for.body.preheader ], [ %indvars.iv.next, %for.body ]
  %0 = shl nuw nsw i64 %indvars.iv, 1
  %arrayidx = getelementptr inbounds i16, ptr %s1, i64 %0
  %1 = load i16, ptr %arrayidx, align 2
  %conv = sext i16 %1 to i32
  %arrayidx4 = getelementptr inbounds i16, ptr %s2, i64 %0
  %2 = load i16, ptr %arrayidx4, align 2
  %conv5 = sext i16 %2 to i32
  %mul6 = mul nsw i32 %conv5, %conv
  %3 = or disjoint i64 %0, 1
  %arrayidx10 = getelementptr inbounds i16, ptr %s1, i64 %3
  %4 = load i16, ptr %arrayidx10, align 2
  %conv11 = sext i16 %4 to i32
  %arrayidx15 = getelementptr inbounds i16, ptr %s2, i64 %3
  %5 = load i16, ptr %arrayidx15, align 2
  %conv16 = sext i16 %5 to i32
  %mul17 = mul nsw i32 %conv16, %conv11
  %add18 = add nsw i32 %mul17, %mul6
  %arrayidx20 = getelementptr inbounds i32, ptr %d1, i64 %indvars.iv
  store i32 %add18, ptr %arrayidx20, align 4
  %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
  %exitcond.not = icmp eq i64 %indvars.iv.next, %wide.trip.count
  br i1 %exitcond.not, label %for.end.loopexit, label %for.body

for.end.loopexit:
  br label %for.end

for.end:
  ret void
}