define i32 @sad(i8* nocapture readonly %pix1, i8* nocapture readonly %pix2) #0 {
entry:
  br label %vector.body

vector.body:                                      ; preds = %entry
  %0 = bitcast i8* %pix1 to <4 x i8>*
  %wide.load = load <4 x i8>* %0, align 1
  %1 = zext <4 x i8> %wide.load to <4 x i32>
  %2 = bitcast i8* %pix2 to <4 x i8>*
  %wide.load7 = load <4 x i8>* %2, align 1
  %3 = zext <4 x i8> %wide.load7 to <4 x i32>
  %4 = sub nsw <4 x i32> %1, %3
  %5 = icmp uge <4 x i8> %wide.load, %wide.load7
  %6 = sub nsw <4 x i32> zeroinitializer, %4
  %7 = select <4 x i1> %5, <4 x i32> %4, <4 x i32> %6
  %8 = getelementptr inbounds i8* %pix1, i64 4
  %9 = bitcast i8* %8 to <4 x i8>*
  %wide.load.1 = load <4 x i8>* %9, align 1
  %10 = zext <4 x i8> %wide.load.1 to <4 x i32>
  %11 = getelementptr inbounds i8* %pix2, i64 4
  %12 = bitcast i8* %11 to <4 x i8>*
  %wide.load7.1 = load <4 x i8>* %12, align 1
  %13 = zext <4 x i8> %wide.load7.1 to <4 x i32>
  %14 = sub nsw <4 x i32> %10, %13
  %15 = icmp uge <4 x i8> %wide.load.1, %wide.load7.1
  %16 = sub nsw <4 x i32> zeroinitializer, %14
  %17 = select <4 x i1> %15, <4 x i32> %14, <4 x i32> %16
  %18 = add nsw <4 x i32> %17, %7
  %19 = getelementptr inbounds i8* %pix1, i64 8
  %20 = bitcast i8* %19 to <4 x i8>*
  %wide.load.2 = load <4 x i8>* %20, align 1
  %21 = zext <4 x i8> %wide.load.2 to <4 x i32>
  %22 = getelementptr inbounds i8* %pix2, i64 8
  %23 = bitcast i8* %22 to <4 x i8>*
  %wide.load7.2 = load <4 x i8>* %23, align 1
  %24 = zext <4 x i8> %wide.load7.2 to <4 x i32>
  %25 = sub nsw <4 x i32> %21, %24
  %26 = icmp uge <4 x i8> %wide.load.2, %wide.load7.2
  %27 = sub nsw <4 x i32> zeroinitializer, %25
  %28 = select <4 x i1> %26, <4 x i32> %25, <4 x i32> %27
  %29 = add nsw <4 x i32> %28, %18
  %30 = getelementptr inbounds i8* %pix1, i64 12
  %31 = bitcast i8* %30 to <4 x i8>*
  %wide.load.3 = load <4 x i8>* %31, align 1
  %32 = zext <4 x i8> %wide.load.3 to <4 x i32>
  %33 = getelementptr inbounds i8* %pix2, i64 12
  %34 = bitcast i8* %33 to <4 x i8>*
  %wide.load7.3 = load <4 x i8>* %34, align 1
  %35 = zext <4 x i8> %wide.load7.3 to <4 x i32>
  %36 = sub nsw <4 x i32> %32, %35
  %37 = icmp uge <4 x i8> %wide.load.3, %wide.load7.3
  %38 = sub nsw <4 x i32> zeroinitializer, %36
  %39 = select <4 x i1> %37, <4 x i32> %36, <4 x i32> %38
  %40 = add nsw <4 x i32> %39, %29
  %rdx.shuf = shufflevector <4 x i32> %40, <4 x i32> undef, <4 x i32> <i32 2, i32 3, i32 undef, i32 undef>
  %bin.rdx = add <4 x i32> %40, %rdx.shuf
  %rdx.shuf8 = shufflevector <4 x i32> %bin.rdx, <4 x i32> undef, <4 x i32> <i32 1, i32 undef, i32 undef, i32 undef>
  %bin.rdx9 = add <4 x i32> %bin.rdx, %rdx.shuf8
  %41 = extractelement <4 x i32> %bin.rdx9, i32 0
  ret i32 %41
}