Multiply the low unsigned 32-bit integers from each packed 64-bit element in a and b, and return the unsigned 64-bit results.