And see if it can be sped up through vectorization: https://software.intel.com/en-us/articles/fast-computation-of-adler32-checksums