Make inner loop process 16 bytes at a time with SSE instructions This is an attempt to speed up the inner loop by using 128-bit registers. Currently stalled; I think there's no way to do the IMULQ equivalent on two packed quadwords so this doesn't seem feasible.
xxhash is a Go implementation of the 64-bit xxHash algorithm, XXH64. This is a high-quality hashing algorithm that is much faster than anything in the Go standard library.
On amd64 there is an even faster assembly implementation that runs at over 10 GB/s on my laptop.