In writing a fast Hamming distance library in Go, dropping inline popcnt64 in using cgo was far slower than good, old-fashioned lookup table method for a single

In writing a fast Hamming distance library in Go, dropping inline popcnt64 in using cgo was far slower than good, old-fashioned lookup table method for a single byte on a Haswell mobile processor. For the uint64 case, bitfolding was still faster than popcnt64. Perhaps for the vector buffer case, popcnt64 is faster.