2 ms·
To be honest I could have probably used signed integers but conceptually representing DNA sequence (4 possibilities at each position) was far easier to think ab
by collyw 12y ago
To be honest I could have probably used signed integers but conceptually representing DNA sequence (4 possibilities at each position) was far easier to think about that two's compliment. The idea was to make everything as small and lean as possible, to get the maximum amount into memory, and it seemed to work.
Bitcount / popcount / hamming distance as a CPU operation will have saved a lot of CPU cycles over the PG method I tried.
You sound like you know Postgres fairly well. How feasible would it be to write a CPU based popcount extension?