Hi,
I'm close to finishing vectorizing the mem* and str* functions in glibc and was thinking of trying to work on optimizing some MySQL internals. Apart from the strings/ folder which include many mem*/str* functions that will be easily vectorized now that I know the generic mechanism, I'd like to work also on sth else. After Neko's suggestion, I plan to take a look at the hashing functions in bdb/hash/, and specifically, the functions in bdb/hash/hash_func.c. Apparently some or all of them can be vectorized and the performance will be probably quite improved. What I would like to ask is if there's any other place I should look first. I'm checking on the 4.1 tree (using the Debian package because i'm lazy to configure everything myself

Konstantinos