Closed nikostr closed 2 years ago
As far as I remember, that option was only helpful when mash is run on multiple files so each file is processed within a thread (which Skmer does the same thing), but not using multithreading on a single file. Can you double-check that? If they have added that feature, we should definitely add it to Skmer. Otherwise, it requires re-implementing the whole hashing part to make parallelization possible which is a standalone (and not easy to me) problem.
Looking at the documentation of
mash sketch
it looks as if this is possible to parallelize beyond running one thread per sample by using the-p
flag. Would this be desirable? I could probably help implement this in the coming days if that would help. I guess it would basically just mean updating thesketch
-call to look like the call tojellyfish
?