mmmagic
mmmagic copied to clipboard
mscdex
Better Performance
Briefly looking over the changes it looks like the purpose of this is to add some sort of magic_set pool? Have you compared performance with a std::queue? Maybe a better solution might be to have a global pool (using a MPMC queue such as this) with some sensible upper bound on size?
I don't want to sound rude, but only looking at the numbers provided by @magicode it seems to be arguably faster than the current way - in case it would be even faster using std::queue, does that matter as long as nobody implements it?
I think that this should be merged (unless I am overlooking some bugs or bad code), as it seems to be around 5x faster than the current solution (looking at the provided benchmark) - regardless of wether or not it could be even faster, that still looks like a valuable improvement.
