100trillionUSD / bitcoin

258 stars 65 forks source link

10% performance gain #13

Open FrancisBehnen opened 3 years ago

FrancisBehnen commented 3 years ago

I'm a bit stumped that pandas is so bad at working with ragged csv's (https://github.com/100trillionUSD/bitcoin/pull/11)...

Anyway, here are some improvements I thought of along the way. I guess this is it on a single thread. Multiprocessing or an implementation in C directly seems the way to go from here.