The moment you're glad you put that big multi-core CPU in your PC instead of a fancier GPU two years ago, when you manage to run your statistics simulation run on all cores.
0 Boosts
0 Likes
@fullywoolly i started using doParallel in R and on each run it saves me about a minute, and probably hours during the whole day
TIL: parallelizing R code is surprisingly easy with the doParallel package!
Also: holy shit dplyr is slow sometimes
I stripped out all the convenience stuff I used it for in places that churned through a lot of data and replaced it with base R code and I gained orders of magnitude in speed
Be careful when you use it
@aesthr I have to confess I'm not good at R, I'm always suffer a bit from the unusual syntax.
But I'm currently digging in deeper as my niece started to study sociology (with a specialization in technology) and they will be doing the data analysis stuff in R 
@geeeero @sashag it takes some getting used to, that's true. but it's a very powerful language for this particular domain.
The tidyverse stuff is helpful but also sometimes quite slow. I just sped up my simulation by orders of magnitude by stripping out all the dplyr stuff and doing base R instead where performance matters
@aesthr what did you do to make it utilize the cores like that? And what kind of time savings did it offer you in actual time ... Like hours to minutes? Because that feels so good when it happens.