Processor ressource management on windows


#1

Hello!

I usually fit my models using rstan or brms in my personnal laptop running under ubuntu. In order to free my laptop, I tried to run distributional gamm on windows at my lab (with brms). However, I detected a ressource allocation which seems quite strange (given my habits on ubuntu).

As shown in the attached screen shot, the processor ressources are allocated to a unique process (RStudio R Session), while what I’m used to consider as my sampling processes (R for windows front-end) almost do not use the processor (~7%). The “Rstudio R Session” did not take such ressources at the beginning of the sampling process, though.

Anyone has an idea about why this is happening? Is it slowing the sampling process, or just an informatic technicality (I’m really not used to windows nor in love with it…)

Thank you!
Lucas


#2

For precision, this computer runs under windows 10, and ressources allocated to “Rstudio R Session” process increase with time!


#3

You might want to try grabbing a 64-bit R. Looks like you have some 32-bit processors lying around.

Try running from the R terminal rather than from RStudio—it won’t introduce any additional overhead. Then you can see if the problem’s coming from RStudio or from brms (which includes coming from Stan).