Determining number of CPUs for queries and iterators

Question

Determining number of CPUs for queries and iterators

vadim opened this issue 3 years ago · comments

Using timsPy to make queries on a 40-core machine, I notice that the python interpreter usage spikes to 2000+% CPU. How is the maximal number of threads set? I've been unable to find references to OpenMP in the code.

Matteo · Answer 1 · Fri Aug 06 2021 17:42:44 GMT+0800 (China Standard Time)

Hello Vadim,

I think it might be because of the opentims_bruker_bridge which is basically Bruker's code for doing transformations between TOF and MZ and scan an inverse ion mobility. We are dealing with that part right now, especially that Michal has written a nice new multithreaded version of the code and is trying to turn them off too.

We will stay in touch,

Best,

Vadim Patsalo · Answer 2 · Sat Aug 07 2021 00:28:52 GMT+0800 (China Standard Time)

Thank you so much for your reply, @MatteoLacki, appreciate it! That's what I suspected as well. Look forward to hearing more of your progress. The project is fantastic, and it's great to hear that Bruker is supporting you guys.

Michał Startek · Answer 3 · Wed Aug 11 2021 01:58:54 GMT+0800 (China Standard Time)

Hi,

The aforementioned code for setting the number of threads is finished, so if you want to be our beta teste^H^H^H^H^H early adopter, you can grab opentims v1.0.9a2 (or master branch on github) and call opentims.set_num_threads(...)

Note: it's in opentims, not timspy - timspy just uses opentims under the hood.