CPU usage

[wangjiawen2013]

Hi,
I trained a model (70000 cells and 4000 genes) on my server, I found that it used all the CPUs(4 physical cores and 64 logical processors) available by default. How to limit the computation resource it occupied ?

here are info of my server:
LSB Version: :core-4.1-amd64:core-4.1-noarch:cxx-4.1-amd64:cxx-4.1-noarch:desktop-4.1-amd64:desktop-4.1-noarch:languages-4.1-amd64:languages-4.1-noarch:printing-4.1-amd64:printing-4.1-noarch
Distributor ID: CentOS
Description: CentOS Linux release 7.5.1804 (Core)
Release: 7.5.1804
Codename: Core

64 Intel(R) Xeon(R) Gold 6130 CPU @ 2.10GHz

[adamgayoso]

You'll want to use this:

https://pytorch.org/docs/stable/generated/torch.set_num_threads.html

On that note, we should provide a convenience wrapper of this in scvi.settings

[wangjiawen2013]

Thanks for your help ! It works well.
And I find an interesting thing. It ran faster using only 20 threads (set torch.set_num_threads(20) explicitly) than all the threads (withouting setting torch.set_num_threads, all the 64 threads by default in my case).