Number of cores settings in sampling

Dear all,

Good day.
I find some strange behavior during sampling.
whether we set core=1 or cores = 4, I found out that the the program still use the maximum number of cores available in the computer.

Did someone notice this or can pymc developer please help answer my enquiry?

Thank you very much

Dear PyMC Developers @junpenglao
Good day.


May I know why the system use all cores even I set the number of cores = 1?
May I know the proper way to set up or constraint the number of cores?

Thank you very much[quote=“GMCobraz_T, post:1, topic:7941, full:true”]
Dear all,

Good day.
I find some strange behavior during sampling.
whether we set core=1 or cores = 4, I found out that the the program still use the maximum number of cores available in the computer.

Did someone notice this or can pymc developer please help answer my enquiry?

Thank you very much.

I think most of what @aseyboldt said here should still apply: NUTS uses all cores - #10 by aseyboldt
Could you try the solution listed there?

Hi @junpenglao ,

The solution still applies.
After I adjust the OMP_NUM_THREADS, it solve my problem.
https://theano-pymc.readthedocs.io/en/latest/tutorial/multi_cores.html

Thanks alot.