Sklearn used n_jobs parameter in multiple functions. We can pass how many cores we want to use. For example 1 for one core and -1 for all cores. My system has six cores and I want to use [0,1,3,4] cores only. Is this possible?
Asked
Active
Viewed 233 times
1
Yahya
- 13,349
- 6
- 30
- 42
Talha Anwar
- 2,699
- 4
- 23
- 62
2 Answers
1
I've never tried nor heard about this possibility before, with scikit-learn.
Still, as scikit-learn uses joblib, it might be the right direction, for you to investigate, to start looking for how to implement your custom ParallelBackendBase, which is a helper abc that defines all methods a ParallelBackend must implement 1, 2, 3; you might be able to implement internally CPU Affinity to pin a process to a particular CPU core.
Nevertheless, it is purely experimental.
Yahya
- 13,349
- 6
- 30
- 42
0
No, that is not possible: you can't specify specific cores, that is up to the system.
9769953
- 10,344
- 3
- 26
- 37