0

I am trying same GluonCV inferencing examples. However I want to benchmark the results on a single core. I tried to play with ctx gave cpu(0) but still unable to limit the number of cores which are getting utilized.

I Tried the following it works but this will set for the complete script I want to set for just for inferencing import os os.environ["OMP_NUM_THREADS"] = 1 # export OMP_NUM_THREADS=1

Scorpio
  • 511
  • 4
  • 14
  • why don't you make a separate script just for your inference? Why do you want to benchmark on one core? – Olivier Cruchant Apr 16 '20 at 22:22
  • I want to benchmark only on one core because in my production env I have the luxury to use only a single core. I am not sure If forking a separate process will guarantee to run it on the same core. – Scorpio Apr 18 '20 at 10:25

0 Answers0