running Glomap with subprocess In the Jupyter Notebook, during "estimate pose" phase, only one CPU core is running.

colmap / glomap

GLOMAP - Global Structured-from-Motion Revisited

BSD 3-Clause "New" or "Revised" License

1.3k stars 76 forks source link

running Glomap with subprocess In the Jupyter Notebook, during "estimate pose" phase, only one CPU core is running. #72

Open zhj296409022 opened 3 weeks ago

zhj296409022 commented 3 weeks ago

I can't figure out what the problem is.

lpanaf commented 3 weeks ago

Hi, this should be the expected behavior, we are now not utilizing CUDA

zhj296409022 commented 3 weeks ago

i mean, when i run glomap in terminal, it will use almost my cpu core, but it only use one core when i run glomap with subprocess in jupyter notebook environment.

run in jupyter notebook environment

run in terminal

lpanaf commented 3 weeks ago

It seems that the openmp is not properly functioning from your screenshot. Can you maybe check to openmp condition for your python environment? I tried to call GLOMAP from python with subprocess as you did, and it seems to use a similar amount of resources as directly calling it from the terminal