porespy
porespy copied to clipboard
Memory leak in `diffusive_size_factor_dns`
Discussed in https://github.com/PMEAL/porespy/discussions/834
Originally posted by oscarhx May 16, 2023 Hello Profs. As stated in the title, the 64G memory goes out when the number of pore pairs reaching several thousands, far behind the number of pore pairs and conductivity required for AI trainning.
Thanks for reporting this. It's a new function so was bound to have a few rough edges.
I think this was coming from openpnm which was used to do the finite difference calculations in the backend. I have pushed a fix for this and it's on pypi. I wonder if you're able to check now and see if the problem is still there?