Hello, I have been running a reaction-diffusion problem which is getting quite large. Before deploying it on an HPC, I played around with a few cores on my laptop but noticed that running time increases substantially with increasing number of cores used. I have looked at the similar discussion top…

[image] cvillette: Also, I get the following warning: [WARNING] yaksa: 3 leaked handle pool objects. I don’t understand it. This is due to garbage management in PETSc, see: Add explicit destruction of PETSc Python objects by garth-wells · Pull Request #2560 · FEniCS/dolfinx · GitHub which re…

Thank you for your quick reply! [image] dokken: This is due to garbage management in PETSc, see: Add explicit destruction of PETSc Python objects by garth-wells · Pull Request #2560 · FEniCS/dolfinx · GitHub which resolved this. Sorry but I am not familiar with pull requests. Was this edit…

[image] cvillette: Sorry but I am not familiar with pull requests. Was this edit added into the latest dolfinx version? If not, how can I best use it? Depending on how you create your singularity image, you should be able to pull the latest version of dolfinx/dolfinx:nightly and rebuild your …

[image] dokken: Depending on how you create your singularity image, you should be able to pull the latest version of dolfinx/dolfinx:nightly and rebuild your image on that. This is what I did but the warning remained. Any additional command I should add in my code itself? Thank you

What is the output of python3 -c "import dolfinx.common; print(dolfinx.common.git_commit_hash)"? When I run the code with dfee3c4449625a1141650e3c678487627a6a01b9 I do not get any of these leaks.

My output is 72c70a59b8bed7cc5caa4c68a84182c9c4be87cf. Is that an older version then?

[image] cvillette: 72c70a59b8bed7cc5caa4c68a84182c9c4be87cf Your version corresponds to: Bump to PETSc 3.18 FEniCS:main ← FEniCS:garth/docker-petsc-bump opened 08:25PM - 26 Nov 22 UTC (UTC) …

I was indeed not updating the container version properly. I tried again pulling the latest version of dolfinx:nightly > singularity pull docker://dolfinx/dolfinx:nightly I ended up with version fecad69716105cadda8cd43e52a630127a3af853 but still got the leaked handle pool object warning.

I cannot reproduce the leaked objects with the docker container with the same commit hash on my system.

All right. Then maybe it’s an issue with the singularity container translation from Docker? After changing the dolfinx version, I now get an error in my code when computing the mesh size. I think I understand some function variable formats have been modified but I don’t know how to make the compati…

Problem solving in parallel slower than in serial

General

dokken April 3, 2023, 10:42am 12

See Applying BC in parallel (elastic bar in dynamic explicit) - #6 by dokken

Topic		Replies	Views
How to further speed up the calculation in doflinx? Uncategorized	5	1174	February 26, 2025
Parallel slower runtime, MPI Anaconda dolfinx	7	646	January 17, 2024
Different results in serial and parallel run dolfinx dolfinx	23	3062	October 26, 2020
Dolfinx crashes with a bit computation load; How? dolfinx mesh , dolfinx , mpi	10	430	March 3, 2024
Parallel slower runtime dolfinx	6	1222	February 19, 2024

Problem solving in parallel slower than in serial

Related topics