Skip to content

LAMMPS multiple MPI instances over limited GPUs (multiple repllica simulations)#1

Open
Felixrccs wants to merge 61 commits into
ACEsuit:macefrom
Felixrccs:mace
Open

LAMMPS multiple MPI instances over limited GPUs (multiple repllica simulations)#1
Felixrccs wants to merge 61 commits into
ACEsuit:macefrom
Felixrccs:mace

Conversation

@Felixrccs

Copy link
Copy Markdown

All details in ACEsuit/mace#195

Debug of MPI distibution over GPUs in ML-MACE

@ilyes319

ilyes319 commented Dec 4, 2023

Copy link
Copy Markdown

Thank you! @wcwitt @bernstei Could you have a look?

@bernstei

bernstei commented Dec 4, 2023

Copy link
Copy Markdown

Looks the same as the syntax I suggested. The only optional missing bit is error checking on the CUDA API call, but the functionality it's using is so basic it hardly feels worth it.

@wcwitt wcwitt mentioned this pull request Jan 6, 2024
10 tasks
@tomasfbouvier

tomasfbouvier commented Dec 20, 2024

Copy link
Copy Markdown

was this pull request merged into the original lammps repository? It seems a PR which goes beyond MACE and inherent to the REPLICA package

@Felixrccs

Copy link
Copy Markdown
Author

I dont think it was ever merged. It mainly deals with how mace-lammps calls multiple GPUs. Should be independent of the Replica package. (That was just my usecase)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants