Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Determine --procsPerNode automatically #141

Closed
vasdommes opened this issue Oct 10, 2023 · 0 comments
Closed

Determine --procsPerNode automatically #141

vasdommes opened this issue Oct 10, 2023 · 0 comments
Assignees
Milestone

Comments

@vasdommes
Copy link
Collaborator

Currently --procsPerNode is passed by user.

However, in MPI-3 we can determine it using shared memory communicator:

MPI_Comm shared_memory_comm;
MPI_Comm_split_type(MPI_COMM_WORLD, MPI_COMM_TYPE_SHARED, 0, MPI_INFO_NULL, &shared_memory_comm);
int procsPerNode = El::mpi::Rank(shared_memory_comm);

NB: we should also assert that each node has the same number of processes, otherwise the block distribution algorithm will fail.

@vasdommes vasdommes self-assigned this Oct 10, 2023
@vasdommes vasdommes added this to the Backlog milestone Nov 14, 2023
@vasdommes vasdommes modified the milestones: Backlog, 2.6.2 Jan 4, 2024
vasdommes added a commit that referenced this issue Jan 5, 2024
Fix #141 Determine --procsPerNode automatically, reuse shared memory communicator, improve output
bharathr98 pushed a commit to bharathr98/sdpb that referenced this issue Mar 1, 2024
--procsPerNode is marked as obsolete and left for backward compatibility.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

1 participant