Skip to content

Comments

Compiler option for GCHP load balance in chemistry#532

Open
lizziel wants to merge 1 commit intodev/no-diff-to-benchmarkfrom
feature/gchp_mpi_load_balance_in_chemistry
Open

Compiler option for GCHP load balance in chemistry#532
lizziel wants to merge 1 commit intodev/no-diff-to-benchmarkfrom
feature/gchp_mpi_load_balance_in_chemistry

Conversation

@lizziel
Copy link
Contributor

@lizziel lizziel commented Jan 22, 2026

Name and Institution (Required)

Name: Lizzie Lundgren
Institution: Harvard University

Describe the update

Add CMake option for turning MPI load balancing on/off. It will be on by default. Turn off by building with:
-DMPI_LOAD_BALANCE=n

There is a companion PR with this update:
geoschem/geos-chem#3015

Expected changes

MPI_LOAD_BALANCE will appear in the messages during the configure stage indicating whether it is on or off. GCHP chemistry will be faster with it on (default). I find the benchmark simulation to be 20% faster on the Harvard Cannon cluster using 96 cores.

Reference(s)

None

Related Github Issue

@lizziel lizziel marked this pull request as draft January 22, 2026 16:39
@lizziel lizziel changed the title Feature/gchp mpi load balance in chemistry Compiler option for GCHP load balance in chemistry Jan 22, 2026
@lizziel lizziel added topic: Build Related to makefiles or the build sequence no-diff-to-benchmark This update will not change the results of fullchem benchmark simulations topic: Performance Related to GCHP model speed and/or memory labels Jan 22, 2026
@lizziel lizziel added this to the 14.7.1 milestone Jan 22, 2026
@yantosca yantosca force-pushed the main branch 6 times, most recently from 82d6639 to 3216f28 Compare February 8, 2026 02:05
@yantosca yantosca marked this pull request as ready for review February 9, 2026 20:27
@yantosca yantosca changed the base branch from main to dev/no-diff-to-benchmark February 9, 2026 20:28
Copy link
Contributor

@yantosca yantosca left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks good but we need a CHANGELOG.md update.

@yantosca yantosca self-assigned this Feb 9, 2026
@yantosca yantosca added the category: Feature Request New feature or request label Feb 9, 2026
@yantosca yantosca force-pushed the dev/no-diff-to-benchmark branch from 55c0e55 to 2904178 Compare February 12, 2026 15:20
…lt=on

To turn off, build with -DMPI_LOAD_BALANCE=n

Signed-off-by: Lizzie Lundgren <elundgren@seas.harvard.edu>
@lizziel lizziel force-pushed the feature/gchp_mpi_load_balance_in_chemistry branch from 8f5214d to 654920b Compare February 17, 2026 14:17
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

category: Feature Request New feature or request no-diff-to-benchmark This update will not change the results of fullchem benchmark simulations topic: Build Related to makefiles or the build sequence topic: Performance Related to GCHP model speed and/or memory

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants