Skip to content

Add CUDA-aware MPI halo exchange for MPMesh#88

Open
Shahrear2000 wants to merge 1 commit into
SCOREC:dn/Velocity_solver_improvementfrom
Shahrear2000:pmpo_MPMesh_improved
Open

Add CUDA-aware MPI halo exchange for MPMesh#88
Shahrear2000 wants to merge 1 commit into
SCOREC:dn/Velocity_solver_improvementfrom
Shahrear2000:pmpo_MPMesh_improved

Conversation

@Shahrear2000

Copy link
Copy Markdown

Main changes:

  • Adds CUDA-aware MPI communication path guarded by CUDA_AWARE_MPI.
  • Keeps CPU-GPU deep_copy path as fallback.
  • Uses cached metadata and per-neighbor GPU buffers for repeated halo exchanges.

Runtime notes:

  • Requires MPICH_GPU_SUPPORT_ENABLED=1 for CUDA-aware MPI on Perlmutter.
  • Can disable CUDA-aware path at runtime with POLYMPO_DISABLE_CUDA_AWARE_MPI=1.

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.

1 participant