64-BitBrainstorm_2026/AMSS-NCKU
8
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Optimize MPI communication in RestrictProlong and surface_integral

Browse Source 
Cache Sync in RestrictProlong: replace 11 basic Parallel::Sync() calls
with Parallel::Sync_cached() across RestrictProlong, RestrictProlong_aux,
and ProlongRestrict to avoid rebuilding grid segment lists every call.

Merge paired MPI_Allreduce in surface_integral: combine 9 pairs of
consecutive RP/IP Allreduce calls into single calls with count=2*NN.

Merge scalar MPI_Allreduce in surf_MassPAng: combine 3 groups of 7
scalar Allreduce calls (mass + angular/linear momentum) into single
calls with count=7.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
CGH0S7
2026-02-09 22:07:12 +08:00
parent 42b9cf1ad9
commit 738498cb28
3 changed files with 154 additions and 63 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
AMSS_NCKU_source/bssn_class.h
View File

@@ -128,6 +128,8 @@ public:
Parallel::SyncCache *sync_cache_pre; // per-level cache for predictor sync
Parallel::SyncCache *sync_cache_cor; // per-level cache for corrector sync
Parallel::SyncCache *sync_cache_rp_coarse; // RestrictProlong sync on PatL[lev-1]
Parallel::SyncCache *sync_cache_rp_fine; // RestrictProlong sync on PatL[lev]
monitor *ErrorMonitor, *Psi4Monitor, *BHMonitor, *MAPMonitor;
monitor *ConVMonitor;