Checkpoint Z4C CUDA optimization progress

This commit is contained in:
2026-05-02 08:55:25 +08:00
parent a5c8188305
commit fcd98649f6
4 changed files with 180 additions and 24 deletions

View File

@@ -679,7 +679,11 @@ void Z4c_class::Step(int lev, int YN)
MPI_Abort(MPI_COMM_WORLD, 1);
}
Parallel::Sync_cached(GH->PatL[lev], SynchList_pre, Symmetry, sync_cache_pre[lev]);
{
Parallel::AsyncSyncState async_pre;
Parallel::Sync_start(GH->PatL[lev], SynchList_pre, Symmetry, sync_cache_pre[lev], async_pre);
Parallel::Sync_finish(sync_cache_pre[lev], async_pre, SynchList_pre, Symmetry);
}
if ((lev == a_lev) && (LastAnas + dT_lev >= AnasTime))
z4c_cuda_download_level_state(GH->PatL[lev], SynchList_pre, myrank, false);
@@ -778,7 +782,11 @@ void Z4c_class::Step(int lev, int YN)
MPI_Abort(MPI_COMM_WORLD, 1);
}
Parallel::Sync_cached(GH->PatL[lev], SynchList_cor, Symmetry, sync_cache_cor[lev]);
{
Parallel::AsyncSyncState async_cor;
Parallel::Sync_start(GH->PatL[lev], SynchList_cor, Symmetry, sync_cache_cor[lev], async_cor);
Parallel::Sync_finish(sync_cache_cor[lev], async_cor, SynchList_cor, Symmetry);
}
if (BH_num > 0 && lev == GH->levels - 1)
{