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

perf(RestrictProlong): 用 Restrict_cached/OutBdLow2Hi_cached 替换非缓存版本，Sync_finish 改为渐进式解包

Browse Source 
- RestrictProlong/RestrictProlong_aux 中的 Restrict() 和 OutBdLow2Hi() 替换为 _cached 版本，
  复用 gridseg 列表和 MPI 缓冲区，避免每次调用重新分配
- 新增 sync_cache_restrict/sync_cache_outbd 两组 per-level 缓存
- Sync_finish 从 MPI_Waitall 改为 MPI_Waitsome 渐进式解包，降低尾延迟
- AsyncSyncState 扩展 req_node/req_is_recv/pending_recv 字段支持渐进解包

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
CGH0S7
2026-03-02 20:48:38 +08:00
parent b3c367f15b
commit 4012e9d068
4 changed files with 527 additions and 483 deletions
Download Patch File Download Diff File Expand all files Collapse all files

53
AMSS_NCKU_source/Parallel.C
View File

@@ -4471,7 +4471,6 @@ void Parallel::transfer_cached(MyList<Parallel::gridseg> **src, MyList<Parallel:
delete[] req_is_recv; delete[] req_is_recv;
delete[] completed; delete[] completed;
} }
// Sync_cached: build grid segment lists on first call, reuse on subsequent calls
void Parallel::Sync_cached(MyList<Patch> *PatL, MyList<var> *VarList, int Symmetry, SyncCache &cache) void Parallel::Sync_cached(MyList<Patch> *PatL, MyList<var> *VarList, int Symmetry, SyncCache &cache)
{ {
if (!cache.valid) if (!cache.valid)
@@ -4669,6 +4668,11 @@ void Parallel::Sync_start(MyList<Patch> *PatL, MyList<var> *VarList, int Symmetr
int cpusize = cache.cpusize; int cpusize = cache.cpusize;
state.req_no = 0; state.req_no = 0;
state.active = true; state.active = true;
state.pending_recv = 0;
// Allocate tracking arrays
delete[] state.req_node; delete[] state.req_is_recv;
state.req_node = new int[cache.max_reqs];
state.req_is_recv = new int[cache.max_reqs];
MyList<Parallel::gridseg> **src = cache.combined_src; MyList<Parallel::gridseg> **src = cache.combined_src;
MyList<Parallel::gridseg> **dst = cache.combined_dst; MyList<Parallel::gridseg> **dst = cache.combined_dst;
@@ -4713,6 +4717,8 @@ void Parallel::Sync_start(MyList<Patch> *PatL, MyList<var> *VarList, int Symmetr
cache.send_buf_caps[node] = slength; cache.send_buf_caps[node] = slength;
} }
data_packer(cache.send_bufs[node], src[myrank], dst[myrank], node, PACK, VarList, VarList, Symmetry); data_packer(cache.send_bufs[node], src[myrank], dst[myrank], node, PACK, VarList, VarList, Symmetry);
state.req_node[state.req_no] = node;
state.req_is_recv[state.req_no] = 0;
MPI_Isend((void *)cache.send_bufs[node], slength, MPI_DOUBLE, node, 2, MPI_COMM_WORLD, cache.reqs + state.req_no++); MPI_Isend((void *)cache.send_bufs[node], slength, MPI_DOUBLE, node, 2, MPI_COMM_WORLD, cache.reqs + state.req_no++);
} }
int rlength; int rlength;
@@ -4730,29 +4736,60 @@ void Parallel::Sync_start(MyList<Patch> *PatL, MyList<var> *VarList, int Symmetr
cache.recv_bufs[node] = new double[rlength]; cache.recv_bufs[node] = new double[rlength];
cache.recv_buf_caps[node] = rlength; cache.recv_buf_caps[node] = rlength;
} }
state.req_node[state.req_no] = node;
state.req_is_recv[state.req_no] = 1;
state.pending_recv++;
MPI_Irecv((void *)cache.recv_bufs[node], rlength, MPI_DOUBLE, node, 2, MPI_COMM_WORLD, cache.reqs + state.req_no++); MPI_Irecv((void *)cache.recv_bufs[node], rlength, MPI_DOUBLE, node, 2, MPI_COMM_WORLD, cache.reqs + state.req_no++);
} }
} }
} }
cache.lengths_valid = true; cache.lengths_valid = true;
} }
// Sync_finish: wait for async MPI operations and unpack // Sync_finish: progressive unpack as receives complete, then wait for sends
void Parallel::Sync_finish(SyncCache &cache, AsyncSyncState &state, void Parallel::Sync_finish(SyncCache &cache, AsyncSyncState &state,
MyList<var> *VarList, int Symmetry) MyList<var> *VarList, int Symmetry)
{ {
if (!state.active) if (!state.active)
return; return;
MPI_Waitall(state.req_no, cache.reqs, cache.stats); int myrank;
MPI_Comm_rank(MPI_COMM_WORLD, &myrank);
int cpusize = cache.cpusize;
MyList<Parallel::gridseg> **src = cache.combined_src; MyList<Parallel::gridseg> **src = cache.combined_src;
MyList<Parallel::gridseg> **dst = cache.combined_dst; MyList<Parallel::gridseg> **dst = cache.combined_dst;
for (int node = 0; node < cpusize; node++) // Unpack local data first (no MPI needed)
if (cache.recv_bufs[node] && cache.recv_lengths[node] > 0) if (cache.recv_bufs[myrank] && cache.recv_lengths[myrank] > 0)
data_packer(cache.recv_bufs[node], src[node], dst[node], node, UNPACK, VarList, VarList, Symmetry); data_packer(cache.recv_bufs[myrank], src[myrank], dst[myrank], myrank, UNPACK, VarList, VarList, Symmetry);
// Progressive unpack of remote receives
if (state.pending_recv > 0 && state.req_no > 0)
{
int pending = state.pending_recv;
int *completed = new int[cache.max_reqs];
while (pending > 0)
{
int outcount = 0;
MPI_Waitsome(state.req_no, cache.reqs, &outcount, completed, cache.stats);
if (outcount == MPI_UNDEFINED) break;
for (int i = 0; i < outcount; i++)
{
int idx = completed[i];
if (idx >= 0 && state.req_is_recv[idx])
{
int recv_node = state.req_node[idx];
data_packer(cache.recv_bufs[recv_node], src[recv_node], dst[recv_node], recv_node, UNPACK, VarList, VarList, Symmetry);
pending--;
}
}
}
delete[] completed;
}
// Wait for remaining sends
if (state.req_no > 0) MPI_Waitall(state.req_no, cache.reqs, cache.stats);
delete[] state.req_node; state.req_node = 0;
delete[] state.req_is_recv; state.req_is_recv = 0;
state.active = false; state.active = false;
} }
// collect buffer grid segments or blocks for the periodic boundary condition of given patch // collect buffer grid segments or blocks for the periodic boundary condition of given patch

5
AMSS_NCKU_source/Parallel.h
View File

@@ -121,7 +121,10 @@ namespace Parallel
struct AsyncSyncState { struct AsyncSyncState {
int req_no; int req_no;
bool active; bool active;
AsyncSyncState() : req_no(0), active(false) {} int *req_node;
int *req_is_recv;
int pending_recv;
AsyncSyncState() : req_no(0), active(false), req_node(0), req_is_recv(0), pending_recv(0) {}
}; };
void Sync_start(MyList<Patch> *PatL, MyList<var> *VarList, int Symmetry, void Sync_start(MyList<Patch> *PatL, MyList<var> *VarList, int Symmetry,

44
AMSS_NCKU_source/bssn_class.C
View File

@@ -736,6 +736,8 @@ void bssn_class::Initialize()
sync_cache_cor = new Parallel::SyncCache[GH->levels]; sync_cache_cor = new Parallel::SyncCache[GH->levels];
sync_cache_rp_coarse = new Parallel::SyncCache[GH->levels]; sync_cache_rp_coarse = new Parallel::SyncCache[GH->levels];
sync_cache_rp_fine = new Parallel::SyncCache[GH->levels]; sync_cache_rp_fine = new Parallel::SyncCache[GH->levels];
sync_cache_restrict = new Parallel::SyncCache[GH->levels];
sync_cache_outbd = new Parallel::SyncCache[GH->levels];
} }
//================================================================================================ //================================================================================================
@@ -2213,7 +2215,7 @@ void bssn_class::Evolve(int Steps)
GH->Regrid(Symmetry, BH_num, Porgbr, Porg0, GH->Regrid(Symmetry, BH_num, Porgbr, Porg0,
SynchList_cor, OldStateList, StateList, SynchList_pre, SynchList_cor, OldStateList, StateList, SynchList_pre,
fgt(PhysTime - dT_mon, StartTime, dT_mon / 2), ErrorMonitor); fgt(PhysTime - dT_mon, StartTime, dT_mon / 2), ErrorMonitor);
for (int il = 0; il < GH->levels; il++) { sync_cache_pre[il].invalidate(); sync_cache_cor[il].invalidate(); sync_cache_rp_coarse[il].invalidate(); sync_cache_rp_fine[il].invalidate(); } for (int il = 0; il < GH->levels; il++) { sync_cache_pre[il].invalidate(); sync_cache_cor[il].invalidate(); sync_cache_rp_coarse[il].invalidate(); sync_cache_rp_fine[il].invalidate(); sync_cache_restrict[il].invalidate(); sync_cache_outbd[il].invalidate(); }
#endif #endif
#if (REGLEV == 0 && (PSTR == 1 || PSTR == 2)) #if (REGLEV == 0 && (PSTR == 1 || PSTR == 2))
@@ -2429,7 +2431,7 @@ void bssn_class::RecursiveStep(int lev)
if (GH->Regrid_Onelevel(lev, Symmetry, BH_num, Porgbr, Porg0, if (GH->Regrid_Onelevel(lev, Symmetry, BH_num, Porgbr, Porg0,
SynchList_cor, OldStateList, StateList, SynchList_pre, SynchList_cor, OldStateList, StateList, SynchList_pre,
fgt(PhysTime - dT_lev, StartTime, dT_lev / 2), ErrorMonitor)) fgt(PhysTime - dT_lev, StartTime, dT_lev / 2), ErrorMonitor))
for (int il = 0; il < GH->levels; il++) { sync_cache_pre[il].invalidate(); sync_cache_cor[il].invalidate(); sync_cache_rp_coarse[il].invalidate(); sync_cache_rp_fine[il].invalidate(); } for (int il = 0; il < GH->levels; il++) { sync_cache_pre[il].invalidate(); sync_cache_cor[il].invalidate(); sync_cache_rp_coarse[il].invalidate(); sync_cache_rp_fine[il].invalidate(); sync_cache_restrict[il].invalidate(); sync_cache_outbd[il].invalidate(); }
#endif #endif
} }
@@ -2608,7 +2610,7 @@ void bssn_class::ParallelStep()
if (GH->Regrid_Onelevel(GH->mylev, Symmetry, BH_num, Porgbr, Porg0, if (GH->Regrid_Onelevel(GH->mylev, Symmetry, BH_num, Porgbr, Porg0,
SynchList_cor, OldStateList, StateList, SynchList_pre, SynchList_cor, OldStateList, StateList, SynchList_pre,
fgt(PhysTime - dT_lev, StartTime, dT_lev / 2), ErrorMonitor)) fgt(PhysTime - dT_lev, StartTime, dT_lev / 2), ErrorMonitor))
for (int il = 0; il < GH->levels; il++) { sync_cache_pre[il].invalidate(); sync_cache_cor[il].invalidate(); sync_cache_rp_coarse[il].invalidate(); sync_cache_rp_fine[il].invalidate(); } for (int il = 0; il < GH->levels; il++) { sync_cache_pre[il].invalidate(); sync_cache_cor[il].invalidate(); sync_cache_rp_coarse[il].invalidate(); sync_cache_rp_fine[il].invalidate(); sync_cache_restrict[il].invalidate(); sync_cache_outbd[il].invalidate(); }
#endif #endif
} }
@@ -2775,7 +2777,7 @@ void bssn_class::ParallelStep()
if (GH->Regrid_Onelevel(lev + 1, Symmetry, BH_num, Porgbr, Porg0, if (GH->Regrid_Onelevel(lev + 1, Symmetry, BH_num, Porgbr, Porg0,
SynchList_cor, OldStateList, StateList, SynchList_pre, SynchList_cor, OldStateList, StateList, SynchList_pre,
fgt(PhysTime - dT_levp1, StartTime, dT_levp1 / 2), ErrorMonitor)) fgt(PhysTime - dT_levp1, StartTime, dT_levp1 / 2), ErrorMonitor))
for (int il = 0; il < GH->levels; il++) { sync_cache_pre[il].invalidate(); sync_cache_cor[il].invalidate(); sync_cache_rp_coarse[il].invalidate(); sync_cache_rp_fine[il].invalidate(); } for (int il = 0; il < GH->levels; il++) { sync_cache_pre[il].invalidate(); sync_cache_cor[il].invalidate(); sync_cache_rp_coarse[il].invalidate(); sync_cache_rp_fine[il].invalidate(); sync_cache_restrict[il].invalidate(); sync_cache_outbd[il].invalidate(); }
// a_stream.clear(); // a_stream.clear();
// a_stream.str(""); // a_stream.str("");
@@ -2790,7 +2792,7 @@ void bssn_class::ParallelStep()
if (GH->Regrid_Onelevel(lev, Symmetry, BH_num, Porgbr, Porg0, if (GH->Regrid_Onelevel(lev, Symmetry, BH_num, Porgbr, Porg0,
SynchList_cor, OldStateList, StateList, SynchList_pre, SynchList_cor, OldStateList, StateList, SynchList_pre,
fgt(PhysTime - dT_lev, StartTime, dT_lev / 2), ErrorMonitor)) fgt(PhysTime - dT_lev, StartTime, dT_lev / 2), ErrorMonitor))
for (int il = 0; il < GH->levels; il++) { sync_cache_pre[il].invalidate(); sync_cache_cor[il].invalidate(); sync_cache_rp_coarse[il].invalidate(); sync_cache_rp_fine[il].invalidate(); } for (int il = 0; il < GH->levels; il++) { sync_cache_pre[il].invalidate(); sync_cache_cor[il].invalidate(); sync_cache_rp_coarse[il].invalidate(); sync_cache_rp_fine[il].invalidate(); sync_cache_restrict[il].invalidate(); sync_cache_outbd[il].invalidate(); }
// a_stream.clear(); // a_stream.clear();
// a_stream.str(""); // a_stream.str("");
@@ -2809,7 +2811,7 @@ void bssn_class::ParallelStep()
if (GH->Regrid_Onelevel(lev - 1, Symmetry, BH_num, Porgbr, Porg0, if (GH->Regrid_Onelevel(lev - 1, Symmetry, BH_num, Porgbr, Porg0,
SynchList_cor, OldStateList, StateList, SynchList_pre, SynchList_cor, OldStateList, StateList, SynchList_pre,
fgt(PhysTime - dT_lev, StartTime, dT_levm1 / 2), ErrorMonitor)) fgt(PhysTime - dT_lev, StartTime, dT_levm1 / 2), ErrorMonitor))
for (int il = 0; il < GH->levels; il++) { sync_cache_pre[il].invalidate(); sync_cache_cor[il].invalidate(); sync_cache_rp_coarse[il].invalidate(); sync_cache_rp_fine[il].invalidate(); } for (int il = 0; il < GH->levels; il++) { sync_cache_pre[il].invalidate(); sync_cache_cor[il].invalidate(); sync_cache_rp_coarse[il].invalidate(); sync_cache_rp_fine[il].invalidate(); sync_cache_restrict[il].invalidate(); sync_cache_outbd[il].invalidate(); }
// a_stream.clear(); // a_stream.clear();
// a_stream.str(""); // a_stream.str("");
@@ -2825,7 +2827,7 @@ void bssn_class::ParallelStep()
if (GH->Regrid_Onelevel(lev - 1, Symmetry, BH_num, Porgbr, Porg0, if (GH->Regrid_Onelevel(lev - 1, Symmetry, BH_num, Porgbr, Porg0,
SynchList_cor, OldStateList, StateList, SynchList_pre, SynchList_cor, OldStateList, StateList, SynchList_pre,
fgt(PhysTime - dT_lev, StartTime, dT_levm1 / 2), ErrorMonitor)) fgt(PhysTime - dT_lev, StartTime, dT_levm1 / 2), ErrorMonitor))
for (int il = 0; il < GH->levels; il++) { sync_cache_pre[il].invalidate(); sync_cache_cor[il].invalidate(); sync_cache_rp_coarse[il].invalidate(); sync_cache_rp_fine[il].invalidate(); } for (int il = 0; il < GH->levels; il++) { sync_cache_pre[il].invalidate(); sync_cache_cor[il].invalidate(); sync_cache_rp_coarse[il].invalidate(); sync_cache_rp_fine[il].invalidate(); sync_cache_restrict[il].invalidate(); sync_cache_outbd[il].invalidate(); }
// a_stream.clear(); // a_stream.clear();
// a_stream.str(""); // a_stream.str("");
@@ -5796,7 +5798,7 @@ void bssn_class::RestrictProlong(int lev, int YN, bool BB,
#endif #endif
#if (RPB == 0) #if (RPB == 0)
Parallel::Restrict(GH->PatL[lev - 1], GH->PatL[lev], SL, SynchList_pre, Symmetry); Parallel::Restrict_cached(GH->PatL[lev - 1], GH->PatL[lev], SL, SynchList_pre, Symmetry, sync_cache_restrict[lev]);
#elif (RPB == 1) #elif (RPB == 1)
// Parallel::Restrict_bam(GH->PatL[lev-1],GH->PatL[lev],SL,SynchList_pre,Symmetry); // Parallel::Restrict_bam(GH->PatL[lev-1],GH->PatL[lev],SL,SynchList_pre,Symmetry);
Parallel::Restrict_bam(GH->PatL[lev - 1], GH->PatL[lev], SL, SynchList_pre, GH->rsul[lev], Symmetry); Parallel::Restrict_bam(GH->PatL[lev - 1], GH->PatL[lev], SL, SynchList_pre, GH->rsul[lev], Symmetry);
@@ -5820,7 +5822,7 @@ void bssn_class::RestrictProlong(int lev, int YN, bool BB,
#if (RPB == 0) #if (RPB == 0)
#if (MIXOUTB == 0) #if (MIXOUTB == 0)
Parallel::OutBdLow2Hi(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SL, Symmetry); Parallel::OutBdLow2Hi_cached(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SL, Symmetry, sync_cache_outbd[lev]);
#elif (MIXOUTB == 1) #elif (MIXOUTB == 1)
Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SL, Symmetry); Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SL, Symmetry);
#endif #endif
@@ -5847,7 +5849,7 @@ void bssn_class::RestrictProlong(int lev, int YN, bool BB,
#endif #endif
#if (RPB == 0) #if (RPB == 0)
Parallel::Restrict(GH->PatL[lev - 1], GH->PatL[lev], SL, SL, Symmetry); Parallel::Restrict_cached(GH->PatL[lev - 1], GH->PatL[lev], SL, SL, Symmetry, sync_cache_restrict[lev]);
#elif (RPB == 1) #elif (RPB == 1)
// Parallel::Restrict_bam(GH->PatL[lev-1],GH->PatL[lev],SL,SL,Symmetry); // Parallel::Restrict_bam(GH->PatL[lev-1],GH->PatL[lev],SL,SL,Symmetry);
Parallel::Restrict_bam(GH->PatL[lev - 1], GH->PatL[lev], SL, SL, GH->rsul[lev], Symmetry); Parallel::Restrict_bam(GH->PatL[lev - 1], GH->PatL[lev], SL, SL, GH->rsul[lev], Symmetry);
@@ -5871,7 +5873,7 @@ void bssn_class::RestrictProlong(int lev, int YN, bool BB,
#if (RPB == 0) #if (RPB == 0)
#if (MIXOUTB == 0) #if (MIXOUTB == 0)
Parallel::OutBdLow2Hi(GH->PatL[lev - 1], GH->PatL[lev], SL, SL, Symmetry); Parallel::OutBdLow2Hi_cached(GH->PatL[lev - 1], GH->PatL[lev], SL, SL, Symmetry, sync_cache_outbd[lev]);
#elif (MIXOUTB == 1) #elif (MIXOUTB == 1)
Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], SL, SL, Symmetry); Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], SL, SL, Symmetry);
#endif #endif
@@ -5940,7 +5942,7 @@ void bssn_class::RestrictProlong_aux(int lev, int YN, bool BB,
} }
#if (RPB == 0) #if (RPB == 0)
Parallel::Restrict(GH->PatL[lev - 1], GH->PatL[lev], SL, SynchList_pre, Symmetry); Parallel::Restrict_cached(GH->PatL[lev - 1], GH->PatL[lev], SL, SynchList_pre, Symmetry, sync_cache_restrict[lev]);
#elif (RPB == 1) #elif (RPB == 1)
// Parallel::Restrict_bam(GH->PatL[lev-1],GH->PatL[lev],SL,SynchList_pre,Symmetry); // Parallel::Restrict_bam(GH->PatL[lev-1],GH->PatL[lev],SL,SynchList_pre,Symmetry);
Parallel::Restrict_bam(GH->PatL[lev - 1], GH->PatL[lev], SL, SynchList_pre, GH->rsul[lev], Symmetry); Parallel::Restrict_bam(GH->PatL[lev - 1], GH->PatL[lev], SL, SynchList_pre, GH->rsul[lev], Symmetry);
@@ -5950,7 +5952,7 @@ void bssn_class::RestrictProlong_aux(int lev, int YN, bool BB,
#if (RPB == 0) #if (RPB == 0)
#if (MIXOUTB == 0) #if (MIXOUTB == 0)
Parallel::OutBdLow2Hi(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SL, Symmetry); Parallel::OutBdLow2Hi_cached(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SL, Symmetry, sync_cache_outbd[lev]);
#elif (MIXOUTB == 1) #elif (MIXOUTB == 1)
Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SL, Symmetry); Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SL, Symmetry);
#endif #endif
@@ -5962,7 +5964,7 @@ void bssn_class::RestrictProlong_aux(int lev, int YN, bool BB,
else // no time refinement levels and for all same time levels else // no time refinement levels and for all same time levels
{ {
#if (RPB == 0) #if (RPB == 0)
Parallel::Restrict(GH->PatL[lev - 1], GH->PatL[lev], SL, SL, Symmetry); Parallel::Restrict_cached(GH->PatL[lev - 1], GH->PatL[lev], SL, SL, Symmetry, sync_cache_restrict[lev]);
#elif (RPB == 1) #elif (RPB == 1)
// Parallel::Restrict_bam(GH->PatL[lev-1],GH->PatL[lev],SL,SL,Symmetry); // Parallel::Restrict_bam(GH->PatL[lev-1],GH->PatL[lev],SL,SL,Symmetry);
Parallel::Restrict_bam(GH->PatL[lev - 1], GH->PatL[lev], SL, SL, GH->rsul[lev], Symmetry); Parallel::Restrict_bam(GH->PatL[lev - 1], GH->PatL[lev], SL, SL, GH->rsul[lev], Symmetry);
@@ -5972,7 +5974,7 @@ void bssn_class::RestrictProlong_aux(int lev, int YN, bool BB,
#if (RPB == 0) #if (RPB == 0)
#if (MIXOUTB == 0) #if (MIXOUTB == 0)
Parallel::OutBdLow2Hi(GH->PatL[lev - 1], GH->PatL[lev], SL, SL, Symmetry); Parallel::OutBdLow2Hi_cached(GH->PatL[lev - 1], GH->PatL[lev], SL, SL, Symmetry, sync_cache_outbd[lev]);
#elif (MIXOUTB == 1) #elif (MIXOUTB == 1)
Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], SL, SL, Symmetry); Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], SL, SL, Symmetry);
#endif #endif
@@ -6027,7 +6029,7 @@ void bssn_class::RestrictProlong(int lev, int YN, bool BB)
} }
#if (RPB == 0) #if (RPB == 0)
Parallel::Restrict(GH->PatL[lev - 1], GH->PatL[lev], SynchList_cor, SynchList_pre, Symmetry); Parallel::Restrict_cached(GH->PatL[lev - 1], GH->PatL[lev], SynchList_cor, SynchList_pre, Symmetry, sync_cache_restrict[lev]);
#elif (RPB == 1) #elif (RPB == 1)
// Parallel::Restrict_bam(GH->PatL[lev-1],GH->PatL[lev],SynchList_cor,SynchList_pre,Symmetry); // Parallel::Restrict_bam(GH->PatL[lev-1],GH->PatL[lev],SynchList_cor,SynchList_pre,Symmetry);
Parallel::Restrict_bam(GH->PatL[lev - 1], GH->PatL[lev], SynchList_cor, SynchList_pre, GH->rsul[lev], Symmetry); Parallel::Restrict_bam(GH->PatL[lev - 1], GH->PatL[lev], SynchList_cor, SynchList_pre, GH->rsul[lev], Symmetry);
@@ -6037,7 +6039,7 @@ void bssn_class::RestrictProlong(int lev, int YN, bool BB)
#if (RPB == 0) #if (RPB == 0)
#if (MIXOUTB == 0) #if (MIXOUTB == 0)
Parallel::OutBdLow2Hi(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SynchList_cor, Symmetry); Parallel::OutBdLow2Hi_cached(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SynchList_cor, Symmetry, sync_cache_outbd[lev]);
#elif (MIXOUTB == 1) #elif (MIXOUTB == 1)
Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SynchList_cor, Symmetry); Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SynchList_cor, Symmetry);
#endif #endif
@@ -6051,7 +6053,7 @@ void bssn_class::RestrictProlong(int lev, int YN, bool BB)
if (myrank == 0) if (myrank == 0)
cout << "===: " << GH->Lt[lev - 1] << "," << GH->Lt[lev] + dT_lev << endl; cout << "===: " << GH->Lt[lev - 1] << "," << GH->Lt[lev] + dT_lev << endl;
#if (RPB == 0) #if (RPB == 0)
Parallel::Restrict(GH->PatL[lev - 1], GH->PatL[lev], SynchList_cor, StateList, Symmetry); Parallel::Restrict_cached(GH->PatL[lev - 1], GH->PatL[lev], SynchList_cor, StateList, Symmetry, sync_cache_restrict[lev]);
#elif (RPB == 1) #elif (RPB == 1)
// Parallel::Restrict_bam(GH->PatL[lev-1],GH->PatL[lev],SynchList_cor,StateList,Symmetry); // Parallel::Restrict_bam(GH->PatL[lev-1],GH->PatL[lev],SynchList_cor,StateList,Symmetry);
Parallel::Restrict_bam(GH->PatL[lev - 1], GH->PatL[lev], SynchList_cor, StateList, GH->rsul[lev], Symmetry); Parallel::Restrict_bam(GH->PatL[lev - 1], GH->PatL[lev], SynchList_cor, StateList, GH->rsul[lev], Symmetry);
@@ -6061,7 +6063,7 @@ void bssn_class::RestrictProlong(int lev, int YN, bool BB)
#if (RPB == 0) #if (RPB == 0)
#if (MIXOUTB == 0) #if (MIXOUTB == 0)
Parallel::OutBdLow2Hi(GH->PatL[lev - 1], GH->PatL[lev], StateList, SynchList_cor, Symmetry); Parallel::OutBdLow2Hi_cached(GH->PatL[lev - 1], GH->PatL[lev], StateList, SynchList_cor, Symmetry, sync_cache_outbd[lev]);
#elif (MIXOUTB == 1) #elif (MIXOUTB == 1)
Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], StateList, SynchList_cor, Symmetry); Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], StateList, SynchList_cor, Symmetry);
#endif #endif
@@ -6102,7 +6104,7 @@ void bssn_class::ProlongRestrict(int lev, int YN, bool BB)
#if (RPB == 0) #if (RPB == 0)
#if (MIXOUTB == 0) #if (MIXOUTB == 0)
Parallel::OutBdLow2Hi(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SynchList_cor, Symmetry); Parallel::OutBdLow2Hi_cached(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SynchList_cor, Symmetry, sync_cache_outbd[lev]);
#elif (MIXOUTB == 1) #elif (MIXOUTB == 1)
Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SynchList_cor, Symmetry); Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SynchList_cor, Symmetry);
#endif #endif
@@ -6115,7 +6117,7 @@ void bssn_class::ProlongRestrict(int lev, int YN, bool BB)
{ {
#if (RPB == 0) #if (RPB == 0)
#if (MIXOUTB == 0) #if (MIXOUTB == 0)
Parallel::OutBdLow2Hi(GH->PatL[lev - 1], GH->PatL[lev], StateList, SynchList_cor, Symmetry); Parallel::OutBdLow2Hi_cached(GH->PatL[lev - 1], GH->PatL[lev], StateList, SynchList_cor, Symmetry, sync_cache_outbd[lev]);
#elif (MIXOUTB == 1) #elif (MIXOUTB == 1)
Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], StateList, SynchList_cor, Symmetry); Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], StateList, SynchList_cor, Symmetry);
#endif #endif

2
AMSS_NCKU_source/bssn_class.h
View File

@@ -130,6 +130,8 @@ public:
Parallel::SyncCache *sync_cache_cor; // per-level cache for corrector 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_coarse; // RestrictProlong sync on PatL[lev-1]
Parallel::SyncCache *sync_cache_rp_fine; // RestrictProlong sync on PatL[lev] Parallel::SyncCache *sync_cache_rp_fine; // RestrictProlong sync on PatL[lev]
Parallel::SyncCache *sync_cache_restrict; // cached Restrict in RestrictProlong
Parallel::SyncCache *sync_cache_outbd; // cached OutBdLow2Hi in RestrictProlong
monitor *ErrorMonitor, *Psi4Monitor, *BHMonitor, *MAPMonitor; monitor *ErrorMonitor, *Psi4Monitor, *BHMonitor, *MAPMonitor;
monitor *ConVMonitor; monitor *ConVMonitor;