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

Add safe BSSN-EScalar kernel and transfer toggles

Browse Source
This commit is contained in:
CGH0S7
2026-04-25 01:41:55 +08:00
parent 0f1d0de1e7
commit 0cf58176d9
6 changed files with 379 additions and 413 deletions
Download Patch File Download Diff File Expand all files Collapse all files

606
AMSS_NCKU_source/bssn_class.C
View File

@@ -283,7 +283,7 @@ namespace rhs_kernel_timing_report
bssn_class::bssn_class(double Couranti, double StartTimei, double TotalTimei,
double DumpTimei, double d2DumpTimei, double CheckTimei, double AnasTimei,
int Symmetryi, int checkruni, char *checkfilenamei,
double numepssi, double numepsbi, double numepshi,
double numepssi, double numepsbi, double numepshi,
int a_levi, int maxli, int decni, double maxrexi, double drexi)
: Courant(Couranti), StartTime(StartTimei), TotalTime(TotalTimei),
DumpTime(DumpTimei), d2DumpTime(d2DumpTimei), CheckTime(CheckTimei), AnasTime(AnasTimei),
@@ -1008,21 +1008,7 @@ void bssn_class::Initialize()
Setup_Black_Hole_position();
}
// BSSN-EScalar uses the uncached communication fallback paths.
sync_cache_pre = 0;
sync_cache_cor = 0;
sync_cache_rp_coarse = 0;
sync_cache_rp_fine = 0;
sync_cache_restrict = 0;
sync_cache_outbd = 0;
#if (ABEtype != 1)
sync_cache_pre = 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_fine = new Parallel::SyncCache[GH->levels];
sync_cache_restrict = new Parallel::SyncCache[GH->levels];
sync_cache_outbd = new Parallel::SyncCache[GH->levels];
#endif
setup_transfer_caches();
}
//================================================================================================
@@ -1037,15 +1023,8 @@ void bssn_class::Initialize()
bssn_class::~bssn_class()
{
#if (ABEtype == 1)
if (myrank == 0)
{
cout << "[dtor] begin" << endl;
cout.flush();
}
#endif
#ifdef With_AHF
AHList->clearList();
AHList->clearList();
AHDList->clearList();
GaugeList->clearList();
if (lastahdumpid)
@@ -1078,13 +1057,6 @@ bssn_class::~bssn_class()
ConstraintList->clearList();
delete[] ConstraintRefreshLevels;
#if (ABEtype == 1)
if (myrank == 0)
{
cout << "[dtor] lists cleared" << endl;
cout.flush();
}
#endif
delete phio;
delete trKo;
@@ -1257,20 +1229,13 @@ bssn_class::~bssn_class()
delete Cons_Ham;
delete Cons_Px;
delete Cons_Py;
delete Cons_Pz;
delete Cons_Gx;
delete Cons_Py;
delete Cons_Pz;
delete Cons_Gx;
delete Cons_Gy;
delete Cons_Gz;
#if (ABEtype == 1)
if (myrank == 0)
{
cout << "[dtor] core vars freed" << endl;
cout.flush();
}
#endif
#ifdef Point_Psi4
#ifdef Point_Psi4
delete phix;
delete phiy;
delete phiz;
@@ -1296,78 +1261,17 @@ bssn_class::~bssn_class()
delete Azzy;
delete Azzz;
#endif
// Destroy sync caches before GH
if (sync_cache_pre)
{
#if (ABEtype != 1)
for (int i = 0; i < GH->levels; i++)
sync_cache_pre[i].destroy();
#endif
delete[] sync_cache_pre;
}
if (sync_cache_cor)
{
#if (ABEtype != 1)
for (int i = 0; i < GH->levels; i++)
sync_cache_cor[i].destroy();
#endif
delete[] sync_cache_cor;
}
if (sync_cache_rp_coarse)
{
#if (ABEtype != 1)
for (int i = 0; i < GH->levels; i++)
sync_cache_rp_coarse[i].destroy();
#endif
delete[] sync_cache_rp_coarse;
}
if (sync_cache_rp_fine)
{
#if (ABEtype != 1)
for (int i = 0; i < GH->levels; i++)
sync_cache_rp_fine[i].destroy();
#endif
delete[] sync_cache_rp_fine;
}
if (sync_cache_restrict)
{
#if (ABEtype != 1)
for (int i = 0; i < GH->levels; i++)
sync_cache_restrict[i].destroy();
#endif
delete[] sync_cache_restrict;
}
if (sync_cache_outbd)
{
#if (ABEtype != 1)
for (int i = 0; i < GH->levels; i++)
sync_cache_outbd[i].destroy();
#endif
delete[] sync_cache_outbd;
}
#if (ABEtype == 1)
if (myrank == 0)
{
cout << "[dtor] caches freed" << endl;
cout.flush();
}
#endif
// Destroy sync caches before GH
destroy_transfer_caches();
delete GH;
#ifdef WithShell
delete SH;
#endif
#if (ABEtype == 1)
if (myrank == 0)
for (int i = 0; i < BH_num; i++)
{
cout << "[dtor] grids freed" << endl;
cout.flush();
}
#endif
for (int i = 0; i < BH_num; i++)
{
delete[] Porg0[i];
delete[] Porgbr[i];
delete[] Porg[i];
@@ -1380,17 +1284,10 @@ bssn_class::~bssn_class()
delete[] Porg;
delete[] Porg1;
delete[] Porg_rhs;
delete[] Mass;
delete[] Spin;
delete[] Pmom;
#if (ABEtype == 1)
if (myrank == 0)
{
cout << "[dtor] puncture arrays freed" << endl;
cout.flush();
}
#endif
delete ErrorMonitor;
delete Psi4Monitor;
@@ -1399,22 +1296,8 @@ bssn_class::~bssn_class()
delete ConVMonitor;
delete TimingMonitor;
delete Waveshell;
#if (ABEtype == 1)
if (myrank == 0)
{
cout << "[dtor] monitors freed" << endl;
cout.flush();
}
#endif
delete CheckPoint;
#if (ABEtype == 1)
if (myrank == 0)
{
cout << "[dtor] checkpoint freed" << endl;
cout.flush();
}
#endif
}
//================================================================================================
@@ -2599,9 +2482,7 @@ void bssn_class::Evolve(int Steps)
GH->Regrid(Symmetry, BH_num, Porgbr, Porg0,
SynchList_cor, OldStateList, StateList, SynchList_pre,
fgt(PhysTime - dT_mon, StartTime, dT_mon / 2), ErrorMonitor);
#if (ABEtype != 1)
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
invalidate_transfer_caches();
STEP_TIMER_ADD(TB_REGRID, timer_regrid);
#endif
@@ -2842,9 +2723,7 @@ void bssn_class::RecursiveStep(int lev)
{
if (ConstraintRefreshLevels)
ConstraintRefreshLevels[lev] = 1;
#if (ABEtype != 1)
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
invalidate_transfer_caches();
}
STEP_TIMER_ADD(TB_REGRID, timer_regrid_onelevel);
#endif
@@ -3022,13 +2901,11 @@ void bssn_class::ParallelStep()
delete[] tporg;
delete[] tporgo;
#if (REGLEV == 0)
if (GH->Regrid_Onelevel(GH->mylev, Symmetry, BH_num, Porgbr, Porg0,
SynchList_cor, OldStateList, StateList, SynchList_pre,
fgt(PhysTime - dT_lev, StartTime, dT_lev / 2), ErrorMonitor))
#if (ABEtype != 1)
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(); }
if (GH->Regrid_Onelevel(GH->mylev, Symmetry, BH_num, Porgbr, Porg0,
SynchList_cor, OldStateList, StateList, SynchList_pre,
fgt(PhysTime - dT_lev, StartTime, dT_lev / 2), ErrorMonitor))
invalidate_transfer_caches();
#endif
#endif
}
//================================================================================================
@@ -3191,12 +3068,10 @@ void bssn_class::ParallelStep()
if (lev + 1 >= GH->movls)
{
// GH->Regrid_Onelevel_aux(lev,Symmetry,BH_num,Porgbr,Porg0,
if (GH->Regrid_Onelevel(lev + 1, Symmetry, BH_num, Porgbr, Porg0,
SynchList_cor, OldStateList, StateList, SynchList_pre,
fgt(PhysTime - dT_levp1, StartTime, dT_levp1 / 2), ErrorMonitor))
#if (ABEtype != 1)
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
if (GH->Regrid_Onelevel(lev + 1, Symmetry, BH_num, Porgbr, Porg0,
SynchList_cor, OldStateList, StateList, SynchList_pre,
fgt(PhysTime - dT_levp1, StartTime, dT_levp1 / 2), ErrorMonitor))
invalidate_transfer_caches();
// a_stream.clear();
// a_stream.str("");
@@ -3208,12 +3083,10 @@ void bssn_class::ParallelStep()
// for this level
if (YN == 1)
{
if (GH->Regrid_Onelevel(lev, Symmetry, BH_num, Porgbr, Porg0,
SynchList_cor, OldStateList, StateList, SynchList_pre,
fgt(PhysTime - dT_lev, StartTime, dT_lev / 2), ErrorMonitor))
#if (ABEtype != 1)
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
if (GH->Regrid_Onelevel(lev, Symmetry, BH_num, Porgbr, Porg0,
SynchList_cor, OldStateList, StateList, SynchList_pre,
fgt(PhysTime - dT_lev, StartTime, dT_lev / 2), ErrorMonitor))
invalidate_transfer_caches();
// a_stream.clear();
// a_stream.str("");
@@ -3229,12 +3102,10 @@ void bssn_class::ParallelStep()
if (YN == 1)
{
// GH->Regrid_Onelevel_aux(lev-2,Symmetry,BH_num,Porgbr,Porg0,
if (GH->Regrid_Onelevel(lev - 1, Symmetry, BH_num, Porgbr, Porg0,
SynchList_cor, OldStateList, StateList, SynchList_pre,
fgt(PhysTime - dT_lev, StartTime, dT_levm1 / 2), ErrorMonitor))
#if (ABEtype != 1)
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
if (GH->Regrid_Onelevel(lev - 1, Symmetry, BH_num, Porgbr, Porg0,
SynchList_cor, OldStateList, StateList, SynchList_pre,
fgt(PhysTime - dT_lev, StartTime, dT_levm1 / 2), ErrorMonitor))
invalidate_transfer_caches();
// a_stream.clear();
// a_stream.str("");
@@ -3247,12 +3118,10 @@ void bssn_class::ParallelStep()
if (i % 4 == 3)
{
// GH->Regrid_Onelevel_aux(lev-2,Symmetry,BH_num,Porgbr,Porg0,
if (GH->Regrid_Onelevel(lev - 1, Symmetry, BH_num, Porgbr, Porg0,
SynchList_cor, OldStateList, StateList, SynchList_pre,
fgt(PhysTime - dT_lev, StartTime, dT_levm1 / 2), ErrorMonitor))
#if (ABEtype != 1)
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
if (GH->Regrid_Onelevel(lev - 1, Symmetry, BH_num, Porgbr, Porg0,
SynchList_cor, OldStateList, StateList, SynchList_pre,
fgt(PhysTime - dT_lev, StartTime, dT_levm1 / 2), ErrorMonitor))
invalidate_transfer_caches();
// a_stream.clear();
// a_stream.str("");
@@ -3783,11 +3652,7 @@ void bssn_class::Step(int lev, int YN)
STEP_TIMER_DECL(timer_predictor_sync);
Parallel::AsyncSyncState async_pre;
#if (ABEtype == 1)
Parallel::Sync(GH->PatL[lev], SynchList_pre, Symmetry);
#else
Parallel::Sync_start(GH->PatL[lev], SynchList_pre, Symmetry, sync_cache_pre[lev], async_pre);
#endif
sync_predictor_start(lev, SynchList_pre, async_pre);
#ifdef WithShell
if (lev == 0)
@@ -3806,9 +3671,7 @@ void bssn_class::Step(int lev, int YN)
}
}
#endif
#if (ABEtype != 1)
Parallel::Sync_finish(sync_cache_pre[lev], async_pre, SynchList_pre, Symmetry);
#endif
sync_predictor_finish(lev, async_pre, SynchList_pre);
#ifdef WithShell
// Complete non-blocking error reduction and check
@@ -4154,11 +4017,7 @@ void bssn_class::Step(int lev, int YN)
STEP_TIMER_DECL(timer_corrector_sync);
Parallel::AsyncSyncState async_cor;
#if (ABEtype == 1)
Parallel::Sync(GH->PatL[lev], SynchList_cor, Symmetry);
#else
Parallel::Sync_start(GH->PatL[lev], SynchList_cor, Symmetry, sync_cache_cor[lev], async_cor);
#endif
sync_corrector_start(lev, SynchList_cor, async_cor);
#ifdef WithShell
if (lev == 0)
@@ -4177,9 +4036,7 @@ void bssn_class::Step(int lev, int YN)
}
}
#endif
#if (ABEtype != 1)
Parallel::Sync_finish(sync_cache_cor[lev], async_cor, SynchList_cor, Symmetry);
#endif
sync_corrector_finish(lev, async_cor, SynchList_cor);
#ifdef WithShell
// Complete non-blocking error reduction and check
@@ -4664,15 +4521,11 @@ void bssn_class::Step(int lev, int YN)
{
int erh = ERROR;
MPI_Iallreduce(&erh, &ERROR, 1, MPI_INT, MPI_SUM, MPI_COMM_WORLD, &err_req);
}
#endif
Parallel::AsyncSyncState async_pre;
#if (ABEtype == 1)
Parallel::Sync(GH->PatL[lev], SynchList_pre, Symmetry);
#else
Parallel::Sync_start(GH->PatL[lev], SynchList_pre, Symmetry, sync_cache_pre[lev], async_pre);
}
#endif
Parallel::AsyncSyncState async_pre;
sync_predictor_start(lev, SynchList_pre, async_pre);
#ifdef WithShell
if (lev == 0)
@@ -4691,9 +4544,7 @@ void bssn_class::Step(int lev, int YN)
}
}
#endif
#if (ABEtype != 1)
Parallel::Sync_finish(sync_cache_pre[lev], async_pre, SynchList_pre, Symmetry);
#endif
sync_predictor_finish(lev, async_pre, SynchList_pre);
#ifdef WithShell
// Complete non-blocking error reduction and check
@@ -5021,12 +4872,8 @@ void bssn_class::Step(int lev, int YN)
}
#endif
Parallel::AsyncSyncState async_cor;
#if (ABEtype == 1)
Parallel::Sync(GH->PatL[lev], SynchList_cor, Symmetry);
#else
Parallel::Sync_start(GH->PatL[lev], SynchList_cor, Symmetry, sync_cache_cor[lev], async_cor);
#endif
Parallel::AsyncSyncState async_cor;
sync_corrector_start(lev, SynchList_cor, async_cor);
#ifdef WithShell
if (lev == 0)
@@ -5045,9 +4892,7 @@ void bssn_class::Step(int lev, int YN)
}
}
#endif
#if (ABEtype != 1)
Parallel::Sync_finish(sync_cache_cor[lev], async_cor, SynchList_cor, Symmetry);
#endif
sync_corrector_finish(lev, async_cor, SynchList_cor);
#ifdef WithShell
// Complete non-blocking error reduction and check
@@ -5439,11 +5284,7 @@ void bssn_class::Step(int lev, int YN)
// misc::tillherecheck(GH->Commlev[lev],GH->start_rank[lev],"before Predictor sync");
#if (ABEtype == 1)
Parallel::Sync(GH->PatL[lev], SynchList_pre, Symmetry);
#else
Parallel::Sync_cached(GH->PatL[lev], SynchList_pre, Symmetry, sync_cache_pre[lev]);
#endif
sync_evolution(lev, SynchList_pre, sync_cache_pre);
// Complete non-blocking error reduction and check
MPI_Wait(&err_req, MPI_STATUS_IGNORE);
@@ -5644,11 +5485,7 @@ void bssn_class::Step(int lev, int YN)
// misc::tillherecheck(GH->Commlev[lev],GH->start_rank[lev],"before Corrector sync");
#if (ABEtype == 1)
Parallel::Sync(GH->PatL[lev], SynchList_cor, Symmetry);
#else
Parallel::Sync_cached(GH->PatL[lev], SynchList_cor, Symmetry, sync_cache_cor[lev]);
#endif
sync_evolution(lev, SynchList_cor, sync_cache_cor);
// misc::tillherecheck(GH->Commlev[lev],GH->start_rank[lev],"after Corrector sync");
@@ -6365,15 +6202,11 @@ void bssn_class::RestrictProlong(int lev, int YN, bool BB,
#endif
#if (RPB == 0)
#if (ABEtype == 1)
Parallel::Restrict(GH->PatL[lev - 1], GH->PatL[lev], SL, SynchList_pre, Symmetry);
#else
Parallel::Restrict_cached(GH->PatL[lev - 1], GH->PatL[lev], SL, SynchList_pre, Symmetry, sync_cache_restrict[lev]);
#endif
restrict_evolution(lev, SL, SynchList_pre);
#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, GH->rsul[lev], Symmetry);
#endif
// 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);
#endif
#if (PSTR == 1 || PSTR == 2)
// a_stream.clear();
@@ -6382,11 +6215,7 @@ void bssn_class::RestrictProlong(int lev, int YN, bool BB,
// misc::tillherecheck(GH->Commlev[GH->mylev],GH->start_rank[GH->mylev],a_stream.str());
#endif
#if (ABEtype == 1)
Parallel::Sync(GH->PatL[lev - 1], SynchList_pre, Symmetry);
#else
Parallel::Sync_cached(GH->PatL[lev - 1], SynchList_pre, Symmetry, sync_cache_rp_coarse[lev]);
#endif
sync_evolution(lev - 1, SynchList_pre, sync_cache_rp_coarse);
#if (PSTR == 1 || PSTR == 2)
// a_stream.clear();
@@ -6397,21 +6226,7 @@ void bssn_class::RestrictProlong(int lev, int YN, bool BB,
#if (RPB == 0)
#if (MIXOUTB == 0)
#if (ABEtype == 1)
Ppc = GH->PatL[lev - 1];
while (Ppc)
{
Pp = GH->PatL[lev];
while (Pp)
{
Parallel::OutBdLow2Hi(Ppc->data, Pp->data, SynchList_pre, SL, Symmetry);
Pp = Pp->next;
}
Ppc = Ppc->next;
}
#else
Parallel::OutBdLow2Hi_cached(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SL, Symmetry, sync_cache_outbd[lev]);
#endif
outbdlow2hi_evolution(lev, SynchList_pre, SL);
#elif (MIXOUTB == 1)
Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SL, Symmetry);
#endif
@@ -6438,15 +6253,11 @@ void bssn_class::RestrictProlong(int lev, int YN, bool BB,
#endif
#if (RPB == 0)
#if (ABEtype == 1)
Parallel::Restrict(GH->PatL[lev - 1], GH->PatL[lev], SL, SL, Symmetry);
#else
Parallel::Restrict_cached(GH->PatL[lev - 1], GH->PatL[lev], SL, SL, Symmetry, sync_cache_restrict[lev]);
#endif
restrict_evolution(lev, SL, SL);
#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, GH->rsul[lev], Symmetry);
#endif
// 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);
#endif
#if (PSTR == 1 || PSTR == 2)
// a_stream.clear();
@@ -6455,11 +6266,7 @@ void bssn_class::RestrictProlong(int lev, int YN, bool BB,
// misc::tillherecheck(GH->Commlev[GH->mylev],GH->start_rank[GH->mylev],a_stream.str());
#endif
#if (ABEtype == 1)
Parallel::Sync(GH->PatL[lev - 1], SL, Symmetry);
#else
Parallel::Sync_cached(GH->PatL[lev - 1], SL, Symmetry, sync_cache_rp_coarse[lev]);
#endif
sync_evolution(lev - 1, SL, sync_cache_rp_coarse);
#if (PSTR == 1 || PSTR == 2)
// a_stream.clear();
@@ -6470,21 +6277,7 @@ void bssn_class::RestrictProlong(int lev, int YN, bool BB,
#if (RPB == 0)
#if (MIXOUTB == 0)
#if (ABEtype == 1)
Ppc = GH->PatL[lev - 1];
while (Ppc)
{
Pp = GH->PatL[lev];
while (Pp)
{
Parallel::OutBdLow2Hi(Ppc->data, Pp->data, SL, SL, Symmetry);
Pp = Pp->next;
}
Ppc = Ppc->next;
}
#else
Parallel::OutBdLow2Hi_cached(GH->PatL[lev - 1], GH->PatL[lev], SL, SL, Symmetry, sync_cache_outbd[lev]);
#endif
outbdlow2hi_evolution(lev, SL, SL);
#elif (MIXOUTB == 1)
Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], SL, SL, Symmetry);
#endif
@@ -6685,7 +6478,7 @@ void bssn_class::RestrictProlong_aux(int lev, int YN, bool BB,
#if (ABEtype == 1)
Parallel::Sync(GH->PatL[lev], SL, Symmetry);
#else
Parallel::Sync_cached(GH->PatL[lev], SL, Symmetry, sync_cache_rp_fine[lev]);
sync_evolution(lev, SL, sync_cache_rp_fine);
#endif
}
STEP_TIMER_ADD(TB_RESTRICT_PROLONG, timer_restrict_prolong);
@@ -6818,39 +6611,17 @@ void bssn_class::RestrictProlong(int lev, int YN, bool BB)
}
#if (RPB == 0)
#if (ABEtype == 1)
Parallel::Restrict(GH->PatL[lev - 1], GH->PatL[lev], SynchList_cor, SynchList_pre, Symmetry);
#else
Parallel::Restrict_cached(GH->PatL[lev - 1], GH->PatL[lev], SynchList_cor, SynchList_pre, Symmetry, sync_cache_restrict[lev]);
#endif
restrict_evolution(lev, SynchList_cor, SynchList_pre);
#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, GH->rsul[lev], Symmetry);
#endif
#if (ABEtype == 1)
Parallel::Sync(GH->PatL[lev - 1], SynchList_pre, Symmetry);
#else
Parallel::Sync_cached(GH->PatL[lev - 1], SynchList_pre, Symmetry, sync_cache_rp_coarse[lev]);
// 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);
#endif
sync_evolution(lev - 1, SynchList_pre, sync_cache_rp_coarse);
#if (RPB == 0)
#if (MIXOUTB == 0)
#if (ABEtype == 1)
Ppc = GH->PatL[lev - 1];
while (Ppc)
{
Pp = GH->PatL[lev];
while (Pp)
{
Parallel::OutBdLow2Hi(Ppc->data, Pp->data, SynchList_pre, SynchList_cor, Symmetry);
Pp = Pp->next;
}
Ppc = Ppc->next;
}
#else
Parallel::OutBdLow2Hi_cached(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SynchList_cor, Symmetry, sync_cache_outbd[lev]);
#endif
outbdlow2hi_evolution(lev, SynchList_pre, SynchList_cor);
#elif (MIXOUTB == 1)
Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SynchList_cor, Symmetry);
#endif
@@ -6864,39 +6635,17 @@ void bssn_class::RestrictProlong(int lev, int YN, bool BB)
if (myrank == 0)
cout << "===: " << GH->Lt[lev - 1] << "," << GH->Lt[lev] + dT_lev << endl;
#if (RPB == 0)
#if (ABEtype == 1)
Parallel::Restrict(GH->PatL[lev - 1], GH->PatL[lev], SynchList_cor, StateList, Symmetry);
#else
Parallel::Restrict_cached(GH->PatL[lev - 1], GH->PatL[lev], SynchList_cor, StateList, Symmetry, sync_cache_restrict[lev]);
#endif
restrict_evolution(lev, SynchList_cor, StateList);
#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, GH->rsul[lev], Symmetry);
#endif
#if (ABEtype == 1)
Parallel::Sync(GH->PatL[lev - 1], StateList, Symmetry);
#else
Parallel::Sync_cached(GH->PatL[lev - 1], StateList, Symmetry, sync_cache_rp_coarse[lev]);
// 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);
#endif
sync_evolution(lev - 1, StateList, sync_cache_rp_coarse);
#if (RPB == 0)
#if (MIXOUTB == 0)
#if (ABEtype == 1)
Ppc = GH->PatL[lev - 1];
while (Ppc)
{
Pp = GH->PatL[lev];
while (Pp)
{
Parallel::OutBdLow2Hi(Ppc->data, Pp->data, StateList, SynchList_cor, Symmetry);
Pp = Pp->next;
}
Ppc = Ppc->next;
}
#else
Parallel::OutBdLow2Hi_cached(GH->PatL[lev - 1], GH->PatL[lev], StateList, SynchList_cor, Symmetry, sync_cache_outbd[lev]);
#endif
outbdlow2hi_evolution(lev, StateList, SynchList_cor);
#elif (MIXOUTB == 1)
Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], StateList, SynchList_cor, Symmetry);
#endif
@@ -6906,11 +6655,7 @@ void bssn_class::RestrictProlong(int lev, int YN, bool BB)
#endif
}
#if (ABEtype == 1)
Parallel::Sync(GH->PatL[lev], SynchList_cor, Symmetry);
#else
Parallel::Sync_cached(GH->PatL[lev], SynchList_cor, Symmetry, sync_cache_rp_fine[lev]);
#endif
sync_evolution(lev, SynchList_cor, sync_cache_rp_fine);
}
STEP_TIMER_ADD(TB_RESTRICT_PROLONG, timer_restrict_prolong);
}
@@ -6940,12 +6685,12 @@ void bssn_class::ProlongRestrict(int lev, int YN, bool BB)
Pp = Pp->next;
}
#if (RPB == 0)
#if (MIXOUTB == 0)
Parallel::OutBdLow2Hi_cached(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SynchList_cor, Symmetry, sync_cache_outbd[lev]);
#elif (MIXOUTB == 1)
Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SynchList_cor, Symmetry);
#endif
#if (RPB == 0)
#if (MIXOUTB == 0)
outbdlow2hi_evolution(lev, SynchList_pre, SynchList_cor);
#elif (MIXOUTB == 1)
Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SynchList_cor, Symmetry);
#endif
#elif (RPB == 1)
// Parallel::OutBdLow2Hi_bam(GH->PatL[lev-1],GH->PatL[lev],SynchList_pre,SynchList_cor,Symmetry);
Parallel::OutBdLow2Hi_bam(GH->PatL[lev - 1], GH->PatL[lev], SynchList_pre, SynchList_cor, GH->bdsul[lev], Symmetry);
@@ -6953,12 +6698,12 @@ void bssn_class::ProlongRestrict(int lev, int YN, bool BB)
}
else // no time refinement levels and for all same time levels
{
#if (RPB == 0)
#if (MIXOUTB == 0)
Parallel::OutBdLow2Hi_cached(GH->PatL[lev - 1], GH->PatL[lev], StateList, SynchList_cor, Symmetry, sync_cache_outbd[lev]);
#elif (MIXOUTB == 1)
Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], StateList, SynchList_cor, Symmetry);
#endif
#if (RPB == 0)
#if (MIXOUTB == 0)
outbdlow2hi_evolution(lev, StateList, SynchList_cor);
#elif (MIXOUTB == 1)
Parallel::OutBdLow2Himix(GH->PatL[lev - 1], GH->PatL[lev], StateList, SynchList_cor, Symmetry);
#endif
#elif (RPB == 1)
// Parallel::OutBdLow2Hi_bam(GH->PatL[lev-1],GH->PatL[lev],StateList,SynchList_cor,Symmetry);
Parallel::OutBdLow2Hi_bam(GH->PatL[lev - 1], GH->PatL[lev], StateList, SynchList_cor, GH->bdsul[lev], Symmetry);
@@ -6974,12 +6719,12 @@ void bssn_class::ProlongRestrict(int lev, int YN, bool BB)
#else
Parallel::Restrict_after(GH->PatL[lev - 1], GH->PatL[lev], SynchList_cor, StateList, Symmetry);
#endif
Parallel::Sync_cached(GH->PatL[lev - 1], StateList, Symmetry, sync_cache_rp_coarse[lev]);
}
Parallel::Sync_cached(GH->PatL[lev], SynchList_cor, Symmetry, sync_cache_rp_fine[lev]);
}
}
sync_evolution(lev - 1, StateList, sync_cache_rp_coarse);
}
sync_evolution(lev, SynchList_cor, sync_cache_rp_fine);
}
}
#undef MIXOUTB
//================================================================================================
@@ -7707,6 +7452,169 @@ void bssn_class::compute_Porg_rhs(double **BH_PS, double **BH_RHS, var *forx, va
}
}
}
bool bssn_class::use_transfer_cache() const
{
#if BSSN_USE_TRANSFER_CACHE
return true;
#else
return false;
#endif
}
void bssn_class::setup_transfer_caches()
{
sync_cache_pre = 0;
sync_cache_cor = 0;
sync_cache_rp_coarse = 0;
sync_cache_rp_fine = 0;
sync_cache_restrict = 0;
sync_cache_outbd = 0;
if (!use_transfer_cache() || !GH)
return;
sync_cache_pre = 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_fine = new Parallel::SyncCache[GH->levels];
sync_cache_restrict = new Parallel::SyncCache[GH->levels];
sync_cache_outbd = new Parallel::SyncCache[GH->levels];
}
void bssn_class::invalidate_transfer_caches()
{
if (!use_transfer_cache() || !GH || !sync_cache_pre || !sync_cache_cor ||
!sync_cache_rp_coarse || !sync_cache_rp_fine || !sync_cache_restrict || !sync_cache_outbd)
return;
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();
}
}
void bssn_class::destroy_transfer_caches()
{
if (sync_cache_pre)
{
if (use_transfer_cache() && GH)
for (int i = 0; i < GH->levels; i++)
sync_cache_pre[i].destroy();
delete[] sync_cache_pre;
sync_cache_pre = 0;
}
if (sync_cache_cor)
{
if (use_transfer_cache() && GH)
for (int i = 0; i < GH->levels; i++)
sync_cache_cor[i].destroy();
delete[] sync_cache_cor;
sync_cache_cor = 0;
}
if (sync_cache_rp_coarse)
{
if (use_transfer_cache() && GH)
for (int i = 0; i < GH->levels; i++)
sync_cache_rp_coarse[i].destroy();
delete[] sync_cache_rp_coarse;
sync_cache_rp_coarse = 0;
}
if (sync_cache_rp_fine)
{
if (use_transfer_cache() && GH)
for (int i = 0; i < GH->levels; i++)
sync_cache_rp_fine[i].destroy();
delete[] sync_cache_rp_fine;
sync_cache_rp_fine = 0;
}
if (sync_cache_restrict)
{
if (use_transfer_cache() && GH)
for (int i = 0; i < GH->levels; i++)
sync_cache_restrict[i].destroy();
delete[] sync_cache_restrict;
sync_cache_restrict = 0;
}
if (sync_cache_outbd)
{
if (use_transfer_cache() && GH)
for (int i = 0; i < GH->levels; i++)
sync_cache_outbd[i].destroy();
delete[] sync_cache_outbd;
sync_cache_outbd = 0;
}
}
void bssn_class::sync_predictor_start(int lev, MyList<var> *VarList, Parallel::AsyncSyncState &async_state)
{
if (use_transfer_cache())
Parallel::Sync_start(GH->PatL[lev], VarList, Symmetry, sync_cache_pre[lev], async_state);
else
Parallel::Sync(GH->PatL[lev], VarList, Symmetry);
}
void bssn_class::sync_predictor_finish(int lev, Parallel::AsyncSyncState &async_state, MyList<var> *VarList)
{
if (use_transfer_cache())
Parallel::Sync_finish(sync_cache_pre[lev], async_state, VarList, Symmetry);
}
void bssn_class::sync_corrector_start(int lev, MyList<var> *VarList, Parallel::AsyncSyncState &async_state)
{
if (use_transfer_cache())
Parallel::Sync_start(GH->PatL[lev], VarList, Symmetry, sync_cache_cor[lev], async_state);
else
Parallel::Sync(GH->PatL[lev], VarList, Symmetry);
}
void bssn_class::sync_corrector_finish(int lev, Parallel::AsyncSyncState &async_state, MyList<var> *VarList)
{
if (use_transfer_cache())
Parallel::Sync_finish(sync_cache_cor[lev], async_state, VarList, Symmetry);
}
void bssn_class::sync_evolution(int lev, MyList<var> *VarList, Parallel::SyncCache *cache_array)
{
if (use_transfer_cache() && cache_array)
Parallel::Sync_cached(GH->PatL[lev], VarList, Symmetry, cache_array[lev]);
else
Parallel::Sync(GH->PatL[lev], VarList, Symmetry);
}
void bssn_class::restrict_evolution(int lev, MyList<var> *src_var_list, MyList<var> *dst_var_list)
{
if (use_transfer_cache())
Parallel::Restrict_cached(GH->PatL[lev - 1], GH->PatL[lev], src_var_list, dst_var_list, Symmetry, sync_cache_restrict[lev]);
else
Parallel::Restrict(GH->PatL[lev - 1], GH->PatL[lev], src_var_list, dst_var_list, Symmetry);
}
void bssn_class::outbdlow2hi_evolution(int lev, MyList<var> *src_var_list, MyList<var> *dst_var_list)
{
if (use_transfer_cache())
{
Parallel::OutBdLow2Hi_cached(GH->PatL[lev - 1], GH->PatL[lev], src_var_list, dst_var_list, Symmetry, sync_cache_outbd[lev]);
return;
}
MyList<Patch> *Ppc = GH->PatL[lev - 1];
while (Ppc)
{
MyList<Patch> *Pp = GH->PatL[lev];
while (Pp)
{
Parallel::OutBdLow2Hi(Ppc->data, Pp->data, src_var_list, dst_var_list, Symmetry);
Pp = Pp->next;
}
Ppc = Ppc->next;
}
}
#endif
//================================================================================================