修改block划分，对负载高的rank所在block进行划分，添加到空rank，空rank是平移得到的

Rank [  20]: Calc 0.000003 s

Rank [  35]: Calc 0.000003 s

Rank [  10]: Calc 0.000010 s

Rank [  17]: Calc 0.000005 s

Rank [  32]: Calc 0.000003 s，而且rank不固定，一般就是rank 10 和 rank 52；
但尽管有很多，比前者时间还是少很多
对于第二个Surf_Wave 调用 Interp_points，我发现前四个rank时间最长，比较固定，就是下面四个rank

Rank [  27]: Calc 0.331978 s

Rank [  35]: Calc 0.242219 s

Rank [  28]: Calc 0.242132 s

Rank [  36]: Calc 0.197024 s
因此下面surf_wave是核心

AMSS_NCKU_Program.py

@@ -66,7 +66,8 @@ if os.path.exists(File_directory):
     ## Prompt whether to overwrite the existing directory
     while True:
         try:
-            inputvalue = input()
+            ## inputvalue = input()
+            inputvalue = "continue"
             ## If the user agrees to overwrite, proceed and remove the existing directory
             if ( inputvalue == "continue" ):
                 print( " Continue the calculation !!! " )

AMSS_NCKU_source/NullShellPatch.h

@@ -24,6 +24,7 @@ using namespace std;
 #endif
 
 #include <mpi.h>
+#include <memory.h>
 #include "MyList.h"
 #include "Block.h"
 #include "Parallel.h"

AMSS_NCKU_source/Parallel.h

@@ -1,213 +1,235 @@
-
-#ifndef PARALLEL_H
-#define PARALLEL_H
-
-#include <iostream>
-#include <iomanip>
-#include <fstream>
-#include <cstdlib>
-#include <cstdio>
-#include <string>
-#include <cmath>
-#include <new>
-using namespace std;
-
-#include "Parallel_bam.h"
-#include "var.h"
-#include "MPatch.h"
-#include "Block.h"
-#include "MyList.h"
-#include "macrodef.h" //need dim; ghost_width; CONTRACT
-namespace Parallel
-{
-  struct gridseg
-  {
-    double llb[dim];
-    double uub[dim];
-    int shape[dim];
-    double illb[dim], iuub[dim]; // only use for OutBdLow2Hi
-    Block *Bg;
-  };
-  int partition1(int &nx, int split_size, int min_width, int cpusize, int shape);    // special for 1 diemnsion
-  int partition2(int *nxy, int split_size, int *min_width, int cpusize, int *shape); // special for 2 diemnsions
-  int partition3(int *nxyz, int split_size, int *min_width, int cpusize, int *shape);
-  MyList<Block> *distribute(MyList<Patch> *PatchLIST, int cpusize, int ingfsi, int fngfs, bool periodic, int nodes = 0); // produce corresponding Blocks
-  void KillBlocks(MyList<Patch> *PatchLIST);
-
-  void setfunction(MyList<Block> *BlL, var *vn, double func(double x, double y, double z));
-  void setfunction(int rank, MyList<Block> *BlL, var *vn, double func(double x, double y, double z));
-  void writefile(double time, int nx, int ny, int nz, double xmin, double xmax, double ymin, double ymax,
-                 double zmin, double zmax, char *filename, double *data_out);
-  void writefile(double time, int nx, int ny, double xmin, double xmax, double ymin, double ymax,
-                 char *filename, double *datain);
-  void getarrayindex(int DIM, int *shape, int *index, int n);
-  int getarraylocation(int DIM, int *shape, int *index);
-  void copy(int DIM, double *llbout, double *uubout, int *Dshape, double *DD, double *llbin, double *uubin,
-            int *shape, double *datain, double *llb, double *uub);
-  void Dump_CPU_Data(MyList<Block> *BlL, MyList<var> *DumpList, char *tag, double time, double dT);
-  void Dump_Data(MyList<Patch> *PL, MyList<var> *DumpList, char *tag, double time, double dT);
-  void Dump_Data(Patch *PP, MyList<var> *DumpList, char *tag, double time, double dT, int grd);
-  double *Collect_Data(Patch *PP, var *VP);
-  void d2Dump_Data(MyList<Patch> *PL, MyList<var> *DumpList, char *tag, double time, double dT);
-  void d2Dump_Data(Patch *PP, MyList<var> *DumpList, char *tag, double time, double dT, int grd);
-  void Dump_Data0(Patch *PP, MyList<var> *DumpList, char *tag, double time, double dT);
-  double global_interp(int DIM, int *ext, double **CoX, double *datain,
-                       double *poX, int ordn, double *SoA, int Symmetry);
-  double global_interp(int DIM, int *ext, double **CoX, double *datain,
-                       double *poX, int ordn);
-  double Lagrangian_Int(double x, int npts, double *xpts, double *funcvals);
-  double LagrangePoly(double x, int pt, int npts, double *xpts);
-  MyList<gridseg> *build_complete_gsl(Patch *Pat);
-  MyList<gridseg> *build_complete_gsl(MyList<Patch> *PatL);
-  MyList<gridseg> *build_complete_gsl_virtual(MyList<Patch> *PatL);
-  MyList<gridseg> *build_complete_gsl_virtual2(MyList<Patch> *PatL);        // - buffer
-  MyList<gridseg> *build_owned_gsl0(Patch *Pat, int rank_in);               // - ghost without extension, special for Sync usage
-  MyList<gridseg> *build_owned_gsl1(Patch *Pat, int rank_in);               // - ghost, similar to build_owned_gsl0 but extend one point on left side for vertex grid
-  MyList<gridseg> *build_owned_gsl2(Patch *Pat, int rank_in);               // - buffer - ghost
-  MyList<gridseg> *build_owned_gsl3(Patch *Pat, int rank_in, int Symmetry); // - ghost - BD ghost
-  MyList<gridseg> *build_owned_gsl4(Patch *Pat, int rank_in, int Symmetry); // - buffer - ghost - BD ghost
-  MyList<gridseg> *build_owned_gsl5(Patch *Pat, int rank_in);               // similar to build_owned_gsl2 but no extension
-  MyList<gridseg> *build_owned_gsl(MyList<Patch> *PatL, int rank_in, int type, int Symmetry);
-  void build_gstl(MyList<gridseg> *srci, MyList<gridseg> *dsti, MyList<gridseg> **out_src, MyList<gridseg> **out_dst);
-  int data_packer(double *data, MyList<gridseg> *src, MyList<gridseg> *dst, int rank_in, int dir,
-                  MyList<var> *VarLists, MyList<var> *VarListd, int Symmetry);
-  void transfer(MyList<gridseg> **src, MyList<gridseg> **dst,
-                MyList<var> *VarList1 /* source */, MyList<var> *VarList2 /*target */,
-                int Symmetry);
-  int data_packermix(double *data, MyList<gridseg> *src, MyList<gridseg> *dst, int rank_in, int dir,
-                     MyList<var> *VarLists, MyList<var> *VarListd, int Symmetry);
-  void transfermix(MyList<gridseg> **src, MyList<gridseg> **dst,
-                   MyList<var> *VarList1 /* source */, MyList<var> *VarList2 /*target */,
-                   int Symmetry);
-  void Sync(Patch *Pat, MyList<var> *VarList, int Symmetry);
-  void Sync(MyList<Patch> *PatL, MyList<var> *VarList, int Symmetry);
-  void Sync_merged(MyList<Patch> *PatL, MyList<var> *VarList, int Symmetry);
-
-  struct SyncCache {
-    bool valid;
-    int cpusize;
-    MyList<gridseg> **combined_src;
-    MyList<gridseg> **combined_dst;
-    int *send_lengths;
-    int *recv_lengths;
-    double **send_bufs;
-    double **recv_bufs;
-    int *send_buf_caps;
-    int *recv_buf_caps;
-    MPI_Request *reqs;
-    MPI_Status *stats;
-    int max_reqs;
-    bool lengths_valid;
-    SyncCache();
-    void invalidate();
-    void destroy();
-  };
-
-  void Sync_cached(MyList<Patch> *PatL, MyList<var> *VarList, int Symmetry, SyncCache &cache);
-  void transfer_cached(MyList<gridseg> **src, MyList<gridseg> **dst,
-                       MyList<var> *VarList1, MyList<var> *VarList2,
-                       int Symmetry, SyncCache &cache);
-
-  struct AsyncSyncState {
-    int req_no;
-    bool active;
-    AsyncSyncState() : req_no(0), active(false) {}
-  };
-
-  void Sync_start(MyList<Patch> *PatL, MyList<var> *VarList, int Symmetry,
-                  SyncCache &cache, AsyncSyncState &state);
-  void Sync_finish(SyncCache &cache, AsyncSyncState &state,
-                   MyList<var> *VarList, int Symmetry);
-  void OutBdLow2Hi(Patch *Patc, Patch *Patf,
-                   MyList<var> *VarList1 /* source */, MyList<var> *VarList2 /* target */,
-                   int Symmetry);
-  void OutBdLow2Hi(MyList<Patch> *PatcL, MyList<Patch> *PatfL,
-                   MyList<var> *VarList1 /* source */, MyList<var> *VarList2 /* target */,
-                   int Symmetry);
-  void OutBdLow2Himix(Patch *Patc, Patch *Patf,
-                      MyList<var> *VarList1 /* source */, MyList<var> *VarList2 /* target */,
-                      int Symmetry);
-  void OutBdLow2Himix(MyList<Patch> *PatcL, MyList<Patch> *PatfL,
-                      MyList<var> *VarList1 /* source */, MyList<var> *VarList2 /* target */,
-                      int Symmetry);
-  void Restrict_cached(MyList<Patch> *PatcL, MyList<Patch> *PatfL,
-                       MyList<var> *VarList1, MyList<var> *VarList2,
-                       int Symmetry, SyncCache &cache);
-  void OutBdLow2Hi_cached(MyList<Patch> *PatcL, MyList<Patch> *PatfL,
-                          MyList<var> *VarList1, MyList<var> *VarList2,
-                          int Symmetry, SyncCache &cache);
-  void OutBdLow2Himix_cached(MyList<Patch> *PatcL, MyList<Patch> *PatfL,
-                             MyList<var> *VarList1, MyList<var> *VarList2,
-                             int Symmetry, SyncCache &cache);
-  void Prolong(Patch *Patc, Patch *Patf,
-               MyList<var> *VarList1 /* source */, MyList<var> *VarList2 /* target */,
-               int Symmetry);
-  void Prolongint(Patch *Patc, Patch *Patf,
-                  MyList<var> *VarList1 /* source */, MyList<var> *VarList2 /* target */,
-                  int Symmetry);
-  void Restrict(MyList<Patch> *PatcL, MyList<Patch> *PatfL,
-                MyList<var> *VarList1 /* source */, MyList<var> *VarList2 /* target */,
-                int Symmetry);
-  void Restrict_after(MyList<Patch> *PatcL, MyList<Patch> *PatfL,
-                      MyList<var> *VarList1 /* source */, MyList<var> *VarList2 /* target */,
-                      int Symmetry); // for -ghost - BDghost
-  MyList<Parallel::gridseg> *build_PhysBD_gsl(Patch *Pat);
-  MyList<Parallel::gridseg> *build_ghost_gsl(MyList<Patch> *PatL);
-  MyList<Parallel::gridseg> *build_ghost_gsl(Patch *Pat);
-  MyList<Parallel::gridseg> *build_buffer_gsl(Patch *Pat);
-  MyList<Parallel::gridseg> *build_buffer_gsl(MyList<Patch> *PatL);
-  MyList<Parallel::gridseg> *gsl_subtract(MyList<Parallel::gridseg> *A, MyList<Parallel::gridseg> *B);
-  MyList<Parallel::gridseg> *gs_subtract(MyList<Parallel::gridseg> *A, MyList<Parallel::gridseg> *B);
-  MyList<Parallel::gridseg> *gsl_and(MyList<Parallel::gridseg> *A, MyList<Parallel::gridseg> *B);
-  MyList<Parallel::gridseg> *gs_and(MyList<Parallel::gridseg> *A, MyList<Parallel::gridseg> *B);
-  MyList<Parallel::gridseg> *clone_gsl(MyList<Parallel::gridseg> *p, bool first_only);
-  MyList<Parallel::gridseg> *build_bulk_gsl(Patch *Pat); // similar to build_owned_gsl0 but does not care rank issue
-  MyList<Parallel::gridseg> *build_bulk_gsl(Block *bp, Patch *Pat);
-  void build_PhysBD_gstl(Patch *Pat, MyList<Parallel::gridseg> *srci, MyList<Parallel::gridseg> *dsti,
-                         MyList<Parallel::gridseg> **out_src, MyList<Parallel::gridseg> **out_dst);
-  void PeriodicBD(Patch *Pat, MyList<var> *VarList, int Symmetry);
-  double L2Norm(Patch *Pat, var *vf);
-  void checkgsl(MyList<Parallel::gridseg> *pp, bool first_only);
-  void checkvarl(MyList<var> *pp, bool first_only);
-  MyList<Parallel::gridseg> *divide_gsl(MyList<Parallel::gridseg> *p, Patch *Pat);
-  MyList<Parallel::gridseg> *divide_gs(MyList<Parallel::gridseg> *p, Patch *Pat);
-  void prepare_inter_time_level(Patch *Pat,
-                                MyList<var> *VarList1 /* source (t+dt) */, MyList<var> *VarList2 /* source (t) */,
-                                MyList<var> *VarList3 /* target (t+a*dt) */, int tindex);
-  void prepare_inter_time_level(Patch *Pat,
-                                MyList<var> *VarList1 /* source (t+dt) */, MyList<var> *VarList2 /* source (t) */,
-                                MyList<var> *VarList3 /* source (t-dt) */, MyList<var> *VarList4 /* target (t+a*dt) */, int tindex);
-  void prepare_inter_time_level(MyList<Patch> *PatL,
-                                MyList<var> *VarList1 /* source (t+dt) */, MyList<var> *VarList2 /* source (t) */,
-                                MyList<var> *VarList3 /* target (t+a*dt) */, int tindex);
-  void prepare_inter_time_level(MyList<Patch> *Pat,
-                                MyList<var> *VarList1 /* source (t+dt) */, MyList<var> *VarList2 /* source (t) */,
-                                MyList<var> *VarList3 /* source (t-dt) */, MyList<var> *VarList4 /* target (t+a*dt) */, int tindex);
-  void merge_gsl(MyList<gridseg> *&A, const double ratio);
-  bool merge_gs(MyList<gridseg> *D, MyList<gridseg> *B, MyList<gridseg> *&C, const double ratio);
-  // Add ghost region to tangent plane
-  // we assume the grids have the same resolution
-  void add_ghost_touch(MyList<gridseg> *&A);
-  void cut_gsl(MyList<gridseg> *&A);
-  bool cut_gs(MyList<gridseg> *D, MyList<gridseg> *B, MyList<gridseg> *&C);
-  MyList<Parallel::gridseg> *gs_subtract_virtual(MyList<Parallel::gridseg> *A, MyList<Parallel::gridseg> *B);
-  void fill_level_data(MyList<Patch> *PatLd, MyList<Patch> *PatLs, MyList<Patch> *PatcL,
-                       MyList<var> *OldList, MyList<var> *StateList, MyList<var> *FutureList,
-                       MyList<var> *tmList, int Symmetry, bool BB, bool CC);
-  bool PatList_Interp_Points(MyList<Patch> *PatL, MyList<var> *VarList,
-                             int NN, double **XX,
-                             double *Shellf, int Symmetry);
-  void aligncheck(double *bbox0, double *bboxl, int lev, double *DH0, int *shape);
-  bool point_locat_gsl(double *pox, MyList<Parallel::gridseg> *gsl);
-  void checkpatchlist(MyList<Patch> *PatL, bool buflog);
-
-  double L2Norm(Patch *Pat, var *vf, MPI_Comm Comm_here);
-  bool PatList_Interp_Points(MyList<Patch> *PatL, MyList<var> *VarList,
-                             int NN, double **XX,
-                             double *Shellf, int Symmetry, MPI_Comm Comm_here);
-#if (PSTR == 1 || PSTR == 2 || PSTR == 3)
-  MyList<Block> *distribute(MyList<Patch> *PatchLIST, int cpusize, int ingfsi, int fngfsi,
-                            bool periodic, int start_rank, int end_rank, int nodes = 0);
-#endif
-}
-#endif /*PARALLEL_H */
+
+#ifndef PARALLEL_H
+#define PARALLEL_H
+
+#include <iostream>
+#include <iomanip>
+#include <fstream>
+#include <cstdlib>
+#include <cstdio>
+#include <string>
+#include <cmath>
+#include <new>
+using namespace std;
+#include <memory.h>
+#include "Parallel_bam.h"
+#include "var.h"
+#include "MPatch.h"
+#include "Block.h"
+#include "MyList.h"
+#include "macrodef.h" //need dim; ghost_width; CONTRACT
+namespace Parallel
+{
+  struct gridseg
+  {
+    double llb[dim];
+    double uub[dim];
+    int shape[dim];
+    double illb[dim], iuub[dim]; // only use for OutBdLow2Hi
+    Block *Bg;
+  };
+  int partition1(int &nx, int split_size, int min_width, int cpusize, int shape);    // special for 1 diemnsion
+  int partition2(int *nxy, int split_size, int *min_width, int cpusize, int *shape); // special for 2 diemnsions
+  int partition3(int *nxyz, int split_size, int *min_width, int cpusize, int *shape);
+  MyList<Block> *distribute(MyList<Patch> *PatchLIST, int cpusize, int ingfsi, int fngfs, bool periodic, int nodes = 0); // produce corresponding Blocks
+  MyList<Block> *distribute_hard(MyList<Patch> *PatchLIST, int cpusize, int ingfsi, int fngfs, bool periodic, int nodes = 0); // produce corresponding Blocks
+  Block* splitHotspotBlock(MyList<Block>* &BlL, int _dim, 
+                                 int ib0_orig, int ib3_orig, 
+                                 int jb1_orig, int jb4_orig, 
+                                 int kb2_orig, int kb5_orig, 
+                                 Patch* PP, int r_left, int r_right, 
+                                 int ingfsi, int fngfsi, bool periodic,
+                                 Block* &split_first_block, Block* &split_last_block);
+  Block* createMappedBlock(MyList<Block>* &BlL, int _dim, int* shape, double* bbox,
+                        int block_id, int ingfsi, int fngfsi, int lev); 
+  void KillBlocks(MyList<Patch> *PatchLIST);
+
+  void setfunction(MyList<Block> *BlL, var *vn, double func(double x, double y, double z));
+  void setfunction(int rank, MyList<Block> *BlL, var *vn, double func(double x, double y, double z));
+  void writefile(double time, int nx, int ny, int nz, double xmin, double xmax, double ymin, double ymax,
+                 double zmin, double zmax, char *filename, double *data_out);
+  void writefile(double time, int nx, int ny, double xmin, double xmax, double ymin, double ymax,
+                 char *filename, double *datain);
+  void getarrayindex(int DIM, int *shape, int *index, int n);
+  int getarraylocation(int DIM, int *shape, int *index);
+  void copy(int DIM, double *llbout, double *uubout, int *Dshape, double *DD, double *llbin, double *uubin,
+            int *shape, double *datain, double *llb, double *uub);
+  void Dump_CPU_Data(MyList<Block> *BlL, MyList<var> *DumpList, char *tag, double time, double dT);
+  void Dump_Data(MyList<Patch> *PL, MyList<var> *DumpList, char *tag, double time, double dT);
+  void Dump_Data(Patch *PP, MyList<var> *DumpList, char *tag, double time, double dT, int grd);
+  double *Collect_Data(Patch *PP, var *VP);
+  void d2Dump_Data(MyList<Patch> *PL, MyList<var> *DumpList, char *tag, double time, double dT);
+  void d2Dump_Data(Patch *PP, MyList<var> *DumpList, char *tag, double time, double dT, int grd);
+  void Dump_Data0(Patch *PP, MyList<var> *DumpList, char *tag, double time, double dT);
+  double global_interp(int DIM, int *ext, double **CoX, double *datain,
+                       double *poX, int ordn, double *SoA, int Symmetry);
+  double global_interp(int DIM, int *ext, double **CoX, double *datain,
+                       double *poX, int ordn);
+  double Lagrangian_Int(double x, int npts, double *xpts, double *funcvals);
+  double LagrangePoly(double x, int pt, int npts, double *xpts);
+  MyList<gridseg> *build_complete_gsl(Patch *Pat);
+  MyList<gridseg> *build_complete_gsl(MyList<Patch> *PatL);
+  MyList<gridseg> *build_complete_gsl_virtual(MyList<Patch> *PatL);
+  MyList<gridseg> *build_complete_gsl_virtual2(MyList<Patch> *PatL);        // - buffer
+  MyList<gridseg> *build_owned_gsl0(Patch *Pat, int rank_in);               // - ghost without extension, special for Sync usage
+  MyList<gridseg> *build_owned_gsl1(Patch *Pat, int rank_in);               // - ghost, similar to build_owned_gsl0 but extend one point on left side for vertex grid
+  MyList<gridseg> *build_owned_gsl2(Patch *Pat, int rank_in);               // - buffer - ghost
+  MyList<gridseg> *build_owned_gsl3(Patch *Pat, int rank_in, int Symmetry); // - ghost - BD ghost
+  MyList<gridseg> *build_owned_gsl4(Patch *Pat, int rank_in, int Symmetry); // - buffer - ghost - BD ghost
+  MyList<gridseg> *build_owned_gsl5(Patch *Pat, int rank_in);               // similar to build_owned_gsl2 but no extension
+  MyList<gridseg> *build_owned_gsl(MyList<Patch> *PatL, int rank_in, int type, int Symmetry);
+  void build_gstl(MyList<gridseg> *srci, MyList<gridseg> *dsti, MyList<gridseg> **out_src, MyList<gridseg> **out_dst);
+  int data_packer(double *data, MyList<gridseg> *src, MyList<gridseg> *dst, int rank_in, int dir,
+                  MyList<var> *VarLists, MyList<var> *VarListd, int Symmetry);
+  void transfer(MyList<gridseg> **src, MyList<gridseg> **dst,
+                MyList<var> *VarList1 /* source */, MyList<var> *VarList2 /*target */,
+                int Symmetry);
+  int data_packermix(double *data, MyList<gridseg> *src, MyList<gridseg> *dst, int rank_in, int dir,
+                     MyList<var> *VarLists, MyList<var> *VarListd, int Symmetry);
+  void transfermix(MyList<gridseg> **src, MyList<gridseg> **dst,
+                   MyList<var> *VarList1 /* source */, MyList<var> *VarList2 /*target */,
+                   int Symmetry);
+  void Sync(Patch *Pat, MyList<var> *VarList, int Symmetry);
+  void Sync(MyList<Patch> *PatL, MyList<var> *VarList, int Symmetry);
+  void Sync_merged(MyList<Patch> *PatL, MyList<var> *VarList, int Symmetry);
+
+  struct SyncCache {
+    bool valid;
+    int cpusize;
+    MyList<gridseg> **combined_src;
+    MyList<gridseg> **combined_dst;
+    int *send_lengths;
+    int *recv_lengths;
+    double **send_bufs;
+    double **recv_bufs;
+    int *send_buf_caps;
+    int *recv_buf_caps;
+    MPI_Request *reqs;
+    MPI_Status *stats;
+    int max_reqs;
+    bool lengths_valid;
+    SyncCache();
+    void invalidate();
+    void destroy();
+  };
+
+  void Sync_cached(MyList<Patch> *PatL, MyList<var> *VarList, int Symmetry, SyncCache &cache);
+  void transfer_cached(MyList<gridseg> **src, MyList<gridseg> **dst,
+                       MyList<var> *VarList1, MyList<var> *VarList2,
+                       int Symmetry, SyncCache &cache);
+
+  struct AsyncSyncState {
+    int req_no;
+    bool active;
+    AsyncSyncState() : req_no(0), active(false) {}
+  };
+
+  void Sync_start(MyList<Patch> *PatL, MyList<var> *VarList, int Symmetry,
+                  SyncCache &cache, AsyncSyncState &state);
+  void Sync_finish(SyncCache &cache, AsyncSyncState &state,
+                   MyList<var> *VarList, int Symmetry);
+  void OutBdLow2Hi(Patch *Patc, Patch *Patf,
+                   MyList<var> *VarList1 /* source */, MyList<var> *VarList2 /* target */,
+                   int Symmetry);
+  void OutBdLow2Hi(MyList<Patch> *PatcL, MyList<Patch> *PatfL,
+                   MyList<var> *VarList1 /* source */, MyList<var> *VarList2 /* target */,
+                   int Symmetry);
+  void OutBdLow2Himix(Patch *Patc, Patch *Patf,
+                      MyList<var> *VarList1 /* source */, MyList<var> *VarList2 /* target */,
+                      int Symmetry);
+  void OutBdLow2Himix(MyList<Patch> *PatcL, MyList<Patch> *PatfL,
+                      MyList<var> *VarList1 /* source */, MyList<var> *VarList2 /* target */,
+                      int Symmetry);
+  void Restrict_cached(MyList<Patch> *PatcL, MyList<Patch> *PatfL,
+                       MyList<var> *VarList1, MyList<var> *VarList2,
+                       int Symmetry, SyncCache &cache);
+  void OutBdLow2Hi_cached(MyList<Patch> *PatcL, MyList<Patch> *PatfL,
+                          MyList<var> *VarList1, MyList<var> *VarList2,
+                          int Symmetry, SyncCache &cache);
+  void OutBdLow2Himix_cached(MyList<Patch> *PatcL, MyList<Patch> *PatfL,
+                             MyList<var> *VarList1, MyList<var> *VarList2,
+                             int Symmetry, SyncCache &cache);
+  void Prolong(Patch *Patc, Patch *Patf,
+               MyList<var> *VarList1 /* source */, MyList<var> *VarList2 /* target */,
+               int Symmetry);
+  void Prolongint(Patch *Patc, Patch *Patf,
+                  MyList<var> *VarList1 /* source */, MyList<var> *VarList2 /* target */,
+                  int Symmetry);
+  void Restrict(MyList<Patch> *PatcL, MyList<Patch> *PatfL,
+                MyList<var> *VarList1 /* source */, MyList<var> *VarList2 /* target */,
+                int Symmetry);
+  void Restrict_after(MyList<Patch> *PatcL, MyList<Patch> *PatfL,
+                      MyList<var> *VarList1 /* source */, MyList<var> *VarList2 /* target */,
+                      int Symmetry); // for -ghost - BDghost
+  MyList<Parallel::gridseg> *build_PhysBD_gsl(Patch *Pat);
+  MyList<Parallel::gridseg> *build_ghost_gsl(MyList<Patch> *PatL);
+  MyList<Parallel::gridseg> *build_ghost_gsl(Patch *Pat);
+  MyList<Parallel::gridseg> *build_buffer_gsl(Patch *Pat);
+  MyList<Parallel::gridseg> *build_buffer_gsl(MyList<Patch> *PatL);
+  MyList<Parallel::gridseg> *gsl_subtract(MyList<Parallel::gridseg> *A, MyList<Parallel::gridseg> *B);
+  MyList<Parallel::gridseg> *gs_subtract(MyList<Parallel::gridseg> *A, MyList<Parallel::gridseg> *B);
+  MyList<Parallel::gridseg> *gsl_and(MyList<Parallel::gridseg> *A, MyList<Parallel::gridseg> *B);
+  MyList<Parallel::gridseg> *gs_and(MyList<Parallel::gridseg> *A, MyList<Parallel::gridseg> *B);
+  MyList<Parallel::gridseg> *clone_gsl(MyList<Parallel::gridseg> *p, bool first_only);
+  MyList<Parallel::gridseg> *build_bulk_gsl(Patch *Pat); // similar to build_owned_gsl0 but does not care rank issue
+  MyList<Parallel::gridseg> *build_bulk_gsl(Block *bp, Patch *Pat);
+  void build_PhysBD_gstl(Patch *Pat, MyList<Parallel::gridseg> *srci, MyList<Parallel::gridseg> *dsti,
+                         MyList<Parallel::gridseg> **out_src, MyList<Parallel::gridseg> **out_dst);
+  void PeriodicBD(Patch *Pat, MyList<var> *VarList, int Symmetry);
+  double L2Norm(Patch *Pat, var *vf);
+  void checkgsl(MyList<Parallel::gridseg> *pp, bool first_only);
+  void checkvarl(MyList<var> *pp, bool first_only);
+  MyList<Parallel::gridseg> *divide_gsl(MyList<Parallel::gridseg> *p, Patch *Pat);
+  MyList<Parallel::gridseg> *divide_gs(MyList<Parallel::gridseg> *p, Patch *Pat);
+  void prepare_inter_time_level(Patch *Pat,
+                                MyList<var> *VarList1 /* source (t+dt) */, MyList<var> *VarList2 /* source (t) */,
+                                MyList<var> *VarList3 /* target (t+a*dt) */, int tindex);
+  void prepare_inter_time_level(Patch *Pat,
+                                MyList<var> *VarList1 /* source (t+dt) */, MyList<var> *VarList2 /* source (t) */,
+                                MyList<var> *VarList3 /* source (t-dt) */, MyList<var> *VarList4 /* target (t+a*dt) */, int tindex);
+  void prepare_inter_time_level(MyList<Patch> *PatL,
+                                MyList<var> *VarList1 /* source (t+dt) */, MyList<var> *VarList2 /* source (t) */,
+                                MyList<var> *VarList3 /* target (t+a*dt) */, int tindex);
+  void prepare_inter_time_level(MyList<Patch> *Pat,
+                                MyList<var> *VarList1 /* source (t+dt) */, MyList<var> *VarList2 /* source (t) */,
+                                MyList<var> *VarList3 /* source (t-dt) */, MyList<var> *VarList4 /* target (t+a*dt) */, int tindex);
+  void merge_gsl(MyList<gridseg> *&A, const double ratio);
+  bool merge_gs(MyList<gridseg> *D, MyList<gridseg> *B, MyList<gridseg> *&C, const double ratio);
+  // Add ghost region to tangent plane
+  // we assume the grids have the same resolution
+  void add_ghost_touch(MyList<gridseg> *&A);
+  void cut_gsl(MyList<gridseg> *&A);
+  bool cut_gs(MyList<gridseg> *D, MyList<gridseg> *B, MyList<gridseg> *&C);
+  MyList<Parallel::gridseg> *gs_subtract_virtual(MyList<Parallel::gridseg> *A, MyList<Parallel::gridseg> *B);
+  void fill_level_data(MyList<Patch> *PatLd, MyList<Patch> *PatLs, MyList<Patch> *PatcL,
+                       MyList<var> *OldList, MyList<var> *StateList, MyList<var> *FutureList,
+                       MyList<var> *tmList, int Symmetry, bool BB, bool CC);
+  bool PatList_Interp_Points(MyList<Patch> *PatL, MyList<var> *VarList,
+                             int NN, double **XX,
+                             double *Shellf, int Symmetry);
+  void aligncheck(double *bbox0, double *bboxl, int lev, double *DH0, int *shape);
+  bool point_locat_gsl(double *pox, MyList<Parallel::gridseg> *gsl);
+  void checkpatchlist(MyList<Patch> *PatL, bool buflog);
+
+  double L2Norm(Patch *Pat, var *vf, MPI_Comm Comm_here);
+  bool PatList_Interp_Points(MyList<Patch> *PatL, MyList<var> *VarList,
+                             int NN, double **XX,
+                             double *Shellf, int Symmetry, MPI_Comm Comm_here);
+#if (PSTR == 1 || PSTR == 2 || PSTR == 3)
+  MyList<Block> *distribute(MyList<Patch> *PatchLIST, int cpusize, int ingfsi, int fngfsi,
+                            bool periodic, int start_rank, int end_rank, int nodes = 0);
+  
+  // Redistribute blocks with time statistics for load balancing
+  MyList<Block> *distribute(MyList<Patch> *PatchLIST, MyList<Block> *OldBlockL,
+                            int cpusize, int ingfsi, int fngfsi,
+                            bool periodic, int start_rank, int end_rank, int nodes = 0);
+#endif
+
+  // Dynamic load balancing: split blocks for heavy ranks
+    void split_heavy_blocks(MyList<Patch> *PatL, int *heavy_ranks, int num_heavy,
+                            int split_factor, int cpusize, int ingfsi, int fngfsi);
+
+    // Check if load balancing is needed based on interpolation times
+    bool check_load_balance_need(double *rank_times, int nprocs, int &num_heavy, int *heavy_ranks);
+  }
+  #endif /*PARALLEL_H */

AMSS_NCKU_source/Z4c_class.C

@@ -321,7 +321,22 @@ void Z4c_class::Step(int lev, int YN)
     }
     Pp = Pp->next;
   }
-  // NOTE: error check deferred to after Shell Patch computation to reduce MPI_Allreduce calls
+  // check error information
+  {
+    int erh = ERROR;
+    MPI_Allreduce(&erh, &ERROR, 1, MPI_INT, MPI_SUM, MPI_COMM_WORLD);
+  }
+  if (ERROR)
+  {
+    Parallel::Dump_Data(GH->PatL[lev], StateList, 0, PhysTime, dT_lev);
+    if (myrank == 0)
+    {
+      if (ErrorMonitor->outfile)
+        ErrorMonitor->outfile << "find NaN in state variables at t = " << PhysTime 
+                              << ", lev = " << lev << endl;
+      MPI_Abort(MPI_COMM_WORLD, 1);
+    }
+  }
 
 #ifdef WithShell
   // evolve Shell Patches
@@ -339,9 +354,9 @@ void Z4c_class::Step(int lev, int YN)
         {
 #if (AGM == 0)
           f_enforce_ga(cg->shape,
-                       cg->fgfs[gxx0->sgfn], cg->fgfs[gxy0->sgfn], cg->fgfs[gxz0->sgfn],
+                       cg->fgfs[gxx0->sgfn], cg->fgfs[gxy0->sgfn], cg->fgfs[gxz0->sgfn], 
                        cg->fgfs[gyy0->sgfn], cg->fgfs[gyz0->sgfn], cg->fgfs[gzz0->sgfn],
-                       cg->fgfs[Axx0->sgfn], cg->fgfs[Axy0->sgfn], cg->fgfs[Axz0->sgfn],
+                       cg->fgfs[Axx0->sgfn], cg->fgfs[Axy0->sgfn], cg->fgfs[Axz0->sgfn], 
                        cg->fgfs[Ayy0->sgfn], cg->fgfs[Ayz0->sgfn], cg->fgfs[Azz0->sgfn]);
 #endif
 
@@ -453,16 +468,24 @@ void Z4c_class::Step(int lev, int YN)
       sPp = sPp->next;
     }
   }
-  // Non-blocking error reduction overlapped with Sync to hide Allreduce latency
-  MPI_Request err_req_pre;
+  // check error information
   {
     int erh = ERROR;
-    MPI_Iallreduce(&erh, &ERROR, 1, MPI_INT, MPI_SUM, MPI_COMM_WORLD, &err_req_pre);
+    MPI_Allreduce(&erh, &ERROR, 1, MPI_INT, MPI_SUM, MPI_COMM_WORLD);
+  }
+  if (ERROR)
+  {
+    SH->Dump_Data(StateList, 0, PhysTime, dT_lev);
+    if (myrank == 0)
+    {
+      if (ErrorMonitor->outfile)
+        ErrorMonitor->outfile << "find NaN in state variables on Shell Patches at t = " << PhysTime << endl;
+      MPI_Abort(MPI_COMM_WORLD, 1);
+    }
   }
 #endif
 
-  Parallel::AsyncSyncState async_pre;
-  Parallel::Sync_start(GH->PatL[lev], SynchList_pre, Symmetry, sync_cache_pre[lev], async_pre);
+  Parallel::Sync(GH->PatL[lev], SynchList_pre, Symmetry);
 
 #ifdef WithShell
   if (lev == 0)
@@ -475,30 +498,12 @@ void Z4c_class::Step(int lev, int YN)
     {
       prev_clock = curr_clock;
       curr_clock = clock();
-      cout << " Shell stuff synchronization used "
-           << (double)(curr_clock - prev_clock) / ((double)CLOCKS_PER_SEC)
+      cout << " Shell stuff synchronization used " 
+           << (double)(curr_clock - prev_clock) / ((double)CLOCKS_PER_SEC) 
            << " seconds! " << endl;
     }
   }
 #endif
-  Parallel::Sync_finish(sync_cache_pre[lev], async_pre, SynchList_pre, Symmetry);
-
-#ifdef WithShell
-  // Complete non-blocking error reduction and check
-  MPI_Wait(&err_req_pre, MPI_STATUS_IGNORE);
-  if (ERROR)
-  {
-    Parallel::Dump_Data(GH->PatL[lev], StateList, 0, PhysTime, dT_lev);
-    SH->Dump_Data(StateList, 0, PhysTime, dT_lev);
-    if (myrank == 0)
-    {
-      if (ErrorMonitor->outfile)
-        ErrorMonitor->outfile << "find NaN in state variables at t = " << PhysTime
-                              << ", lev = " << lev << endl;
-      MPI_Abort(MPI_COMM_WORLD, 1);
-    }
-  }
-#endif
 
   // for black hole position
   if (BH_num > 0 && lev == GH->levels - 1)
@@ -688,7 +693,23 @@ void Z4c_class::Step(int lev, int YN)
       Pp = Pp->next;
     }
 
-    // NOTE: error check deferred to after Shell Patch computation to reduce MPI_Allreduce calls
+    // check error information
+    {
+      int erh = ERROR;
+      MPI_Allreduce(&erh, &ERROR, 1, MPI_INT, MPI_SUM, MPI_COMM_WORLD);
+    }
+    if (ERROR)
+    {
+      Parallel::Dump_Data(GH->PatL[lev], SynchList_pre, 0, PhysTime, dT_lev);
+      if (myrank == 0)
+      {
+        if (ErrorMonitor->outfile)
+          ErrorMonitor->outfile << "find NaN in RK4 substep#" << iter_count 
+                                << " variables at t = " << PhysTime 
+                                << ", lev = " << lev << endl;
+        MPI_Abort(MPI_COMM_WORLD, 1);
+      }
+    }
 
 #ifdef WithShell
     // evolve Shell Patches
@@ -829,16 +850,25 @@ void Z4c_class::Step(int lev, int YN)
         sPp = sPp->next;
       }
     }
-    // Non-blocking error reduction overlapped with Sync to hide Allreduce latency
-    MPI_Request err_req_cor;
+    // check error information
     {
       int erh = ERROR;
-      MPI_Iallreduce(&erh, &ERROR, 1, MPI_INT, MPI_SUM, MPI_COMM_WORLD, &err_req_cor);
+      MPI_Allreduce(&erh, &ERROR, 1, MPI_INT, MPI_SUM, MPI_COMM_WORLD);
+    }
+    if (ERROR)
+    {
+      SH->Dump_Data(SynchList_pre, 0, PhysTime, dT_lev);
+      if (myrank == 0)
+      {
+        if (ErrorMonitor->outfile)
+          ErrorMonitor->outfile << "find NaN on Shell Patches in RK4 substep#" << iter_count 
+                                << " variables at t = " << PhysTime << endl;
+        MPI_Abort(MPI_COMM_WORLD, 1);
+      }
     }
 #endif
 
-    Parallel::AsyncSyncState async_cor;
-    Parallel::Sync_start(GH->PatL[lev], SynchList_cor, Symmetry, sync_cache_cor[lev], async_cor);
+    Parallel::Sync(GH->PatL[lev], SynchList_cor, Symmetry);
 
 #ifdef WithShell
     if (lev == 0)
@@ -851,30 +881,11 @@ void Z4c_class::Step(int lev, int YN)
       {
         prev_clock = curr_clock;
         curr_clock = clock();
-        cout << " Shell stuff synchronization used "
-             << (double)(curr_clock - prev_clock) / ((double)CLOCKS_PER_SEC)
+        cout << " Shell stuff synchronization used " 
+             << (double)(curr_clock - prev_clock) / ((double)CLOCKS_PER_SEC) 
              << " seconds! " << endl;
       }
     }
-#endif
-    Parallel::Sync_finish(sync_cache_cor[lev], async_cor, SynchList_cor, Symmetry);
-
-#ifdef WithShell
-    // Complete non-blocking error reduction and check
-    MPI_Wait(&err_req_cor, MPI_STATUS_IGNORE);
-    if (ERROR)
-    {
-      Parallel::Dump_Data(GH->PatL[lev], SynchList_pre, 0, PhysTime, dT_lev);
-      SH->Dump_Data(SynchList_pre, 0, PhysTime, dT_lev);
-      if (myrank == 0)
-      {
-        if (ErrorMonitor->outfile)
-          ErrorMonitor->outfile << "find NaN in RK4 substep#" << iter_count
-                                << " variables at t = " << PhysTime
-                                << ", lev = " << lev << endl;
-        MPI_Abort(MPI_COMM_WORLD, 1);
-      }
-    }
 #endif
     // for black hole position
     if (BH_num > 0 && lev == GH->levels - 1)
@@ -1241,7 +1252,22 @@ void Z4c_class::Step(int lev, int YN)
 	 }
   }
 #endif
-  // NOTE: error check deferred to after Shell Patch computation to reduce MPI_Allreduce calls
+  // check error information
+  {
+    int erh = ERROR;
+    MPI_Allreduce(&erh, &ERROR, 1, MPI_INT, MPI_SUM, MPI_COMM_WORLD);
+  }
+  if (ERROR)
+  {
+    Parallel::Dump_Data(GH->PatL[lev], StateList, 0, PhysTime, dT_lev);
+    if (myrank == 0)
+    {
+      if (ErrorMonitor->outfile)
+        ErrorMonitor->outfile << "find NaN in state variables at t = " << PhysTime 
+                              << ", lev = " << lev << endl;
+      MPI_Abort(MPI_COMM_WORLD, 1);
+    }
+  }
 
   // evolve Shell Patches
   if (lev == 0)
@@ -1516,15 +1542,23 @@ void Z4c_class::Step(int lev, int YN)
   }
 #endif
   }
-  // Non-blocking error reduction overlapped with Sync to hide Allreduce latency
-  MPI_Request err_req_pre;
+  // check error information
   {
     int erh = ERROR;
-    MPI_Iallreduce(&erh, &ERROR, 1, MPI_INT, MPI_SUM, MPI_COMM_WORLD, &err_req_pre);
+    MPI_Allreduce(&erh, &ERROR, 1, MPI_INT, MPI_SUM, MPI_COMM_WORLD);
+  }
+  if (ERROR)
+  {
+    SH->Dump_Data(StateList, 0, PhysTime, dT_lev);
+    if (myrank == 0)
+    {
+      if (ErrorMonitor->outfile)
+        ErrorMonitor->outfile << "find NaN in state variables on Shell Patches at t = " << PhysTime << endl;
+      MPI_Abort(MPI_COMM_WORLD, 1);
+    }
   }
 
-  Parallel::AsyncSyncState async_pre;
-  Parallel::Sync_start(GH->PatL[lev], SynchList_pre, Symmetry, sync_cache_pre[lev], async_pre);
+  Parallel::Sync(GH->PatL[lev], SynchList_pre, Symmetry);
 
   if (lev == 0)
   {
@@ -1536,8 +1570,8 @@ void Z4c_class::Step(int lev, int YN)
     {
       prev_clock = curr_clock;
       curr_clock = clock();
-      cout << " Shell stuff synchronization used "
-           << (double)(curr_clock - prev_clock) / ((double)CLOCKS_PER_SEC)
+      cout << " Shell stuff synchronization used " 
+           << (double)(curr_clock - prev_clock) / ((double)CLOCKS_PER_SEC) 
            << " seconds! " << endl;
     }
 
@@ -1586,22 +1620,6 @@ void Z4c_class::Step(int lev, int YN)
   }
 #endif
   }
-  Parallel::Sync_finish(sync_cache_pre[lev], async_pre, SynchList_pre, Symmetry);
-
-  // Complete non-blocking error reduction and check
-  MPI_Wait(&err_req_pre, MPI_STATUS_IGNORE);
-  if (ERROR)
-  {
-    Parallel::Dump_Data(GH->PatL[lev], StateList, 0, PhysTime, dT_lev);
-    SH->Dump_Data(StateList, 0, PhysTime, dT_lev);
-    if (myrank == 0)
-    {
-      if (ErrorMonitor->outfile)
-        ErrorMonitor->outfile << "find NaN in state variables at t = " << PhysTime
-                              << ", lev = " << lev << endl;
-      MPI_Abort(MPI_COMM_WORLD, 1);
-    }
-  }
 
   // for black hole position
   if (BH_num > 0 && lev == GH->levels - 1)
@@ -1823,7 +1841,23 @@ void Z4c_class::Step(int lev, int YN)
       Pp = Pp->next;
     }
 
-    // NOTE: error check deferred to after Shell Patch computation to reduce MPI_Allreduce calls
+    // check error information
+    {
+      int erh = ERROR;
+      MPI_Allreduce(&erh, &ERROR, 1, MPI_INT, MPI_SUM, MPI_COMM_WORLD);
+    }
+    if (ERROR)
+    {
+      Parallel::Dump_Data(GH->PatL[lev], SynchList_pre, 0, PhysTime, dT_lev);
+      if (myrank == 0)
+      {
+        if (ErrorMonitor->outfile)
+          ErrorMonitor->outfile << "find NaN in RK4 substep#" << iter_count 
+                                << " variables at t = " << PhysTime 
+                                << ", lev = " << lev << endl;
+        MPI_Abort(MPI_COMM_WORLD, 1);
+      }
+    }
 
     // evolve Shell Patches
     if (lev == 0)
@@ -2069,15 +2103,24 @@ void Z4c_class::Step(int lev, int YN)
         sPp = sPp->next;
       }
     }
-    // Non-blocking error reduction overlapped with Sync to hide Allreduce latency
-    MPI_Request err_req_cor;
+    // check error information
     {
       int erh = ERROR;
-      MPI_Iallreduce(&erh, &ERROR, 1, MPI_INT, MPI_SUM, MPI_COMM_WORLD, &err_req_cor);
+      MPI_Allreduce(&erh, &ERROR, 1, MPI_INT, MPI_SUM, MPI_COMM_WORLD);
+    }
+    if (ERROR)
+    {
+      SH->Dump_Data(SynchList_pre, 0, PhysTime, dT_lev);
+      if (myrank == 0)
+      {
+        if (ErrorMonitor->outfile)
+          ErrorMonitor->outfile << "find NaN on Shell Patches in RK4 substep#" << iter_count 
+                                << " variables at t = " << PhysTime << endl;
+        MPI_Abort(MPI_COMM_WORLD, 1);
+      }
     }
 
-    Parallel::AsyncSyncState async_cor;
-    Parallel::Sync_start(GH->PatL[lev], SynchList_cor, Symmetry, sync_cache_cor[lev], async_cor);
+    Parallel::Sync(GH->PatL[lev], SynchList_cor, Symmetry);
 
     if (lev == 0)
     {
@@ -2089,8 +2132,8 @@ void Z4c_class::Step(int lev, int YN)
       {
         prev_clock = curr_clock;
         curr_clock = clock();
-        cout << " Shell stuff synchronization used "
-             << (double)(curr_clock - prev_clock) / ((double)CLOCKS_PER_SEC)
+        cout << " Shell stuff synchronization used " 
+             << (double)(curr_clock - prev_clock) / ((double)CLOCKS_PER_SEC) 
              << " seconds! " << endl;
       }
     }
@@ -2127,23 +2170,6 @@ void Z4c_class::Step(int lev, int YN)
     }
 // end smooth
 #endif
-    Parallel::Sync_finish(sync_cache_cor[lev], async_cor, SynchList_cor, Symmetry);
-
-    // Complete non-blocking error reduction and check
-    MPI_Wait(&err_req_cor, MPI_STATUS_IGNORE);
-    if (ERROR)
-    {
-      Parallel::Dump_Data(GH->PatL[lev], SynchList_pre, 0, PhysTime, dT_lev);
-      SH->Dump_Data(SynchList_pre, 0, PhysTime, dT_lev);
-      if (myrank == 0)
-      {
-        if (ErrorMonitor->outfile)
-          ErrorMonitor->outfile << "find NaN in RK4 substep#" << iter_count
-                                << " variables at t = " << PhysTime
-                                << ", lev = " << lev << endl;
-        MPI_Abort(MPI_COMM_WORLD, 1);
-      }
-    }
 
     // for black hole position
     if (BH_num > 0 && lev == GH->levels - 1)

AMSS_NCKU_source/cgh.h

@@ -1,92 +1,107 @@
-
-#ifndef CGH_H
-#define CGH_H
-
-#include <mpi.h>
-#include "MyList.h"
-#include "MPatch.h"
-#include "macrodef.h"
-#include "monitor.h"
-#include "Parallel.h"
-
-class cgh
-{
-
-public:
-   int levels, movls, BH_num_in;
-   // information of boxes
-   int *grids;
-   double ***bbox;
-   int ***shape;
-   double ***handle;
-   double ***Porgls;
-   double *Lt;
-
-   // information of Patch list
-   MyList<Patch> **PatL;
-
-// information of OutBdLow2Hi point list and Restrict point list
-#if (RPB == 1)
-   MyList<Parallel::pointstru_bam> **bdsul, **rsul;
-#endif
-
-#if (PSTR == 1 || PSTR == 2 || PSTR == 3)
-   int mylev;
-   int *start_rank, *end_rank;
-   MPI_Comm *Commlev;
-#endif
-
-protected:
-   int ingfs, fngfs;
-   static constexpr double ratio = 0.75;
-   int trfls;
-
-public:
-   cgh(int ingfsi, int fngfsi, int Symmetry, char *filename, int checkrun, monitor *ErrorMonitor);
-
-   ~cgh();
-
-   void compose_cgh(int nprocs);
-   void sethandle(monitor *ErrorMonitor);
-   void checkPatchList(MyList<Patch> *PatL, bool buflog);
-   void Regrid(int Symmetry, int BH_num, double **Porgbr, double **Porg0,
-               MyList<var> *OldList, MyList<var> *StateList,
-               MyList<var> *FutureList, MyList<var> *tmList, bool BB,
-               monitor *ErrorMonitor);
-   void Regrid_fake(int Symmetry, int BH_num, double **Porgbr, double **Porg0,
-                    MyList<var> *OldList, MyList<var> *StateList,
-                    MyList<var> *FutureList, MyList<var> *tmList, bool BB,
-                    monitor *ErrorMonitor);
-   void recompose_cgh(int nprocs, bool *lev_flag,
-                      MyList<var> *OldList, MyList<var> *StateList,
-                      MyList<var> *FutureList, MyList<var> *tmList,
-                      int Symmetry, bool BB);
-   void recompose_cgh_fake(int nprocs, bool *lev_flag,
-                           MyList<var> *OldList, MyList<var> *StateList,
-                           MyList<var> *FutureList, MyList<var> *tmList,
-                           int Symmetry, bool BB);
-   void read_bbox(int Symmetry, char *filename);
-   MyList<Patch> *construct_patchlist(int lev, int Symmetry);
-   bool Interp_One_Point(MyList<var> *VarList,
-                         double *XX, /*input global Cartesian coordinate*/
-                         double *Shellf, int Symmetry);
-   void recompose_cgh_Onelevel(int nprocs, int lev,
-                               MyList<var> *OldList, MyList<var> *StateList,
-                               MyList<var> *FutureList, MyList<var> *tmList,
-                               int Symmetry, bool BB);
-   void Regrid_Onelevel(int lev, int Symmetry, int BH_num, double **Porgbr, double **Porg0,
-                        MyList<var> *OldList, MyList<var> *StateList,
-                        MyList<var> *FutureList, MyList<var> *tmList, bool BB,
-                        monitor *ErrorMonitor);
-   void Regrid_Onelevel_aux(int lev, int Symmetry, int BH_num, double **Porgbr, double **Porg0,
-                            MyList<var> *OldList, MyList<var> *StateList,
-                            MyList<var> *FutureList, MyList<var> *tmList, bool BB,
-                            monitor *ErrorMonitor);
-   void settrfls(const int lev);
-
-#if (PSTR == 1 || PSTR == 2 || PSTR == 3)
-   void construct_mylev(int nprocs);
-#endif
-};
-
-#endif /* CGH_H */
+
+#ifndef CGH_H
+#define CGH_H
+
+#include <mpi.h>
+#include "MyList.h"
+#include "MPatch.h"
+#include "macrodef.h"
+#include "monitor.h"
+#include "Parallel.h"
+
+class cgh
+{
+
+public:
+   int levels, movls, BH_num_in;
+   // information of boxes
+   int *grids;
+   double ***bbox;
+   int ***shape;
+   double ***handle;
+   double ***Porgls;
+   double *Lt;
+
+   // information of Patch list
+   MyList<Patch> **PatL;
+
+// information of OutBdLow2Hi point list and Restrict point list
+#if (RPB == 1)
+   MyList<Parallel::pointstru_bam> **bdsul, **rsul;
+#endif
+
+#if (PSTR == 1 || PSTR == 2 || PSTR == 3)
+   int mylev;
+   int *start_rank, *end_rank;
+   MPI_Comm *Commlev;
+#endif
+
+protected:
+   int ingfs, fngfs;
+   static constexpr double ratio = 0.75;
+   int trfls;
+
+public:
+   cgh(int ingfsi, int fngfsi, int Symmetry, char *filename, int checkrun, monitor *ErrorMonitor);
+
+   ~cgh();
+
+   void compose_cgh(int nprocs);
+   void sethandle(monitor *ErrorMonitor);
+   void checkPatchList(MyList<Patch> *PatL, bool buflog);
+   void Regrid(int Symmetry, int BH_num, double **Porgbr, double **Porg0,
+               MyList<var> *OldList, MyList<var> *StateList,
+               MyList<var> *FutureList, MyList<var> *tmList, bool BB,
+               monitor *ErrorMonitor);
+   void Regrid_fake(int Symmetry, int BH_num, double **Porgbr, double **Porg0,
+                    MyList<var> *OldList, MyList<var> *StateList,
+                    MyList<var> *FutureList, MyList<var> *tmList, bool BB,
+                    monitor *ErrorMonitor);
+   void recompose_cgh(int nprocs, bool *lev_flag,
+                      MyList<var> *OldList, MyList<var> *StateList,
+                      MyList<var> *FutureList, MyList<var> *tmList,
+                      int Symmetry, bool BB);
+   void recompose_cgh_fake(int nprocs, bool *lev_flag,
+                           MyList<var> *OldList, MyList<var> *StateList,
+                           MyList<var> *FutureList, MyList<var> *tmList,
+                           int Symmetry, bool BB);
+   void read_bbox(int Symmetry, char *filename);
+   MyList<Patch> *construct_patchlist(int lev, int Symmetry);
+   bool Interp_One_Point(MyList<var> *VarList,
+                         double *XX, /*input global Cartesian coordinate*/
+                         double *Shellf, int Symmetry);
+   void recompose_cgh_Onelevel(int nprocs, int lev,
+                               MyList<var> *OldList, MyList<var> *StateList,
+                               MyList<var> *FutureList, MyList<var> *tmList,
+                               int Symmetry, bool BB);
+   void Regrid_Onelevel(int lev, int Symmetry, int BH_num, double **Porgbr, double **Porg0,
+                        MyList<var> *OldList, MyList<var> *StateList,
+                        MyList<var> *FutureList, MyList<var> *tmList, bool BB,
+                        monitor *ErrorMonitor);
+   void Regrid_Onelevel_aux(int lev, int Symmetry, int BH_num, double **Porgbr, double **Porg0,
+                            MyList<var> *OldList, MyList<var> *StateList,
+                            MyList<var> *FutureList, MyList<var> *tmList, bool BB,
+                            monitor *ErrorMonitor);
+   void settrfls(const int lev);
+
+#if (PSTR == 1 || PSTR == 2 || PSTR == 3)
+   void construct_mylev(int nprocs);
+#endif
+
+   // Load balancing support
+   bool enable_load_balance;         // Enable load balancing
+   int load_balance_check_interval;  // Check interval (in time steps)
+   int current_time_step;            // Current time step counter
+   double *rank_interp_times;        // Store interpolation times for each rank
+   int *heavy_ranks;                 // Store heavy rank numbers
+   int num_heavy_ranks;              // Number of heavy ranks
+
+   void init_load_balance(int nprocs);
+   void update_interp_time(int rank, double time);
+   bool check_and_rebalance(int nprocs, int lev,
+                           MyList<var> *OldList, MyList<var> *StateList,
+                           MyList<var> *FutureList, MyList<var> *tmList,
+                           int Symmetry, bool BB);
+};
+
+#endif /* CGH_H */

AMSS_NCKU_source/diff_new.f90

@@ -69,10 +69,12 @@
   fy = ZEO
   fz = ZEO
 
+!DIR$ SIMD VECTORLENGTHFOR(KNOWN_INTEGER=8)
+!DIR$ UNROLL PARTIAL(4)
   do k=1,ex(3)-1
   do j=1,ex(2)-1
   do i=1,ex(1)-1
-! x direction   
+! x direction
         if(i+1 <= imax .and. i-1 >= imin)then
 !
 !              - f(i-1) + f(i+1)
@@ -371,6 +373,8 @@
   fxz = ZEO
   fyz = ZEO
 
+!DIR$ SIMD VECTORLENGTHFOR(KNOWN_INTEGER=8)
+!DIR$ UNROLL PARTIAL(4)
   do k=1,ex(3)-1
   do j=1,ex(2)-1
   do i=1,ex(1)-1

AMSS_NCKU_source/fmisc.f90

@@ -883,13 +883,17 @@ subroutine symmetry_bd(ord,extc,func,funcc,SoA)
 
   integer::i
 
+!DIR$ SIMD VECTORLENGTHFOR(KNOWN_INTEGER=8)
   funcc(1:extc(1),1:extc(2),1:extc(3)) = func
+!DIR$ SIMD VECTORLENGTHFOR(KNOWN_INTEGER=8)
    do i=0,ord-1
       funcc(-i,1:extc(2),1:extc(3)) = funcc(i+1,1:extc(2),1:extc(3))*SoA(1)
    enddo
+!DIR$ SIMD VECTORLENGTHFOR(KNOWN_INTEGER=8)
    do i=0,ord-1
       funcc(:,-i,1:extc(3)) = funcc(:,i+1,1:extc(3))*SoA(2)
    enddo
+!DIR$ SIMD VECTORLENGTHFOR(KNOWN_INTEGER=8)
    do i=0,ord-1
       funcc(:,:,-i) = funcc(:,:,i+1)*SoA(3)
    enddo
@@ -1112,6 +1116,7 @@ end subroutine d2dump
 ! Lagrangian polynomial interpolation
 !------------------------------------------------------------------------------
 
+!DIR$ ATTRIBUTES FORCEINLINE :: polint
   subroutine polint(xa, ya, x, y, dy, ordn)
   implicit none
 

AMSS_NCKU_source/kodiss.f90

@@ -65,6 +65,8 @@ real*8,intent(in) :: eps
 !                       dx^4
 
 !  note the sign (-1)^r-1, now r=2
+!DIR$ SIMD VECTORLENGTHFOR(KNOWN_INTEGER=8)
+!DIR$ UNROLL PARTIAL(4)
   do k=1,ex(3)
   do j=1,ex(2)
   do i=1,ex(1)

AMSS_NCKU_source/makefile.inc

@@ -24,7 +24,7 @@ f90          = ifx
 f77          = ifx
 CXX          = icpx
 CC           = icx
-CLINKER      = mpiicpx
+CLINKER      = mpiicpx 
 
 Cu = nvcc
 CUDA_LIB_PATH = -L/usr/lib/cuda/lib64 -I/usr/include -I/usr/lib/cuda/include