Migrate build system from Intel oneAPI to AMD AOCC/AOCL/OpenMPI

Replace Intel compilers (ifx/icpx/icx) with AOCC (flang/clang++/clang),
Intel MPI (mpiicpx) with AOCC-built OpenMPI (mpicxx), and Intel MKL
with AOCL BLIS/libFLAME. Replace -xHost with -march=znver4, -ipo with
-flto, -fp-model fast=2 with -ffast-math, -qopenmp with -fopenmp.
Remove PGO, TBB allocator, and Intel-specific runtime libraries.
Fix MKL-specific includes in TwoPunctures.C and gaussj.C to use
standard CBLAS/LAPACKE headers from AOCL.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

.gitignore

@@ -1,6 +1,6 @@
 __pycache__
 GW150914
-GW150914-origin
+GW150914*
 docs
 *.tmp
-
+.codex

AMSS_NCKU_Program.py

@@ -174,11 +174,14 @@ import generate_macrodef
 generate_macrodef.generate_macrodef_h()
 print( " AMSS-NCKU macro file macrodef.h has been generated. " )
      
-generate_macrodef.generate_macrodef_fh()
-print( " AMSS-NCKU macro file macrodef.fh has been generated. " )
-
-
-##################################################################
+generate_macrodef.generate_macrodef_fh()
+print( " AMSS-NCKU macro file macrodef.fh has been generated. " )
+
+generate_macrodef.generate_build_config()
+print( " AMSS-NCKU build config AMSS_NCKU_build.mk has been generated. " )
+
+
+##################################################################
 
 # Compile the AMSS-NCKU program according to user requirements
 
@@ -217,11 +220,13 @@ shutil.copytree(AMSS_NCKU_source_path, AMSS_NCKU_source_copy)
 
 # Copy the generated macro files into the AMSS_NCKU source folder
 
-macrodef_h_path  = os.path.join(File_directory, "macrodef.h") 
-macrodef_fh_path = os.path.join(File_directory, "macrodef.fh") 
-
-shutil.copy2(macrodef_h_path,  AMSS_NCKU_source_copy)
-shutil.copy2(macrodef_fh_path, AMSS_NCKU_source_copy)
+macrodef_h_path  = os.path.join(File_directory, "macrodef.h")
+macrodef_fh_path = os.path.join(File_directory, "macrodef.fh")
+build_config_path = os.path.join(File_directory, "AMSS_NCKU_build.mk")
+
+shutil.copy2(macrodef_h_path,  AMSS_NCKU_source_copy)
+shutil.copy2(macrodef_fh_path, AMSS_NCKU_source_copy)
+shutil.copy2(build_config_path, AMSS_NCKU_source_copy)
 
 # Notes on copying files:
 # shutil.copy2 preserves file metadata such as modification time.

AMSS_NCKU_Verify_ASC26.py

@@ -2,13 +2,18 @@
 """
 AMSS-NCKU GW150914 Simulation Regression Test Script (Comprehensive Version)
 
-Verification Requirements:
-1. RMS errors < 1% for:
-   - 3D Vector Total RMS
-   - X Component RMS
-   - Y Component RMS
-   - Z Component RMS
-2. ADM constraint violation < 2 (Grid Level 0)
+Verification Requirements:
+1. RMS errors < 1% for:
+   - 3D Vector Total RMS
+   - X Component RMS
+   - Y Component RMS
+   - Z Component RMS
+2. ADM constraint violation < 2 (Grid Level 0)
+3. The following figure PDFs must match GW150914-origin exactly after rasterization:
+   - ADM_Constraint_Grid_Level_0.pdf
+   - BH_Trajectory_21_XY.pdf
+   - BH_Trajectory_XY.pdf
+   The script also reports the percentage of differing pixels for each figure.
 
 RMS Calculation Method:
 - Computes trajectory deviation on the XY plane independently for BH1 and BH2
@@ -20,9 +25,13 @@ Default: output_dir = GW150914/AMSS_NCKU_output
 Reference: GW150914-origin (baseline simulation)
 """
 
-import numpy as np
-import sys
-import os
+import numpy as np
+import sys
+import os
+import shutil
+import subprocess
+import tempfile
+from PIL import Image
 
 # ANSI Color Codes
 class Color:
@@ -49,17 +58,143 @@ def load_bh_trajectory(filepath):
     }
 
 
-def load_constraint_data(filepath):
-    """Load constraint violation data"""
-    data = []
+def load_constraint_data(filepath):
+    """Load constraint violation data"""
+    data = []
     with open(filepath, 'r') as f:
         for line in f:
             if line.startswith('#'):
                 continue
             parts = line.split()
             if len(parts) >= 8:
-                data.append([float(x) for x in parts[:8]])
-    return np.array(data)
+                data.append([float(x) for x in parts[:8]])
+    return np.array(data)
+
+
+def resolve_figure_dir(path):
+    """Resolve the sibling figure directory from an output or figure path."""
+    normalized = os.path.normpath(path)
+    if os.path.basename(normalized) == "figure":
+        return normalized
+    return os.path.join(os.path.dirname(normalized), "figure")
+
+
+def render_pdf_to_images(pdf_path, dpi=150):
+    """Render a PDF to RGB images using Ghostscript."""
+    gs_path = shutil.which("gs")
+    if gs_path is None:
+        raise RuntimeError("Ghostscript executable 'gs' was not found in PATH")
+
+    with tempfile.TemporaryDirectory(prefix="amss_verify_pdf_") as temp_dir:
+        output_pattern = os.path.join(temp_dir, "page-%03d.ppm")
+        cmd = [
+            gs_path,
+            "-q",
+            "-dSAFER",
+            "-dBATCH",
+            "-dNOPAUSE",
+            "-sDEVICE=ppmraw",
+            f"-r{dpi}",
+            f"-o{output_pattern}",
+            pdf_path
+        ]
+
+        try:
+            subprocess.run(cmd, check=True, stdout=subprocess.DEVNULL, stderr=subprocess.PIPE, text=True)
+        except subprocess.CalledProcessError as exc:
+            message = exc.stderr.strip() or str(exc)
+            raise RuntimeError(f"Failed to render PDF '{pdf_path}': {message}") from exc
+
+        ppm_files = sorted(
+            os.path.join(temp_dir, filename)
+            for filename in os.listdir(temp_dir)
+            if filename.endswith(".ppm")
+        )
+
+        if not ppm_files:
+            raise RuntimeError(f"No rendered pages were produced for '{pdf_path}'")
+
+        images = []
+        for ppm_file in ppm_files:
+            with Image.open(ppm_file) as img:
+                images.append(np.array(img.convert("RGB"), dtype=np.uint8))
+
+        return images
+
+
+def compare_rendered_pages(ref_img, target_img):
+    """Return (different_pixels, total_pixels) for two rendered RGB pages."""
+    ref_h, ref_w = ref_img.shape[:2]
+    tgt_h, tgt_w = target_img.shape[:2]
+    total_pixels = max(ref_h, tgt_h) * max(ref_w, tgt_w)
+
+    if ref_h == tgt_h and ref_w == tgt_w:
+        different_pixels = int(np.count_nonzero(np.any(ref_img != target_img, axis=2)))
+        return different_pixels, total_pixels
+
+    diff_mask = np.ones((max(ref_h, tgt_h), max(ref_w, tgt_w)), dtype=bool)
+    overlap_h = min(ref_h, tgt_h)
+    overlap_w = min(ref_w, tgt_w)
+    overlap_diff = np.any(ref_img[:overlap_h, :overlap_w] != target_img[:overlap_h, :overlap_w], axis=2)
+    diff_mask[:overlap_h, :overlap_w] = overlap_diff
+    different_pixels = int(np.count_nonzero(diff_mask))
+    return different_pixels, total_pixels
+
+
+def compare_pdf_images(ref_pdf, target_pdf, dpi=150, threshold_percent=0.001):
+    """Compare two PDFs by rasterizing them and counting differing pixels."""
+    ref_pages = render_pdf_to_images(ref_pdf, dpi=dpi)
+    target_pages = render_pdf_to_images(target_pdf, dpi=dpi)
+
+    total_pixels = 0
+    different_pixels = 0
+    max_pages = max(len(ref_pages), len(target_pages))
+
+    for page_idx in range(max_pages):
+        if page_idx < len(ref_pages) and page_idx < len(target_pages):
+            page_diff, page_total = compare_rendered_pages(ref_pages[page_idx], target_pages[page_idx])
+        else:
+            existing_page = ref_pages[page_idx] if page_idx < len(ref_pages) else target_pages[page_idx]
+            page_total = existing_page.shape[0] * existing_page.shape[1]
+            page_diff = page_total
+
+        total_pixels += page_total
+        different_pixels += page_diff
+
+    diff_percent = (different_pixels / total_pixels * 100.0) if total_pixels else 0.0
+    return {
+        "different_pixels": different_pixels,
+        "total_pixels": total_pixels,
+        "diff_percent": diff_percent,
+        "pages_ref": len(ref_pages),
+        "pages_target": len(target_pages),
+        "passed": diff_percent < threshold_percent
+    }
+
+
+def compare_required_figures(reference_figure_dir, target_figure_dir):
+    """Compare the required GW150914 figure PDFs."""
+    figure_names = [
+        "ADM_Constraint_Grid_Level_0.pdf",
+        "BH_Trajectory_21_XY.pdf",
+        "BH_Trajectory_XY.pdf"
+    ]
+
+    results = []
+    for figure_name in figure_names:
+        ref_pdf = os.path.join(reference_figure_dir, figure_name)
+        target_pdf = os.path.join(target_figure_dir, figure_name)
+
+        if not os.path.exists(ref_pdf):
+            raise FileNotFoundError(f"Reference figure not found: {ref_pdf}")
+        if not os.path.exists(target_pdf):
+            raise FileNotFoundError(f"Target figure not found: {target_pdf}")
+
+        comparison = compare_pdf_images(ref_pdf, target_pdf)
+        comparison["name"] = figure_name
+        results.append(comparison)
+
+    return results
 
 def calculate_all_rms_errors(bh_data_ref, bh_data_target):
     """
@@ -165,7 +300,7 @@ def print_rms_results(rms_dict, error, threshold=1.0):
 
     return all_passed
 
-def print_constraint_results(results, threshold=2.0):
+def print_constraint_results(results, threshold=2.0):
     print(f"\n{Color.BOLD}2. ADM Constraint Violation Analysis (Grid Level 0){Color.RESET}")
     print("-" * 65)
 
@@ -180,22 +315,49 @@ def print_constraint_results(results, threshold=2.0):
     print(f"\n   Maximum violation:  {results['max_violation']:.6f}")
     print(f"   Requirement:        < {threshold}")
     print(f"   Status:             {get_status_text(passed)}")
-
-    return passed
-
-
-def print_summary(rms_passed, constraint_passed):
-    print("\n" + Color.BLUE + Color.BOLD + "=" * 65 + Color.RESET)
-    print(Color.BOLD + "Verification Summary" + Color.RESET)
-    print(Color.BLUE + Color.BOLD + "=" * 65 + Color.RESET)
-
-    all_passed = rms_passed and constraint_passed
-    
-    res_rms = get_status_text(rms_passed)
-    res_con = get_status_text(constraint_passed)
-
-    print(f"   [1] Comprehensive RMS check:      {res_rms}")
-    print(f"   [2] ADM constraint check:         {res_con}")
+
+    return passed
+
+
+def print_figure_results(results, threshold_percent=0.001):
+    print(f"\n{Color.BOLD}3. Figure Pixel Comparison (PDF Rasterization){Color.RESET}")
+    print("-" * 65)
+    print(f"   Requirement: < {threshold_percent:.3f}% differing pixels\n")
+
+    all_passed = True
+    for result in results:
+        passed = result["passed"]
+        all_passed = all_passed and passed
+        status = get_status_text(passed)
+        print(f"   {result['name']:32}: {result['diff_percent']:10.6f}%   |   Status: {status}")
+
+        if result["pages_ref"] != result["pages_target"]:
+            print(f"   {'':32}  pages(ref/target): {result['pages_ref']}/{result['pages_target']}")
+
+    return all_passed
+
+
+def print_figure_error(error_message):
+    print(f"\n{Color.BOLD}3. Figure Pixel Comparison (PDF Rasterization){Color.RESET}")
+    print("-" * 65)
+    print(f"   {Color.RED}Error: {error_message}{Color.RESET}")
+    return False
+
+
+def print_summary(rms_passed, constraint_passed, figure_passed):
+    print("\n" + Color.BLUE + Color.BOLD + "=" * 65 + Color.RESET)
+    print(Color.BOLD + "Verification Summary" + Color.RESET)
+    print(Color.BLUE + Color.BOLD + "=" * 65 + Color.RESET)
+
+    all_passed = rms_passed and constraint_passed and figure_passed
+    
+    res_rms = get_status_text(rms_passed)
+    res_con = get_status_text(constraint_passed)
+    res_fig = get_status_text(figure_passed)
+
+    print(f"   [1] Comprehensive RMS check:      {res_rms}")
+    print(f"   [2] ADM constraint check:         {res_con}")
+    print(f"   [3] Figure pixel comparison:      {res_fig}")
     
     final_status = f"{Color.GREEN}{Color.BOLD}ALL CHECKS PASSED{Color.RESET}" if all_passed else f"{Color.RED}{Color.BOLD}SOME CHECKS FAILED{Color.RESET}"
     print(f"\n   Overall result: {final_status}")
@@ -210,12 +372,14 @@ def main():
         script_dir = os.path.dirname(os.path.abspath(__file__))
         target_dir = os.path.join(script_dir, "GW150914/AMSS_NCKU_output")
 
-    script_dir = os.path.dirname(os.path.abspath(__file__))
-    reference_dir = os.path.join(script_dir, "GW150914-origin/AMSS_NCKU_output")
-
-    bh_file_ref = os.path.join(reference_dir, "bssn_BH.dat")
-    bh_file_target = os.path.join(target_dir, "bssn_BH.dat")
-    constraint_file = os.path.join(target_dir, "bssn_constraint.dat")
+    script_dir = os.path.dirname(os.path.abspath(__file__))
+    reference_dir = os.path.join(script_dir, "GW150914-origin/AMSS_NCKU_output")
+    target_figure_dir = resolve_figure_dir(target_dir)
+    reference_figure_dir = os.path.join(script_dir, "GW150914-origin/figure")
+
+    bh_file_ref = os.path.join(reference_dir, "bssn_BH.dat")
+    bh_file_target = os.path.join(target_dir, "bssn_BH.dat")
+    constraint_file = os.path.join(target_dir, "bssn_constraint.dat")
 
     if not os.path.exists(bh_file_ref):
         print(f"{Color.RED}{Color.BOLD}Error:{Color.RESET} Baseline trajectory file not found: {bh_file_ref}")
@@ -227,9 +391,11 @@ def main():
         print(f"{Color.RED}{Color.BOLD}Error:{Color.RESET} Constraint data file not found: {constraint_file}")
         sys.exit(1)
 
-    print_header()
-    print(f"\n{Color.BOLD}Reference (Baseline):{Color.RESET} {Color.BLUE}{reference_dir}{Color.RESET}")
-    print(f"{Color.BOLD}Target (Optimized):  {Color.RESET} {Color.BLUE}{target_dir}{Color.RESET}")
+    print_header()
+    print(f"\n{Color.BOLD}Reference (Baseline):{Color.RESET} {Color.BLUE}{reference_dir}{Color.RESET}")
+    print(f"{Color.BOLD}Target (Optimized):  {Color.RESET} {Color.BLUE}{target_dir}{Color.RESET}")
+    print(f"{Color.BOLD}Reference Figures:   {Color.RESET} {Color.BLUE}{reference_figure_dir}{Color.RESET}")
+    print(f"{Color.BOLD}Target Figures:      {Color.RESET} {Color.BLUE}{target_figure_dir}{Color.RESET}")
 
     bh_data_ref = load_bh_trajectory(bh_file_ref)
     bh_data_target = load_bh_trajectory(bh_file_target)
@@ -239,12 +405,18 @@ def main():
     rms_dict, error = calculate_all_rms_errors(bh_data_ref, bh_data_target)
     rms_passed = print_rms_results(rms_dict, error)
 
-    # Output constraint results
-    constraint_results = analyze_constraint_violation(constraint_data)
-    constraint_passed = print_constraint_results(constraint_results)
-
-    all_passed = print_summary(rms_passed, constraint_passed)
-    sys.exit(0 if all_passed else 1)
+    # Output constraint results
+    constraint_results = analyze_constraint_violation(constraint_data)
+    constraint_passed = print_constraint_results(constraint_results)
+
+    try:
+        figure_results = compare_required_figures(reference_figure_dir, target_figure_dir)
+        figure_passed = print_figure_results(figure_results)
+    except (FileNotFoundError, RuntimeError) as exc:
+        figure_passed = print_figure_error(str(exc))
+
+    all_passed = print_summary(rms_passed, constraint_passed, figure_passed)
+    sys.exit(0 if all_passed else 1)
 
 if __name__ == "__main__":
     main()

AMSS_NCKU_source/Parallel.C

@@ -1,14 +1,50 @@
 
-#include "Parallel.h"
-#include "fmisc.h"
-#include "prolongrestrict.h"
-#include "misc.h"
-#include "parameters.h"
-
-int Parallel::partition1(int &nx, int split_size, int min_width, int cpusize, int shape) // special for 1 diemnsion
-{
-  nx = Mymax(1, shape / min_width);
-  nx = Mymin(cpusize, nx);
+#include "Parallel.h"
+#include "fmisc.h"
+#include "prolongrestrict.h"
+#include "misc.h"
+#include "parameters.h"
+
+namespace
+{
+enum { MAX_DATA_PACKER_VARS = 64 };
+
+int expand_var_list_pack_info(MyList<var> *src_list, MyList<var> *dst_list,
+                              int *src_sgfn, int *dst_sgfn, double **src_soa)
+{
+  int count = 0;
+  MyList<var> *src_it = src_list;
+  MyList<var> *dst_it = dst_list;
+
+  while (src_it && dst_it)
+  {
+    if (count >= MAX_DATA_PACKER_VARS)
+    {
+      cout << "Parallel::data_packer: too many variables in communication list." << endl;
+      MPI_Abort(MPI_COMM_WORLD, 1);
+    }
+    src_sgfn[count] = src_it->data->sgfn;
+    dst_sgfn[count] = dst_it->data->sgfn;
+    src_soa[count] = src_it->data->SoA;
+    count++;
+    src_it = src_it->next;
+    dst_it = dst_it->next;
+  }
+
+  if (src_it || dst_it)
+  {
+    cout << "error in short data packer, var lists does not match." << endl;
+    MPI_Abort(MPI_COMM_WORLD, 1);
+  }
+
+  return count;
+}
+}
+
+int Parallel::partition1(int &nx, int split_size, int min_width, int cpusize, int shape) // special for 1 diemnsion
+{
+  nx = Mymax(1, shape / min_width);
+  nx = Mymin(cpusize, nx);
 
   return nx;
 }
@@ -3711,11 +3747,11 @@ void Parallel::build_gstl(MyList<Parallel::gridseg> *srci, MyList<Parallel::grid
 }
 //   PACK: prepare target data in 'data'
 // UNPACK: copy target data from 'data' to corresponding numerical grids
-int Parallel::data_packer(double *data, MyList<Parallel::gridseg> *src, MyList<Parallel::gridseg> *dst, int rank_in, int dir,
-                          MyList<var> *VarLists /* source */, MyList<var> *VarListd /* target */, int Symmetry)
-{
-  int myrank;
-  MPI_Comm_rank(MPI_COMM_WORLD, &myrank);
+int Parallel::data_packer(double *data, MyList<Parallel::gridseg> *src, MyList<Parallel::gridseg> *dst, int rank_in, int dir,
+                          MyList<var> *VarLists /* source */, MyList<var> *VarListd /* target */, int Symmetry)
+{
+  int myrank;
+  MPI_Comm_rank(MPI_COMM_WORLD, &myrank);
 
   int DIM = dim;
 
@@ -3725,86 +3761,89 @@ int Parallel::data_packer(double *data, MyList<Parallel::gridseg> *src, MyList<P
     MPI_Abort(MPI_COMM_WORLD, 1);
   }
 
-  int size_out = 0;
-
-  if (!src || !dst)
-    return size_out;
-
-  MyList<var> *varls, *varld;
-
-  varls = VarLists;
-  varld = VarListd;
-  while (varls && varld)
-  {
-    varls = varls->next;
-    varld = varld->next;
-  }
-
-  if (varls || varld)
-  {
-    cout << "error in short data packer, var lists does not match." << endl;
-    MPI_Abort(MPI_COMM_WORLD, 1);
-  }
-
-  int type; /* 1 copy, 2 restrict, 3 prolong */
-  if (src->data->Bg->lev == dst->data->Bg->lev)
-    type = 1;
-  else if (src->data->Bg->lev > dst->data->Bg->lev)
-    type = 2;
-  else
-    type = 3;
-
-  while (src && dst)
-  {
-    if ((dir == PACK && dst->data->Bg->rank == rank_in && src->data->Bg->rank == myrank) ||
-        (dir == UNPACK && src->data->Bg->rank == rank_in && dst->data->Bg->rank == myrank))
-    {
-      varls = VarLists;
-      varld = VarListd;
-      while (varls && varld)
-      {
-        if (data)
-        {
-          if (dir == PACK)
-            switch (type)
-            {
-              // attention must be paied to the difference between src's llb,uub and dst's llb,uub
-            case 1:
-              f_copy(DIM, dst->data->llb, dst->data->uub, dst->data->shape, data + size_out,
-                     src->data->Bg->bbox, src->data->Bg->bbox + dim, src->data->Bg->shape, src->data->Bg->fgfs[varls->data->sgfn],
-                     dst->data->llb, dst->data->uub);
-              break;
-            case 2:
-              f_restrict3(DIM, dst->data->llb, dst->data->uub, dst->data->shape, data + size_out,
-                          src->data->Bg->bbox, src->data->Bg->bbox + dim, src->data->Bg->shape, src->data->Bg->fgfs[varls->data->sgfn],
-                          dst->data->llb, dst->data->uub, varls->data->SoA, Symmetry);
-              break;
-            case 3:
-              f_prolong3(DIM, src->data->Bg->bbox, src->data->Bg->bbox + dim, src->data->Bg->shape, src->data->Bg->fgfs[varls->data->sgfn],
-                         dst->data->llb, dst->data->uub, dst->data->shape, data + size_out,
-                         dst->data->llb, dst->data->uub, varls->data->SoA, Symmetry);
-            }
-          if (dir == UNPACK) // from target data to corresponding grid
-            f_copy(DIM, dst->data->Bg->bbox, dst->data->Bg->bbox + dim, dst->data->Bg->shape, dst->data->Bg->fgfs[varld->data->sgfn],
-                   dst->data->llb, dst->data->uub, dst->data->shape, data + size_out,
-                   dst->data->llb, dst->data->uub);
-        }
-        size_out += dst->data->shape[0] * dst->data->shape[1] * dst->data->shape[2];
-        varls = varls->next;
-        varld = varld->next;
-      }
-    }
-    dst = dst->next;
-    src = src->next;
-  }
-
+  int size_out = 0;
+
+  if (!src || !dst)
+    return size_out;
+
+  int src_sgfn[MAX_DATA_PACKER_VARS];
+  int dst_sgfn[MAX_DATA_PACKER_VARS];
+  double *src_soa[MAX_DATA_PACKER_VARS];
+  const int var_count = expand_var_list_pack_info(VarLists, VarListd, src_sgfn, dst_sgfn, src_soa);
+
+  int type; /* 1 copy, 2 restrict, 3 prolong */
+  if (src->data->Bg->lev == dst->data->Bg->lev)
+    type = 1;
+  else if (src->data->Bg->lev > dst->data->Bg->lev)
+    type = 2;
+  else
+    type = 3;
+
+  while (src && dst)
+  {
+    const bool rank_match =
+        (dir == PACK && dst->data->Bg->rank == rank_in && src->data->Bg->rank == myrank) ||
+        (dir == UNPACK && src->data->Bg->rank == rank_in && dst->data->Bg->rank == myrank);
+
+    if (rank_match)
+    {
+      const int segment_size = dst->data->shape[0] * dst->data->shape[1] * dst->data->shape[2];
+      int offset = size_out;
+
+      if (data)
+      {
+        if (dir == PACK)
+        {
+          switch (type)
+          {
+            case 1:
+              for (int iv = 0; iv < var_count; iv++, offset += segment_size)
+                f_copy(DIM, dst->data->llb, dst->data->uub, dst->data->shape, data + offset,
+                       src->data->Bg->bbox, src->data->Bg->bbox + dim, src->data->Bg->shape,
+                       src->data->Bg->fgfs[src_sgfn[iv]], dst->data->llb, dst->data->uub);
+              break;
+            case 2:
+              for (int iv = 0; iv < var_count; iv++, offset += segment_size)
+                f_restrict3(DIM, dst->data->llb, dst->data->uub, dst->data->shape, data + offset,
+                            src->data->Bg->bbox, src->data->Bg->bbox + dim, src->data->Bg->shape,
+                            src->data->Bg->fgfs[src_sgfn[iv]], dst->data->llb, dst->data->uub,
+                            src_soa[iv], Symmetry);
+              break;
+            case 3:
+              for (int iv = 0; iv < var_count; iv++, offset += segment_size)
+                f_prolong3(DIM, src->data->Bg->bbox, src->data->Bg->bbox + dim, src->data->Bg->shape,
+                           src->data->Bg->fgfs[src_sgfn[iv]], dst->data->llb, dst->data->uub,
+                           dst->data->shape, data + offset, dst->data->llb, dst->data->uub,
+                           src_soa[iv], Symmetry);
+              break;
+            default:
+              break;
+          }
+        }
+        else
+        {
+          for (int iv = 0; iv < var_count; iv++, offset += segment_size)
+            f_copy(DIM, dst->data->Bg->bbox, dst->data->Bg->bbox + dim, dst->data->Bg->shape,
+                   dst->data->Bg->fgfs[dst_sgfn[iv]], dst->data->llb, dst->data->uub,
+                   dst->data->shape, data + offset, dst->data->llb, dst->data->uub);
+        }
+      }
+
+      size_out = offset + ((!data) ? segment_size * var_count : 0);
+      if (data)
+        size_out = offset;
+    }
+    dst = dst->next;
+    src = src->next;
+  }
+
   return size_out;
 }
-int Parallel::data_packermix(double *data, MyList<Parallel::gridseg> *src, MyList<Parallel::gridseg> *dst, int rank_in, int dir,
-                             MyList<var> *VarLists /* source */, MyList<var> *VarListd /* target */, int Symmetry)
-{
-  int myrank;
-  MPI_Comm_rank(MPI_COMM_WORLD, &myrank);
+int Parallel::data_packermix(double *data, MyList<Parallel::gridseg> *src, MyList<Parallel::gridseg> *dst, int rank_in, int dir,
+                             MyList<var> *VarLists /* source */, MyList<var> *VarListd /* target */, int Symmetry)
+{
+  int myrank;
+  MPI_Comm_rank(MPI_COMM_WORLD, &myrank);
 
   int DIM = dim;
 
@@ -3814,33 +3853,22 @@ int Parallel::data_packermix(double *data, MyList<Parallel::gridseg> *src, MyLis
     MPI_Abort(MPI_COMM_WORLD, 1);
   }
 
-  int size_out = 0;
-
-  if (!src || !dst)
-    return size_out;
-
-  MyList<var> *varls, *varld;
-
-  varls = VarLists;
-  varld = VarListd;
-  while (varls && varld)
-  {
-    varls = varls->next;
-    varld = varld->next;
-  }
-
-  if (varls || varld)
-  {
-    cout << "error in short data packer, var lists does not match." << endl;
-    MPI_Abort(MPI_COMM_WORLD, 1);
-  }
-
-  int type; /* 1 copy, 2 restrict, 3 prolong */
-  if (src->data->Bg->lev == dst->data->Bg->lev)
-    type = 1;
-  else if (src->data->Bg->lev > dst->data->Bg->lev)
-    type = 2;
-  else
+  int size_out = 0;
+
+  if (!src || !dst)
+    return size_out;
+
+  int src_sgfn[MAX_DATA_PACKER_VARS];
+  int dst_sgfn[MAX_DATA_PACKER_VARS];
+  double *src_soa[MAX_DATA_PACKER_VARS];
+  const int var_count = expand_var_list_pack_info(VarLists, VarListd, src_sgfn, dst_sgfn, src_soa);
+
+  int type; /* 1 copy, 2 restrict, 3 prolong */
+  if (src->data->Bg->lev == dst->data->Bg->lev)
+    type = 1;
+  else if (src->data->Bg->lev > dst->data->Bg->lev)
+    type = 2;
+  else
     type = 3;
 
   if (type != 3)
@@ -3848,37 +3876,48 @@ int Parallel::data_packermix(double *data, MyList<Parallel::gridseg> *src, MyLis
     cout << "Parallel::data_packermix: error type " << type << " for data_packermix." << endl;
     MPI_Abort(MPI_COMM_WORLD, 1);
   }
-
-  while (src && dst)
-  {
-    if ((dir == PACK && dst->data->Bg->rank == rank_in && src->data->Bg->rank == myrank) ||
-        (dir == UNPACK && src->data->Bg->rank == rank_in && dst->data->Bg->rank == myrank))
-    {
-      varls = VarLists;
-      varld = VarListd;
-      while (varls && varld)
-      {
-        if (data)
-        {
-          if (dir == PACK)
-            f_prolongcopy3(DIM, src->data->Bg->bbox, src->data->Bg->bbox + dim, src->data->Bg->shape, src->data->Bg->fgfs[varls->data->sgfn],
-                           dst->data->llb, dst->data->uub, src->data->shape, data + size_out,
-                           src->data->llb, src->data->uub, varls->data->SoA, Symmetry);
-          if (dir == UNPACK) // from target data to corresponding grid
-            f_prolongmix3(DIM, dst->data->Bg->bbox, dst->data->Bg->bbox + dim, dst->data->Bg->shape, dst->data->Bg->fgfs[varld->data->sgfn],
-                          src->data->llb, src->data->uub, src->data->shape, data + size_out,
-                          dst->data->llb, dst->data->uub, varls->data->SoA, Symmetry, dst->data->illb, dst->data->iuub);
-        }
-        // the symmetry problem should be dealt in prolongcopy3,
-        // so we always have ghost_width for both sides
-        size_out += (src->data->shape[0] + 2 * ghost_width) * (src->data->shape[1] + 2 * ghost_width) * (src->data->shape[2] + 2 * ghost_width);
-        varls = varls->next;
-        varld = varld->next;
-      }
-    }
-    dst = dst->next;
-    src = src->next;
-  }
+
+  while (src && dst)
+  {
+    const bool rank_match =
+        (dir == PACK && dst->data->Bg->rank == rank_in && src->data->Bg->rank == myrank) ||
+        (dir == UNPACK && src->data->Bg->rank == rank_in && dst->data->Bg->rank == myrank);
+
+    if (rank_match)
+    {
+      const int segment_size =
+          (src->data->shape[0] + 2 * ghost_width) *
+          (src->data->shape[1] + 2 * ghost_width) *
+          (src->data->shape[2] + 2 * ghost_width);
+      int offset = size_out;
+
+      if (data)
+      {
+        if (dir == PACK)
+        {
+          for (int iv = 0; iv < var_count; iv++, offset += segment_size)
+            f_prolongcopy3(DIM, src->data->Bg->bbox, src->data->Bg->bbox + dim, src->data->Bg->shape,
+                           src->data->Bg->fgfs[src_sgfn[iv]], dst->data->llb, dst->data->uub,
+                           src->data->shape, data + offset, src->data->llb, src->data->uub,
+                           src_soa[iv], Symmetry);
+        }
+        else
+        {
+          for (int iv = 0; iv < var_count; iv++, offset += segment_size)
+            f_prolongmix3(DIM, dst->data->Bg->bbox, dst->data->Bg->bbox + dim, dst->data->Bg->shape,
+                          dst->data->Bg->fgfs[dst_sgfn[iv]], src->data->llb, src->data->uub,
+                          src->data->shape, data + offset, dst->data->llb, dst->data->uub,
+                          src_soa[iv], Symmetry, dst->data->illb, dst->data->iuub);
+        }
+      }
+
+      size_out = offset + ((!data) ? segment_size * var_count : 0);
+      if (data)
+        size_out = offset;
+    }
+    dst = dst->next;
+    src = src->next;
+  }
 
   return size_out;
 }

AMSS_NCKU_source/TwoPunctures.C

@@ -27,7 +27,7 @@ using namespace std;
 #endif
 
 #include "TwoPunctures.h"
-#include <mkl_cblas.h>
+#include <cblas.h>
 
 TwoPunctures::TwoPunctures(double mp, double mm, double b,
                            double P_plusx, double P_plusy, double P_plusz,

AMSS_NCKU_source/bssnEM_class.C

@@ -258,6 +258,8 @@ void bssnEM_class::Initialize()
     PhysTime = StartTime;
     Setup_Black_Hole_position();
   }
+
+  setup_transfer_caches();
 }
 
 //================================================================================================

AMSS_NCKU_source/bssnEScalar_class.C

@@ -23,8 +23,14 @@ using namespace std;
 #include "rungekutta4_rout.h"
 #include "sommerfeld_rout.h"
 #include "getnp4.h"
-#include "shellfunctions.h"
-#include "parameters.h"
+#include "shellfunctions.h"
+#include "parameters.h"
+
+#if BSSN_USE_ESCALAR_C_KERNEL
+#define BSSN_ESCALAR_RHS f_compute_rhs_bssn_escalar_c
+#else
+#define BSSN_ESCALAR_RHS f_compute_rhs_bssn_escalar
+#endif
 
 #ifdef With_AHF
 #include "derivatives.h"
@@ -74,8 +80,8 @@ bssnEScalar_class::bssnEScalar_class(double Couranti, double StartTimei, double
 
 //================================================================================================
 
-void bssnEScalar_class::Initialize()
-{
+void bssnEScalar_class::Initialize()
+{
   Sphio = new var("Sphio", ngfs++, 1, 1, 1);
   Spio = new var("Spio", ngfs++, 1, 1, 1);
   Sphi0 = new var("Sphi0", ngfs++, 1, 1, 1);
@@ -132,11 +138,14 @@ void bssnEScalar_class::Initialize()
       }
     } 
    
-  GH = new cgh(0, ngfs, Symmetry, pname, checkrun, ErrorMonitor);
-  if (checkrun)
-    CheckPoint->readcheck_cgh(PhysTime, GH, myrank, nprocs, Symmetry);
-  else
-    GH->compose_cgh(nprocs);
+  GH = new cgh(0, ngfs, Symmetry, pname, checkrun, ErrorMonitor);
+  ConstraintRefreshLevels = new int[GH->levels];
+  for (int il = 0; il < GH->levels; il++)
+    ConstraintRefreshLevels[il] = 0;
+  if (checkrun)
+    CheckPoint->readcheck_cgh(PhysTime, GH, myrank, nprocs, Symmetry);
+  else
+    GH->compose_cgh(nprocs);
 
 #ifdef WithShell
   SH = new ShellPatch(0, ngfs, pname, Symmetry, myrank, ErrorMonitor);
@@ -160,12 +169,14 @@ void bssnEScalar_class::Initialize()
   {
     CheckPoint->read_Black_Hole_position(BH_num_input, BH_num, Porg0, Pmom, Spin, Mass, Porgbr, Porg, Porg1, Porg_rhs);
   }
-  else
-  {
-    PhysTime = StartTime;
-    Setup_Black_Hole_position();
-  }
-}
+  else
+  {
+    PhysTime = StartTime;
+    Setup_Black_Hole_position();
+  }
+
+  setup_transfer_caches();
+}
 
 //================================================================================================
 
@@ -207,10 +218,10 @@ bssnEScalar_class::~bssnEScalar_class()
 
 // Read initial data solved by Ansorg, PRD 70, 064011 (2004)
 
-void bssnEScalar_class::Read_Ansorg()
-{
-  if (!checkrun)
-  {
+void bssnEScalar_class::Read_Ansorg()
+{
+  if (!checkrun)
+  {
     if (myrank == 0)
       cout << "Read initial data from Ansorg's solver,"
            << " please be sure the input parameters for black holes are puncture parameters!!" 
@@ -227,9 +238,12 @@ void bssnEScalar_class::Read_Ansorg()
         cout << "Error inputpar" << endl;
         exit(0);
       }
-    }
-    int BH_NM;
-    double *Porg_here;
+    }
+    int BH_NM;
+    double *Porg_here;
+    double *pmom_local;
+    double *spin_local;
+    double *mass_local;
     // read parameter from file
     {
       const int LEN = 256;
@@ -269,11 +283,11 @@ void bssnEScalar_class::Read_Ansorg()
       }
       inf.close();
     }
-
-    Porg_here = new double[3 * BH_NM];
-    Pmom = new double[3 * BH_NM];
-    Spin = new double[3 * BH_NM];
-    Mass = new double[BH_NM];
+
+    Porg_here = new double[3 * BH_NM];
+    pmom_local = new double[3 * BH_NM];
+    spin_local = new double[3 * BH_NM];
+    mass_local = new double[BH_NM];
     // read parameter from file
     {
       const int LEN = 256;
@@ -305,37 +319,37 @@ void bssnEScalar_class::Read_Ansorg()
         else if (status == 0)
           continue;
 
-        if (sgrp == "BSSN" && sind < BH_NM)
-        {
-          if (skey == "Mass")
-            Mass[sind] = atof(sval.c_str());
-          else if (skey == "Porgx")
-            Porg_here[sind * 3] = atof(sval.c_str());
-          else if (skey == "Porgy")
-            Porg_here[sind * 3 + 1] = atof(sval.c_str());
-          else if (skey == "Porgz")
-            Porg_here[sind * 3 + 2] = atof(sval.c_str());
-          else if (skey == "Spinx")
-            Spin[sind * 3] = atof(sval.c_str());
-          else if (skey == "Spiny")
-            Spin[sind * 3 + 1] = atof(sval.c_str());
-          else if (skey == "Spinz")
-            Spin[sind * 3 + 2] = atof(sval.c_str());
-          else if (skey == "Pmomx")
-            Pmom[sind * 3] = atof(sval.c_str());
-          else if (skey == "Pmomy")
-            Pmom[sind * 3 + 1] = atof(sval.c_str());
-          else if (skey == "Pmomz")
-            Pmom[sind * 3 + 2] = atof(sval.c_str());
-        }
-      }
-      inf.close();
+        if (sgrp == "BSSN" && sind < BH_NM)
+        {
+          if (skey == "Mass")
+            mass_local[sind] = atof(sval.c_str());
+          else if (skey == "Porgx")
+            Porg_here[sind * 3] = atof(sval.c_str());
+          else if (skey == "Porgy")
+            Porg_here[sind * 3 + 1] = atof(sval.c_str());
+          else if (skey == "Porgz")
+            Porg_here[sind * 3 + 2] = atof(sval.c_str());
+          else if (skey == "Spinx")
+            spin_local[sind * 3] = atof(sval.c_str());
+          else if (skey == "Spiny")
+            spin_local[sind * 3 + 1] = atof(sval.c_str());
+          else if (skey == "Spinz")
+            spin_local[sind * 3 + 2] = atof(sval.c_str());
+          else if (skey == "Pmomx")
+            pmom_local[sind * 3] = atof(sval.c_str());
+          else if (skey == "Pmomy")
+            pmom_local[sind * 3 + 1] = atof(sval.c_str());
+          else if (skey == "Pmomz")
+            pmom_local[sind * 3 + 2] = atof(sval.c_str());
+        }
+      }
+      inf.close();
     }
-    int order = 6;
-    Ansorg read_ansorg("Ansorg.psid", order);
-    // set initial data
-    for (int lev = 0; lev < GH->levels; lev++)
-    {
+    int order = 6;
+    Ansorg read_ansorg("Ansorg.psid", order);
+    // set initial data
+    for (int lev = 0; lev < GH->levels; lev++)
+    {
       MyList<Patch> *Pp = GH->PatL[lev];
       while (Pp)
       {
@@ -358,21 +372,21 @@ void bssnEScalar_class::Read_Ansorg()
                                       cg->fgfs[Axx0->sgfn], cg->fgfs[Axy0->sgfn], cg->fgfs[Axz0->sgfn], 
                                       cg->fgfs[Ayy0->sgfn], cg->fgfs[Ayz0->sgfn], cg->fgfs[Azz0->sgfn],
                                       cg->fgfs[Gmx0->sgfn], cg->fgfs[Gmy0->sgfn], cg->fgfs[Gmz0->sgfn],
-                                      cg->fgfs[Lap0->sgfn], 
-                                      cg->fgfs[Sfx0->sgfn], cg->fgfs[Sfy0->sgfn], cg->fgfs[Sfz0->sgfn],
-                                      cg->fgfs[dtSfx0->sgfn], cg->fgfs[dtSfy0->sgfn], cg->fgfs[dtSfz0->sgfn],
-                                      cg->fgfs[Sphi0->sgfn], cg->fgfs[Spi0->sgfn],
-                                      Mass, Porg_here, Pmom, Spin, BH_NM);
+                                      cg->fgfs[Lap0->sgfn], 
+                                      cg->fgfs[Sfx0->sgfn], cg->fgfs[Sfy0->sgfn], cg->fgfs[Sfz0->sgfn],
+                                      cg->fgfs[dtSfx0->sgfn], cg->fgfs[dtSfy0->sgfn], cg->fgfs[dtSfz0->sgfn],
+                                      cg->fgfs[Sphi0->sgfn], cg->fgfs[Spi0->sgfn],
+                                      mass_local, Porg_here, pmom_local, spin_local, BH_NM);
           }
           if (BL == Pp->data->ble)
             break;
           BL = BL->next;
-        }
-        Pp = Pp->next;
-      }
-    }
-#ifdef WithShell
-    // ShellPatch part
+        }
+        Pp = Pp->next;
+      }
+    }
+#ifdef WithShell
+    // ShellPatch part
     MyList<ss_patch> *Pp = SH->PatL;
     while (Pp)
     {
@@ -400,25 +414,28 @@ void bssnEScalar_class::Read_Ansorg()
                                        cg->fgfs[Axx0->sgfn], cg->fgfs[Axy0->sgfn], cg->fgfs[Axz0->sgfn], 
                                        cg->fgfs[Ayy0->sgfn], cg->fgfs[Ayz0->sgfn], cg->fgfs[Azz0->sgfn],
                                        cg->fgfs[Gmx0->sgfn], cg->fgfs[Gmy0->sgfn], cg->fgfs[Gmz0->sgfn],
-                                       cg->fgfs[Lap0->sgfn], 
-                                       cg->fgfs[Sfx0->sgfn], cg->fgfs[Sfy0->sgfn], cg->fgfs[Sfz0->sgfn],
-                                       cg->fgfs[dtSfx0->sgfn], cg->fgfs[dtSfy0->sgfn], cg->fgfs[dtSfz0->sgfn],
-                                       cg->fgfs[Sphi0->sgfn], cg->fgfs[Spi0->sgfn],
-                                       Mass, Porg_here, Pmom, Spin, BH_NM);
+                                       cg->fgfs[Lap0->sgfn], 
+                                       cg->fgfs[Sfx0->sgfn], cg->fgfs[Sfy0->sgfn], cg->fgfs[Sfz0->sgfn],
+                                       cg->fgfs[dtSfx0->sgfn], cg->fgfs[dtSfy0->sgfn], cg->fgfs[dtSfz0->sgfn],
+                                       cg->fgfs[Sphi0->sgfn], cg->fgfs[Spi0->sgfn],
+                                       mass_local, Porg_here, pmom_local, spin_local, BH_NM);
         }
         if (BL == Pp->data->ble)
           break;
         BL = BL->next;
-      }
-      Pp = Pp->next;
-    }
-#endif
-
-    delete[] Porg_here;
-    // dump read_in initial data
-    //   for(int lev=0;lev<GH->levels;lev++) Parallel::Dump_Data(GH->PatL[lev],StateList,0,PhysTime,dT);
-  }
-}
+      }
+      Pp = Pp->next;
+    }
+#endif
+
+    delete[] Porg_here;
+    delete[] pmom_local;
+    delete[] spin_local;
+    delete[] mass_local;
+    // dump read_in initial data
+    //   for(int lev=0;lev<GH->levels;lev++) Parallel::Dump_Data(GH->PatL[lev],StateList,0,PhysTime,dT);
+  }
+}
 
 //================================================================================================
 
@@ -432,10 +449,10 @@ void bssnEScalar_class::Read_Ansorg()
 
 // Read initial data solved by Pablo's Olliptic Phys.Rev.D 82 024005 (2010)
 
-void bssnEScalar_class::Read_Pablo()
-{
-  if (!checkrun)
-  {
+void bssnEScalar_class::Read_Pablo()
+{
+  if (!checkrun)
+  {
     if (myrank == 0)
       cout << "Read initial data from Pablo's solver,"
            << " please be sure the input parameters for black holes are puncture parameters!!" 
@@ -452,9 +469,12 @@ void bssnEScalar_class::Read_Pablo()
         cout << "Error inputpar" << endl;
         exit(0);
       }
-    }
-    int BH_NM;
-    double *Porg_here;
+    }
+    int BH_NM;
+    double *Porg_here;
+    double *pmom_local;
+    double *spin_local;
+    double *mass_local;
     // read parameter from file
     {
       const int LEN = 256;
@@ -494,11 +514,11 @@ void bssnEScalar_class::Read_Pablo()
       }
       inf.close();
     }
-
-    Porg_here = new double[3 * BH_NM];
-    Pmom = new double[3 * BH_NM];
-    Spin = new double[3 * BH_NM];
-    Mass = new double[BH_NM];
+
+    Porg_here = new double[3 * BH_NM];
+    pmom_local = new double[3 * BH_NM];
+    spin_local = new double[3 * BH_NM];
+    mass_local = new double[BH_NM];
     // read parameter from file
     {
       const int LEN = 256;
@@ -530,31 +550,31 @@ void bssnEScalar_class::Read_Pablo()
         else if (status == 0)
           continue;
 
-        if (sgrp == "BSSN" && sind < BH_NM)
-        {
-          if (skey == "Mass")
-            Mass[sind] = atof(sval.c_str());
-          else if (skey == "Porgx")
-            Porg_here[sind * 3] = atof(sval.c_str());
-          else if (skey == "Porgy")
-            Porg_here[sind * 3 + 1] = atof(sval.c_str());
-          else if (skey == "Porgz")
-            Porg_here[sind * 3 + 2] = atof(sval.c_str());
-          else if (skey == "Spinx")
-            Spin[sind * 3] = atof(sval.c_str());
-          else if (skey == "Spiny")
-            Spin[sind * 3 + 1] = atof(sval.c_str());
-          else if (skey == "Spinz")
-            Spin[sind * 3 + 2] = atof(sval.c_str());
-          else if (skey == "Pmomx")
-            Pmom[sind * 3] = atof(sval.c_str());
-          else if (skey == "Pmomy")
-            Pmom[sind * 3 + 1] = atof(sval.c_str());
-          else if (skey == "Pmomz")
-            Pmom[sind * 3 + 2] = atof(sval.c_str());
-        }
-      }
-      inf.close();
+        if (sgrp == "BSSN" && sind < BH_NM)
+        {
+          if (skey == "Mass")
+            mass_local[sind] = atof(sval.c_str());
+          else if (skey == "Porgx")
+            Porg_here[sind * 3] = atof(sval.c_str());
+          else if (skey == "Porgy")
+            Porg_here[sind * 3 + 1] = atof(sval.c_str());
+          else if (skey == "Porgz")
+            Porg_here[sind * 3 + 2] = atof(sval.c_str());
+          else if (skey == "Spinx")
+            spin_local[sind * 3] = atof(sval.c_str());
+          else if (skey == "Spiny")
+            spin_local[sind * 3 + 1] = atof(sval.c_str());
+          else if (skey == "Spinz")
+            spin_local[sind * 3 + 2] = atof(sval.c_str());
+          else if (skey == "Pmomx")
+            pmom_local[sind * 3] = atof(sval.c_str());
+          else if (skey == "Pmomy")
+            pmom_local[sind * 3 + 1] = atof(sval.c_str());
+          else if (skey == "Pmomz")
+            pmom_local[sind * 3 + 2] = atof(sval.c_str());
+        }
+      }
+      inf.close();
     }
     bool flag = false;
     int DIM = dim;
@@ -594,11 +614,11 @@ void bssnEScalar_class::Read_Pablo()
                                         cg->fgfs[Axx0->sgfn], cg->fgfs[Axy0->sgfn], cg->fgfs[Axz0->sgfn], 
                                         cg->fgfs[Ayy0->sgfn], cg->fgfs[Ayz0->sgfn], cg->fgfs[Azz0->sgfn],
                                         cg->fgfs[Gmx0->sgfn], cg->fgfs[Gmy0->sgfn], cg->fgfs[Gmz0->sgfn],
-                                        cg->fgfs[Lap0->sgfn], 
-                                        cg->fgfs[Sfx0->sgfn], cg->fgfs[Sfy0->sgfn], cg->fgfs[Sfz0->sgfn],
-                                        cg->fgfs[dtSfx0->sgfn], cg->fgfs[dtSfy0->sgfn], cg->fgfs[dtSfz0->sgfn],
-                                        cg->fgfs[Sphi0->sgfn], cg->fgfs[Spi0->sgfn],
-                                        Mass, Porg_here, Pmom, Spin, BH_NM);
+                                        cg->fgfs[Lap0->sgfn], 
+                                        cg->fgfs[Sfx0->sgfn], cg->fgfs[Sfy0->sgfn], cg->fgfs[Sfz0->sgfn],
+                                        cg->fgfs[dtSfx0->sgfn], cg->fgfs[dtSfy0->sgfn], cg->fgfs[dtSfz0->sgfn],
+                                        cg->fgfs[Sphi0->sgfn], cg->fgfs[Spi0->sgfn],
+                                        mass_local, Porg_here, pmom_local, spin_local, BH_NM);
             }
             if (BL == Pp->data->ble)
               break;
@@ -658,11 +678,11 @@ void bssnEScalar_class::Read_Pablo()
                                          cg->fgfs[Axx0->sgfn], cg->fgfs[Axy0->sgfn], cg->fgfs[Axz0->sgfn], 
                                          cg->fgfs[Ayy0->sgfn], cg->fgfs[Ayz0->sgfn], cg->fgfs[Azz0->sgfn],
                                          cg->fgfs[Gmx0->sgfn], cg->fgfs[Gmy0->sgfn], cg->fgfs[Gmz0->sgfn],
-                                         cg->fgfs[Lap0->sgfn], 
-                                         cg->fgfs[Sfx0->sgfn], cg->fgfs[Sfy0->sgfn], cg->fgfs[Sfz0->sgfn],
-                                         cg->fgfs[dtSfx0->sgfn], cg->fgfs[dtSfy0->sgfn], cg->fgfs[dtSfz0->sgfn],
-                                         cg->fgfs[Sphi0->sgfn], cg->fgfs[Spi0->sgfn],
-                                         Mass, Porg_here, Pmom, Spin, BH_NM);
+                                         cg->fgfs[Lap0->sgfn], 
+                                         cg->fgfs[Sfx0->sgfn], cg->fgfs[Sfy0->sgfn], cg->fgfs[Sfz0->sgfn],
+                                         cg->fgfs[dtSfx0->sgfn], cg->fgfs[dtSfy0->sgfn], cg->fgfs[dtSfz0->sgfn],
+                                         cg->fgfs[Sphi0->sgfn], cg->fgfs[Spi0->sgfn],
+                                         mass_local, Porg_here, pmom_local, spin_local, BH_NM);
           }
           if (BL == Pp->data->ble)
             break;
@@ -684,10 +704,13 @@ void bssnEScalar_class::Read_Pablo()
       Pp = Pp->next;
     }
 #endif
-
-    delete[] Porg_here;
-    if (flag && myrank == 0)
-      MPI_Abort(MPI_COMM_WORLD, 1);
+
+    delete[] Porg_here;
+    delete[] pmom_local;
+    delete[] spin_local;
+    delete[] mass_local;
+    if (flag && myrank == 0)
+      MPI_Abort(MPI_COMM_WORLD, 1);
     // dump read_in initial data
     for (int lev = 0; lev < GH->levels; lev++)
       Parallel::Dump_Data(GH->PatL[lev], StateList, 0, PhysTime, dT);
@@ -739,10 +762,10 @@ void bssnEScalar_class::Step(int lev, int YN)
                      cg->fgfs[Ayy0->sgfn], cg->fgfs[Ayz0->sgfn], cg->fgfs[Azz0->sgfn]);
 #endif
 
-        if (f_compute_rhs_bssn_escalar(cg->shape, TRK4, cg->X[0], cg->X[1], cg->X[2],
-                                       cg->fgfs[phi0->sgfn], cg->fgfs[trK0->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],
+        if (BSSN_ESCALAR_RHS(cg->shape, TRK4, cg->X[0], cg->X[1], cg->X[2],
+                                       cg->fgfs[phi0->sgfn], cg->fgfs[trK0->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[Ayy0->sgfn], cg->fgfs[Ayz0->sgfn], cg->fgfs[Azz0->sgfn],
                                        cg->fgfs[Gmx0->sgfn], cg->fgfs[Gmy0->sgfn], cg->fgfs[Gmz0->sgfn],
@@ -993,11 +1016,12 @@ void bssnEScalar_class::Step(int lev, int YN)
   }
 #endif
 
-  Parallel::Sync(GH->PatL[lev], SynchList_pre, Symmetry);
+  Parallel::AsyncSyncState async_pre;
+  sync_predictor_start(lev, SynchList_pre, async_pre);
 
-#ifdef WithShell
-  if (lev == 0)
-  {
+#ifdef WithShell
+  if (lev == 0)
+  {
     clock_t prev_clock, curr_clock;
     if (myrank == 0)
       curr_clock = clock();
@@ -1009,9 +1033,10 @@ void bssnEScalar_class::Step(int lev, int YN)
       cout << " Shell stuff synchronization used " 
            << (double)(curr_clock - prev_clock) / ((double)CLOCKS_PER_SEC) 
            << " seconds! " << endl;
-    }
-  }
-#endif
+    }
+  }
+#endif
+  sync_predictor_finish(lev, async_pre, SynchList_pre);
 
   // for black hole position
   if (BH_num > 0 && lev == GH->levels - 1)
@@ -1081,10 +1106,10 @@ void bssnEScalar_class::Step(int lev, int YN)
                          cg->fgfs[Ayy->sgfn], cg->fgfs[Ayz->sgfn], cg->fgfs[Azz->sgfn]);
 #endif
 
-          if (f_compute_rhs_bssn_escalar(cg->shape, TRK4, cg->X[0], cg->X[1], cg->X[2],
-                                         cg->fgfs[phi->sgfn], cg->fgfs[trK->sgfn],
-                                         cg->fgfs[gxx->sgfn], cg->fgfs[gxy->sgfn], cg->fgfs[gxz->sgfn], 
-                                         cg->fgfs[gyy->sgfn], cg->fgfs[gyz->sgfn], cg->fgfs[gzz->sgfn],
+          if (BSSN_ESCALAR_RHS(cg->shape, TRK4, cg->X[0], cg->X[1], cg->X[2],
+                                         cg->fgfs[phi->sgfn], cg->fgfs[trK->sgfn],
+                                         cg->fgfs[gxx->sgfn], cg->fgfs[gxy->sgfn], cg->fgfs[gxz->sgfn], 
+                                         cg->fgfs[gyy->sgfn], cg->fgfs[gyz->sgfn], cg->fgfs[gzz->sgfn],
                                          cg->fgfs[Axx->sgfn], cg->fgfs[Axy->sgfn], cg->fgfs[Axz->sgfn], 
                                          cg->fgfs[Ayy->sgfn], cg->fgfs[Ayz->sgfn], cg->fgfs[Azz->sgfn],
                                          cg->fgfs[Gmx->sgfn], cg->fgfs[Gmy->sgfn], cg->fgfs[Gmz->sgfn],
@@ -1349,11 +1374,12 @@ void bssnEScalar_class::Step(int lev, int YN)
     }
 #endif
 
-    Parallel::Sync(GH->PatL[lev], SynchList_cor, Symmetry);
+    Parallel::AsyncSyncState async_cor;
+    sync_corrector_start(lev, SynchList_cor, async_cor);
 
-#ifdef WithShell
-    if (lev == 0)
-    {
+#ifdef WithShell
+    if (lev == 0)
+    {
       clock_t prev_clock, curr_clock;
       if (myrank == 0)
         curr_clock = clock();
@@ -1365,9 +1391,10 @@ void bssnEScalar_class::Step(int lev, int YN)
         cout << " Shell stuff synchronization used " 
              << (double)(curr_clock - prev_clock) / ((double)CLOCKS_PER_SEC) 
              << " seconds! " << endl;
-      }
-    }
-#endif
+      }
+    }
+#endif
+    sync_corrector_finish(lev, async_cor, SynchList_cor);
     // for black hole position
     if (BH_num > 0 && lev == GH->levels - 1)
     {
@@ -1835,11 +1862,14 @@ void bssnEScalar_class::AnalysisStuff_EScalar(int lev, double dT_lev)
 
 //================================================================================================
 
-void bssnEScalar_class::Interp_Constraint()
-{
-  // we do not support a_lev != 0 yet.
-  if (a_lev > 0)
-    return;
+void bssnEScalar_class::Interp_Constraint(bool infg)
+{
+  if (!infg)
+    return;
+
+  // we do not support a_lev != 0 yet.
+  if (a_lev > 0)
+    return;
 
   for (int lev = 0; lev < GH->levels; lev++)
   {
@@ -1858,10 +1888,10 @@ void bssnEScalar_class::Interp_Constraint()
           if (myrank == cg->rank)
           {
             if (lev > 0)
-              f_compute_rhs_bssn_escalar(cg->shape, TRK4, cg->X[0], cg->X[1], cg->X[2],
-                                         cg->fgfs[phi0->sgfn], cg->fgfs[trK0->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],
+              BSSN_ESCALAR_RHS(cg->shape, TRK4, cg->X[0], cg->X[1], cg->X[2],
+                                         cg->fgfs[phi0->sgfn], cg->fgfs[trK0->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[Ayy0->sgfn], cg->fgfs[Ayz0->sgfn], cg->fgfs[Azz0->sgfn],
                                          cg->fgfs[Gmx0->sgfn], cg->fgfs[Gmy0->sgfn], cg->fgfs[Gmz0->sgfn],
@@ -2078,10 +2108,10 @@ void bssnEScalar_class::Constraint_Out()
             if (myrank == cg->rank)
             {
               if (lev > 0)
-                f_compute_rhs_bssn_escalar(cg->shape, TRK4, cg->X[0], cg->X[1], cg->X[2],
-                                           cg->fgfs[phi0->sgfn], cg->fgfs[trK0->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],
+                BSSN_ESCALAR_RHS(cg->shape, TRK4, cg->X[0], cg->X[1], cg->X[2],
+                                           cg->fgfs[phi0->sgfn], cg->fgfs[trK0->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[Ayy0->sgfn], cg->fgfs[Ayz0->sgfn], cg->fgfs[Azz0->sgfn],
                                            cg->fgfs[Gmx0->sgfn], cg->fgfs[Gmy0->sgfn], cg->fgfs[Gmz0->sgfn],

AMSS_NCKU_source/bssnEScalar_class.h

@@ -51,7 +51,7 @@ public:
      void Compute_Psi4(int lev);
      void Step(int lev, int YN);
      void AnalysisStuff_EScalar(int lev, double dT_lev);
-     void Interp_Constraint();
+     void Interp_Constraint(bool infg);
      void Constraint_Out(); 
 
 protected:

AMSS_NCKU_source/bssn_class.h

@@ -31,11 +31,19 @@ using namespace std;
 #include "surface_integral.h"
 #include "checkpoint.h"
 
-extern void setpbh(int iBHN, double **iPBH, double *iMass, int rBHN);
-
-class bssn_class
-{
-public:
+extern void setpbh(int iBHN, double **iPBH, double *iMass, int rBHN);
+
+#ifndef BSSN_USE_TRANSFER_CACHE
+#define BSSN_USE_TRANSFER_CACHE 1
+#endif
+
+#ifndef BSSN_USE_ESCALAR_C_KERNEL
+#define BSSN_USE_ESCALAR_C_KERNEL 1
+#endif
+
+class bssn_class
+{
+public:
        int ngfs;
        int nprocs, myrank;
        cgh *GH;
@@ -167,14 +175,25 @@ public:
        void Setup_KerrSchild();
        void Enforce_algcon(int lev, int fg);
 
-       void testRestrict();
-       void testOutBd();
-       
-       bool check_Stdin_Abort(); 
-
-       virtual void Setup_Initial_Data_Cao();
-       virtual void Setup_Initial_Data_Lousto();
-       virtual void Initialize();
+       void testRestrict();
+       void testOutBd();
+       
+       bool check_Stdin_Abort(); 
+       bool use_transfer_cache() const;
+       void setup_transfer_caches();
+       void invalidate_transfer_caches();
+       void destroy_transfer_caches();
+       void sync_predictor_start(int lev, MyList<var> *VarList, Parallel::AsyncSyncState &async_state);
+       void sync_predictor_finish(int lev, Parallel::AsyncSyncState &async_state, MyList<var> *VarList);
+       void sync_corrector_start(int lev, MyList<var> *VarList, Parallel::AsyncSyncState &async_state);
+       void sync_corrector_finish(int lev, Parallel::AsyncSyncState &async_state, MyList<var> *VarList);
+       void sync_evolution(int lev, MyList<var> *VarList, Parallel::SyncCache *cache_array = 0);
+       void restrict_evolution(int lev, MyList<var> *src_var_list, MyList<var> *dst_var_list);
+       void outbdlow2hi_evolution(int lev, MyList<var> *src_var_list, MyList<var> *dst_var_list);
+
+       virtual void Setup_Initial_Data_Cao();
+       virtual void Setup_Initial_Data_Lousto();
+       virtual void Initialize();
        virtual void Read_Ansorg();
        virtual void Read_Pablo() {};
        virtual void Compute_Psi4(int lev);

AMSS_NCKU_source/bssn_escalar_rhs_c.C

@@ -0,0 +1,169 @@
+#include "macrodef.h"
+#include "bssn_rhs.h"
+#include "share_func.h"
+#include "tool.h"
+#include <vector>
+
+namespace
+{
+    // Reuse the temporary workspace across block calls to avoid repeated heap churn
+    // in the EScalar wrapper. MPI ranks execute this path sequentially, so a single
+    // process-local buffer is sufficient here.
+    std::vector<double> g_escalar_tmp_store;
+}
+
+#ifdef fortran1
+#define f_frpotential frpotential
+#endif
+#ifdef fortran2
+#define f_frpotential FRPOTENTIAL
+#endif
+#ifdef fortran3
+#define f_frpotential frpotential_
+#endif
+
+extern "C"
+{
+    void f_frpotential(int *, double *, double *, double *);
+}
+
+int f_compute_rhs_bssn_escalar_c(int *ex, double &T,
+                                 double *X, double *Y, double *Z,
+                                 double *chi, double *trK,
+                                 double *dxx, double *gxy, double *gxz, double *dyy, double *gyz, double *dzz,
+                                 double *Axx, double *Axy, double *Axz, double *Ayy, double *Ayz, double *Azz,
+                                 double *Gamx, double *Gamy, double *Gamz,
+                                 double *Lap, double *betax, double *betay, double *betaz,
+                                 double *dtSfx, double *dtSfy, double *dtSfz,
+                                 double *Sphi, double *Spi,
+                                 double *chi_rhs, double *trK_rhs,
+                                 double *gxx_rhs, double *gxy_rhs, double *gxz_rhs, double *gyy_rhs, double *gyz_rhs, double *gzz_rhs,
+                                 double *Axx_rhs, double *Axy_rhs, double *Axz_rhs, double *Ayy_rhs, double *Ayz_rhs, double *Azz_rhs,
+                                 double *Gamx_rhs, double *Gamy_rhs, double *Gamz_rhs,
+                                 double *Lap_rhs, double *betax_rhs, double *betay_rhs, double *betaz_rhs,
+                                 double *dtSfx_rhs, double *dtSfy_rhs, double *dtSfz_rhs,
+                                 double *Sphi_rhs, double *Spi_rhs,
+                                 double *rho, double *Sx, double *Sy, double *Sz,
+                                 double *Sxx, double *Sxy, double *Sxz, double *Syy, double *Syz, double *Szz,
+                                 double *Gamxxx, double *Gamxxy, double *Gamxxz, double *Gamxyy, double *Gamxyz, double *Gamxzz,
+                                 double *Gamyxx, double *Gamyxy, double *Gamyxz, double *Gamyyy, double *Gamyyz, double *Gamyzz,
+                                 double *Gamzxx, double *Gamzxy, double *Gamzxz, double *Gamzyy, double *Gamzyz, double *Gamzzz,
+                                 double *Rxx, double *Rxy, double *Rxz, double *Ryy, double *Ryz, double *Rzz,
+                                 double *ham_Res, double *movx_Res, double *movy_Res, double *movz_Res,
+                                 double *Gmx_Res, double *Gmy_Res, double *Gmz_Res,
+                                 int &Symmetry, int &Lev, double &eps, int &co)
+{
+    const int nx = ex[0], ny = ex[1], nz = ex[2];
+    const int all = nx * ny * nz;
+
+    const size_t workspace_size = size_t(all) * 17;
+    if (g_escalar_tmp_store.size() < workspace_size)
+        g_escalar_tmp_store.resize(workspace_size);
+
+    double *tmp_ptr = g_escalar_tmp_store.data();
+    auto alloc_tmp = [&](int n = 1) -> double *
+    {
+        double *ptr = tmp_ptr;
+        tmp_ptr += size_t(all) * n;
+        return ptr;
+    };
+
+    double *chix = alloc_tmp(), *chiy = alloc_tmp(), *chiz = alloc_tmp();
+    double *Kx = alloc_tmp(), *Ky = alloc_tmp(), *Kz = alloc_tmp();
+    double *fxx = alloc_tmp(), *fxy = alloc_tmp(), *fxz = alloc_tmp();
+    double *fyy = alloc_tmp(), *fyz = alloc_tmp(), *fzz = alloc_tmp();
+    double *Lapx = alloc_tmp(), *Lapy = alloc_tmp(), *Lapz = alloc_tmp();
+    double *V = alloc_tmp(), *dVdSphi = alloc_tmp();
+
+    const double ZEO = 0.0, ONE = 1.0, TWO = 2.0, HALF = 0.5;
+    const double SSS[3] = {1.0, 1.0, 1.0};
+
+    fderivs(ex, chi, chix, chiy, chiz, X, Y, Z, 1.0, 1.0, 1.0, Symmetry, Lev);
+    fderivs(ex, Lap, Lapx, Lapy, Lapz, X, Y, Z, 1.0, 1.0, 1.0, Symmetry, Lev);
+    fderivs(ex, Sphi, Kx, Ky, Kz, X, Y, Z, 1.0, 1.0, 1.0, Symmetry, Lev);
+    fdderivs(ex, Sphi, fxx, fxy, fxz, fyy, fyz, fzz, X, Y, Z, 1.0, 1.0, 1.0, Symmetry, Lev);
+
+    f_frpotential(ex, Sphi, V, dVdSphi);
+
+    for (int i = 0; i < all; ++i)
+    {
+        const double alpn1 = Lap[i] + ONE;
+        const double chin1 = chi[i] + ONE;
+        const double gxx = dxx[i] + ONE;
+        const double gyy = dyy[i] + ONE;
+        const double gzz = dzz[i] + ONE;
+        const double det = gxx * gyy * gzz + gxy[i] * gyz[i] * gxz[i] + gxz[i] * gxy[i] * gyz[i]
+                         - gxz[i] * gyy * gxz[i] - gxy[i] * gxy[i] * gzz - gxx * gyz[i] * gyz[i];
+        const double gupxx = (gyy * gzz - gyz[i] * gyz[i]) / det;
+        const double gupxy = -(gxy[i] * gzz - gyz[i] * gxz[i]) / det;
+        const double gupxz = (gxy[i] * gyz[i] - gyy * gxz[i]) / det;
+        const double gupyy = (gxx * gzz - gxz[i] * gxz[i]) / det;
+        const double gupyz = -(gxx * gyz[i] - gxy[i] * gxz[i]) / det;
+        const double gupzz = (gxx * gyy - gxy[i] * gxy[i]) / det;
+
+        Sphi_rhs[i] = alpn1 * Spi[i];
+
+        Spi_rhs[i] = gupxx * fxx[i] + gupyy * fyy[i] + gupzz * fzz[i]
+                   + TWO * (gupxy * fxy[i] + gupxz * fxz[i] + gupyz * fyz[i])
+                   - ((Gamx[i] + (gupxx * chix[i] + gupxy * chiy[i] + gupxz * chiz[i]) / TWO / chin1) * Kx[i]
+                   +  (Gamy[i] + (gupxy * chix[i] + gupyy * chiy[i] + gupyz * chiz[i]) / TWO / chin1) * Ky[i]
+                   +  (Gamz[i] + (gupxz * chix[i] + gupyz * chiy[i] + gupzz * chiz[i]) / TWO / chin1) * Kz[i]);
+
+        Spi_rhs[i] = Spi_rhs[i] * alpn1
+                   + gupxx * Lapx[i] * Kx[i] + gupxy * Lapx[i] * Ky[i] + gupxz * Lapx[i] * Kz[i]
+                   + gupxy * Lapy[i] * Kx[i] + gupyy * Lapy[i] * Ky[i] + gupyz * Lapy[i] * Kz[i]
+                   + gupxz * Lapz[i] * Kx[i] + gupyz * Lapz[i] * Ky[i] + gupzz * Lapz[i] * Kz[i];
+
+        Spi_rhs[i] = Spi_rhs[i] * chin1 + alpn1 * (trK[i] * Spi[i] - dVdSphi[i]);
+
+        rho[i] = chin1 * ((gupxx * Kx[i] * Kx[i] + gupyy * Ky[i] * Ky[i] + gupzz * Kz[i] * Kz[i]) * HALF
+               + gupxy * Kx[i] * Ky[i] + gupxz * Kx[i] * Kz[i] + gupyz * Ky[i] * Kz[i])
+               + Spi[i] * Spi[i] * HALF + V[i];
+        Sx[i] = -Spi[i] * Kx[i];
+        Sy[i] = -Spi[i] * Ky[i];
+        Sz[i] = -Spi[i] * Kz[i];
+
+        const double pressure = (rho[i] - Spi[i] * Spi[i]) / chin1;
+        Sxx[i] = Kx[i] * Kx[i] - pressure * gxx;
+        Sxy[i] = Kx[i] * Ky[i] - pressure * gxy[i];
+        Sxz[i] = Kx[i] * Kz[i] - pressure * gxz[i];
+        Syy[i] = Ky[i] * Ky[i] - pressure * gyy;
+        Syz[i] = Ky[i] * Kz[i] - pressure * gyz[i];
+        Szz[i] = Kz[i] * Kz[i] - pressure * gzz;
+    }
+
+    if (f_compute_rhs_bssn(ex, T, X, Y, Z,
+                           chi, trK,
+                           dxx, gxy, gxz, dyy, gyz, dzz,
+                           Axx, Axy, Axz, Ayy, Ayz, Azz,
+                           Gamx, Gamy, Gamz,
+                           Lap, betax, betay, betaz,
+                           dtSfx, dtSfy, dtSfz,
+                           chi_rhs, trK_rhs,
+                           gxx_rhs, gxy_rhs, gxz_rhs, gyy_rhs, gyz_rhs, gzz_rhs,
+                           Axx_rhs, Axy_rhs, Axz_rhs, Ayy_rhs, Ayz_rhs, Azz_rhs,
+                           Gamx_rhs, Gamy_rhs, Gamz_rhs,
+                           Lap_rhs, betax_rhs, betay_rhs, betaz_rhs,
+                           dtSfx_rhs, dtSfy_rhs, dtSfz_rhs,
+                           rho, Sx, Sy, Sz,
+                           Sxx, Sxy, Sxz, Syy, Syz, Szz,
+                           Gamxxx, Gamxxy, Gamxxz, Gamxyy, Gamxyz, Gamxzz,
+                           Gamyxx, Gamyxy, Gamyxz, Gamyyy, Gamyyz, Gamyzz,
+                           Gamzxx, Gamzxy, Gamzxz, Gamzyy, Gamzyz, Gamzzz,
+                           Rxx, Rxy, Rxz, Ryy, Ryz, Rzz,
+                           ham_Res, movx_Res, movy_Res, movz_Res,
+                           Gmx_Res, Gmy_Res, Gmz_Res,
+                           Symmetry, Lev, eps, co))
+        return 1;
+
+    lopsided_kodis(ex, X, Y, Z, Sphi, Sphi_rhs, betax, betay, betaz, Symmetry, SSS, eps);
+    lopsided_kodis(ex, X, Y, Z, Spi, Spi_rhs, betax, betay, betaz, Symmetry, SSS, eps);
+
+    for (int i = 0; i < all; ++i)
+    {
+        if (Sphi_rhs[i] != Sphi_rhs[i] || Spi_rhs[i] != Spi_rhs[i] || rho[i] != rho[i])
+            return 1;
+    }
+
+    return 0;
+}

AMSS_NCKU_source/bssn_rhs.h

@@ -63,13 +63,34 @@ extern "C"
                                double *, double *, double *, double *, double *, double *,                                         // Christoffel
                                double *, double *, double *, double *, double *, double *,                                         // Christoffel
                                double *, double *, double *, double *, double *, double *,                                         // Ricci
-                               double *, double *, double *, double *, double *, double *, double *,                               // constraint violation
-                               int &, int &, double &, int &);
-}
-
-extern "C"
-{
-        int f_compute_rhs_bssn_ss(int *, double &, double *, double *, double *,                                                      // ex,T,rho,sigma,R
+                               double *, double *, double *, double *, double *, double *, double *,                               // constraint violation
+                               int &, int &, double &, int &);
+}
+
+int f_compute_rhs_bssn_escalar_c(int *, double &, double *, double *, double *,                                                      // ex,T,X,Y,Z
+                                 double *, double *,                                                                                 // chi, trK
+                                 double *, double *, double *, double *, double *, double *,                                         // gij
+                                 double *, double *, double *, double *, double *, double *,                                         // Aij
+                                 double *, double *, double *,                                                                       // Gam
+                                 double *, double *, double *, double *, double *, double *, double *,                               // Gauge
+                                 double *, double *,                                                                                 // Sphi, Spi
+                                 double *, double *,                                                                                 // chi, trK
+                                 double *, double *, double *, double *, double *, double *,                                         // gij
+                                 double *, double *, double *, double *, double *, double *,                                         // Aij
+                                 double *, double *, double *,                                                                       // Gam
+                                 double *, double *, double *, double *, double *, double *, double *,                               // Gauge
+                                 double *, double *,                                                                                 // Sphi, Spi
+                                 double *, double *, double *, double *, double *, double *, double *, double *, double *, double *, // stress-energy
+                                 double *, double *, double *, double *, double *, double *,                                         // Christoffel
+                                 double *, double *, double *, double *, double *, double *,                                         // Christoffel
+                                 double *, double *, double *, double *, double *, double *,                                         // Christoffel
+                                 double *, double *, double *, double *, double *, double *,                                         // Ricci
+                                 double *, double *, double *, double *, double *, double *, double *,                               // constraint violation
+                                 int &, int &, double &, int &);
+
+extern "C"
+{
+        int f_compute_rhs_bssn_ss(int *, double &, double *, double *, double *,                                                      // ex,T,rho,sigma,R
                                   double *, double *, double *,                                                                       // X,Y,Z
                                   double *, double *, double *,                                                                       // drhodx,drhody,drhodz
                                   double *, double *, double *,                                                                       // dsigmadx,dsigmady,dsigmadz
@@ -96,10 +117,10 @@ extern "C"
                                   int &, int &, double &, int &, int &);
 }
 
-extern "C"
-{
-        int f_compute_rhs_bssn_escalar(int *, double &, double *, double *, double *,                                                      // ex,T,X,Y,Z
-                                       double *, double *,                                                                                 // chi, trK
+extern "C"
+{
+        int f_compute_rhs_bssn_escalar(int *, double &, double *, double *, double *,                                                      // ex,T,X,Y,Z
+                                       double *, double *,                                                                                 // chi, trK
                                        double *, double *, double *, double *, double *, double *,                                         // gij
                                        double *, double *, double *, double *, double *, double *,                                         // Aij
                                        double *, double *, double *,                                                                       // Gam
@@ -116,14 +137,14 @@ extern "C"
                                        double *, double *, double *, double *, double *, double *,                                         // Christoffel
                                        double *, double *, double *, double *, double *, double *,                                         // Christoffel
                                        double *, double *, double *, double *, double *, double *,                                         // Ricci
-                                       double *, double *, double *, double *, double *, double *, double *,                               // constraint violation
-                                       int &, int &, double &, int &);
-}
-
-extern "C"
-{
-        int f_compute_rhs_bssn_escalar_ss(int *, double &, double *, double *, double *,                                                      // ex,T,rho,sigma,R
-                                          double *, double *, double *,                                                                       // X,Y,Z
+                                       double *, double *, double *, double *, double *, double *, double *,                               // constraint violation
+                                       int &, int &, double &, int &);
+}
+
+extern "C"
+{
+        int f_compute_rhs_bssn_escalar_ss(int *, double &, double *, double *, double *,                                                      // ex,T,rho,sigma,R
+                                          double *, double *, double *,                                                                       // X,Y,Z
                                           double *, double *, double *,                                                                       // drhodx,drhody,drhodz
                                           double *, double *, double *,                                                                       // dsigmadx,dsigmady,dsigmadz
                                           double *, double *, double *,                                                                       // dRdx,dRdy,dRdz

AMSS_NCKU_source/gaussj.C

@@ -18,7 +18,7 @@ using namespace std;
 #endif
 
 // Intel oneMKL LAPACK interface
-#include <mkl_lapacke.h>
+#include <lapacke.h>
 /* Linear equation solution using Intel oneMKL LAPACK.
 a[0..n-1][0..n-1] is the input matrix. b[0..n-1] is input
 containing the right-hand side vectors. On output a is

AMSS_NCKU_source/makefile

@@ -1,63 +1,79 @@
-
-
+
+
 include makefile.inc
 
+-include AMSS_NCKU_build.mk
+
+ABE_TYPE ?= $(shell awk '/^[[:space:]]*\#define[[:space:]]+ABEtype/ {print $$3; exit}' macrodef.h 2>/dev/null)
+
+ifeq ($(USE_TRANSFER_CACHE),auto)
+ifeq ($(ABE_TYPE),0)
+EFFECTIVE_USE_TRANSFER_CACHE = 1
+else
+EFFECTIVE_USE_TRANSFER_CACHE = 0
+endif
+else
+EFFECTIVE_USE_TRANSFER_CACHE = $(USE_TRANSFER_CACHE)
+endif
+
+ifeq ($(USE_CXX_ESCALAR_KERNEL),1)
+ifeq ($(ABE_TYPE),1)
+EFFECTIVE_USE_CXX_ESCALAR_KERNEL = 1
+else
+EFFECTIVE_USE_CXX_ESCALAR_KERNEL = 0
+endif
+else
+EFFECTIVE_USE_CXX_ESCALAR_KERNEL = 0
+endif
+
+ifeq ($(EFFECTIVE_USE_CXX_ESCALAR_KERNEL),1)
+ifeq ($(USE_CXX_KERNELS),0)
+$(error USE_CXX_ESCALAR_KERNEL=1 requires USE_CXX_KERNELS=1 because bssn_escalar_rhs_c.C reuses the C BSSN kernel)
+endif
+endif
+
 ## polint(ordn=6) kernel selector:
 ##   1 (default): barycentric fast path
 ##   0          : fallback to Neville path
 POLINT6_USE_BARY ?= 1
 POLINT6_FLAG = -DPOLINT6_USE_BARYCENTRIC=$(POLINT6_USE_BARY)
+TRANSFER_CACHE_FLAG = -DBSSN_USE_TRANSFER_CACHE=$(EFFECTIVE_USE_TRANSFER_CACHE)
+ESCALAR_KERNEL_FLAG = -DBSSN_USE_ESCALAR_C_KERNEL=$(EFFECTIVE_USE_CXX_ESCALAR_KERNEL)
 
-## ABE build flags selected by PGO_MODE (set in makefile.inc, default: opt)
-##   make                        -> opt  (PGO-guided, maximum performance)
-##   make PGO_MODE=instrument    -> instrument (Phase 1: collect fresh profile data)
-PROFDATA = /home/$(shell whoami)/AMSS-NCKU/pgo_profile/default.profdata
-
-ifeq ($(PGO_MODE),instrument)
-## Phase 1: instrumentation — omit -ipo/-fp-model fast=2 for faster build and numerical stability
-CXXAPPFLAGS = -O3 -xHost -fma -fprofile-instr-generate -ipo \
-              -Dfortran3 -Dnewc -I${MKLROOT}/include $(INTERP_LB_FLAGS)
-f90appflags = -O3 -xHost -fma -fprofile-instr-generate -ipo \
-              -align array64byte -fpp -I${MKLROOT}/include $(POLINT6_FLAG)
-else
-## opt (default): maximum performance with PGO profile data -fprofile-instr-use=$(PROFDATA) \
-## PGO has been turned off, now tested and found to be negative optimization
-## INTERP_LB_FLAGS has been turned off too, now tested and found to be negative optimization
+## AMD AOCC build flags optimized for EPYC Zen 4 (-march=znver4)
+CXXAPPFLAGS = -O3 -march=znver4 -ffast-math -flto \
+              -Dfortran3 -Dnewc -I$(AOCL_ROOT)/include $(INTERP_LB_FLAGS) \
+              $(TRANSFER_CACHE_FLAG) $(ESCALAR_KERNEL_FLAG)
+f90appflags = -O3 -march=znver4 -ffast-math -flto \
+              -cpp -I$(AOCL_ROOT)/include $(POLINT6_FLAG)
 
+.SUFFIXES: .o .f90 .C .for .cu
+
+.f90.o:
+	$(f90) $(f90appflags) -c $< -o $@
+
+.C.o:
+	${CXX} $(CXXAPPFLAGS) -c $< $(filein) -o $@
+
+.for.o:
+	$(f77) -c $< -o $@
+
+.cu.o:
+	$(Cu) $(CUDA_APP_FLAGS) -c $< -o $@ $(CUDA_LIB_PATH)
+
+# C rewrite of BSSN RHS kernel and helpers
+bssn_rhs_c.o: bssn_rhs_c.C
+	${CXX} $(CXXAPPFLAGS) -c $< $(filein) -o $@
+
+fderivs_c.o: fderivs_c.C
+	${CXX} $(CXXAPPFLAGS) -c $< $(filein) -o $@
+
+fdderivs_c.o: fdderivs_c.C
+	${CXX} $(CXXAPPFLAGS) -c $< $(filein) -o $@
+
+kodiss_c.o: kodiss_c.C
+	${CXX} $(CXXAPPFLAGS) -c $< $(filein) -o $@
 
-CXXAPPFLAGS = -O3 -xHost -fp-model fast=2 -fma -ipo \
-              -Dfortran3 -Dnewc -I${MKLROOT}/include $(INTERP_LB_FLAGS)
-f90appflags = -O3 -xHost -fp-model fast=2 -fma -ipo \
-              -align array64byte -fpp -I${MKLROOT}/include $(POLINT6_FLAG)
-endif
-
-.SUFFIXES: .o .f90 .C .for .cu
-
-.f90.o:
-	$(f90) $(f90appflags) -c $< -o $@
-
-.C.o:
-	${CXX} $(CXXAPPFLAGS) -c $< $(filein) -o $@
-
-.for.o:
-	$(f77) -c $< -o $@
-
-.cu.o:
-	$(Cu) $(CUDA_APP_FLAGS) -c $< -o $@ $(CUDA_LIB_PATH)
-
-# C rewrite of BSSN RHS kernel and helpers
-bssn_rhs_c.o: bssn_rhs_c.C
-	${CXX} $(CXXAPPFLAGS) -c $< $(filein) -o $@
-
-fderivs_c.o: fderivs_c.C
-	${CXX} $(CXXAPPFLAGS) -c $< $(filein) -o $@
-
-fdderivs_c.o: fdderivs_c.C
-	${CXX} $(CXXAPPFLAGS) -c $< $(filein) -o $@
-
-kodiss_c.o: kodiss_c.C
-	${CXX} $(CXXAPPFLAGS) -c $< $(filein) -o $@
-
 lopsided_c.o: lopsided_c.C
 	${CXX} $(CXXAPPFLAGS) -c $< $(filein) -o $@
 
@@ -66,28 +82,29 @@ lopsided_kodis_c.o: lopsided_kodis_c.C
 
 #interp_lb_profile.o: interp_lb_profile.C interp_lb_profile.h
 #	${CXX} $(CXXAPPFLAGS) -c $< $(filein) -o $@
-
-## TwoPunctureABE uses fixed optimal flags with its own PGO profile, independent of CXXAPPFLAGS
-TP_PROFDATA = /home/$(shell whoami)/AMSS-NCKU/pgo_profile/TwoPunctureABE.profdata
-TP_OPTFLAGS = -O3 -xHost -fp-model fast=2 -fma -ipo \
-              -fprofile-instr-use=$(TP_PROFDATA) \
-              -Dfortran3 -Dnewc -I${MKLROOT}/include
-
-TwoPunctures.o: TwoPunctures.C
-	${CXX} $(TP_OPTFLAGS) -qopenmp -c $< -o $@
-
-TwoPunctureABE.o: TwoPunctureABE.C
-	${CXX} $(TP_OPTFLAGS) -qopenmp -c $< -o $@
-
-# Input files
-
-## Kernel implementation switch (set USE_CXX_KERNELS=0 to fall back to Fortran)
+
+## TwoPunctureABE uses fixed optimal flags (AMD AOCC, no PGO)
+TP_OPTFLAGS = -O3 -march=znver4 -ffast-math -flto \
+              -Dfortran3 -Dnewc -I$(AOCL_ROOT)/include
+
+TwoPunctures.o: TwoPunctures.C
+	${CXX} $(TP_OPTFLAGS) -fopenmp -c $< -o $@
+
+TwoPunctureABE.o: TwoPunctureABE.C
+	${CXX} $(TP_OPTFLAGS) -fopenmp -c $< -o $@
+
+# Input files
+
+## Kernel implementation switch (set USE_CXX_KERNELS=0 to fall back to Fortran)
 ifeq ($(USE_CXX_KERNELS),0)
 # Fortran mode: no C rewrite files; bssn_rhs.o is included via F90FILES below
 CFILES =
 else
-# C++ mode (default): C rewrite of bssn_rhs and helper kernels
+# C++ mode (default): C rewrite of bssn/bssn-escalar rhs and helper kernels
 CFILES = bssn_rhs_c.o fderivs_c.o fdderivs_c.o kodiss_c.o lopsided_c.o lopsided_kodis_c.o
+ifeq ($(EFFECTIVE_USE_CXX_ESCALAR_KERNEL),1)
+CFILES += bssn_escalar_rhs_c.o
+endif
 endif
 
 ## RK4 kernel switch (independent from USE_CXX_KERNELS)
@@ -97,95 +114,95 @@ RK4_F90_OBJ =
 else
 RK4_F90_OBJ = rungekutta4_rout.o
 endif
-
-C++FILES = ABE.o Ansorg.o Block.o misc.o monitor.o Parallel.o MPatch.o var.o\
-           cgh.o bssn_class.o surface_integral.o ShellPatch.o\
-	   bssnEScalar_class.o perf.o Z4c_class.o NullShellPatch.o\
-	   bssnEM_class.o cpbc_util.o z4c_rhs_point.o checkpoint.o\
-           Parallel_bam.o scalar_class.o transpbh.o NullShellPatch2.o\
-	   NullShellPatch2_Evo.o writefile_f.o interp_lb_profile.o
-	   
-C++FILES_GPU = ABE.o Ansorg.o Block.o misc.o monitor.o Parallel.o MPatch.o var.o\
-           cgh.o surface_integral.o ShellPatch.o\
-	   bssnEScalar_class.o perf.o Z4c_class.o NullShellPatch.o\
-	   bssnEM_class.o cpbc_util.o z4c_rhs_point.o checkpoint.o\
-           Parallel_bam.o scalar_class.o transpbh.o NullShellPatch2.o\
-	   NullShellPatch2_Evo.o \
-	   bssn_gpu_class.o bssn_step_gpu.o bssn_macro.o writefile_f.o
-
+
+C++FILES = ABE.o Ansorg.o Block.o misc.o monitor.o Parallel.o MPatch.o var.o\
+           cgh.o bssn_class.o surface_integral.o ShellPatch.o\
+	   bssnEScalar_class.o perf.o Z4c_class.o NullShellPatch.o\
+	   bssnEM_class.o cpbc_util.o z4c_rhs_point.o checkpoint.o\
+           Parallel_bam.o scalar_class.o transpbh.o NullShellPatch2.o\
+	   NullShellPatch2_Evo.o writefile_f.o interp_lb_profile.o
+	   
+C++FILES_GPU = ABE.o Ansorg.o Block.o misc.o monitor.o Parallel.o MPatch.o var.o\
+           cgh.o surface_integral.o ShellPatch.o\
+	   bssnEScalar_class.o perf.o Z4c_class.o NullShellPatch.o\
+	   bssnEM_class.o cpbc_util.o z4c_rhs_point.o checkpoint.o\
+           Parallel_bam.o scalar_class.o transpbh.o NullShellPatch2.o\
+	   NullShellPatch2_Evo.o \
+	   bssn_gpu_class.o bssn_step_gpu.o bssn_macro.o writefile_f.o
+
 F90FILES_BASE = enforce_algebra.o fmisc.o initial_puncture.o prolongrestrict.o\
 	   prolongrestrict_cell.o prolongrestrict_vertex.o\
 	   $(RK4_F90_OBJ) diff_new.o kodiss.o kodiss_sh.o\
 	   lopsidediff.o sommerfeld_rout.o getnp4.o diff_new_sh.o\
 	   shellfunctions.o bssn_rhs_ss.o Set_Rho_ADM.o\
            getnp4EScalar.o bssnEScalar_rhs.o bssn_constraint.o ricci_gamma.o\
-           fadmquantites_bssn.o Z4c_rhs.o Z4c_rhs_ss.o point_diff_new_sh.o\
-	   cpbc.o getnp4old.o NullEvol.o initial_null.o initial_maxwell.o\
-	   getnpem2.o empart.o NullNews.o fourdcurvature.o\
-	   bssn2adm.o adm_constraint.o adm_ricci_gamma.o\
-	   scalar_rhs.o initial_scalar.o NullEvol2.o initial_null2.o\
-	   NullNews2.o tool_f.o
-
-ifeq ($(USE_CXX_KERNELS),0)
-# Fortran mode: include original bssn_rhs.o
-F90FILES = $(F90FILES_BASE) bssn_rhs.o
-else
-# C++ mode (default): bssn_rhs.o replaced by C++ kernel
-F90FILES = $(F90FILES_BASE)
-endif
-
-F77FILES = zbesh.o
-
-AHFDOBJS = expansion.o expansion_Jacobian.o patch.o coords.o patch_info.o patch_interp.o patch_system.o \
-tgrid.o fd_grid.o ghost_zone.o array.o round.o norm.o fuzzy.o error_exit.o miscfp.o \
-linear_map.o cpm_map.o BH_diagnostics.o setup.o horizon_sequence.o find_horizons.o \
-initial_guess.o Newton.o Jacobian.o ilucg.o IntPnts0.o IntPnts.o
-
-TwoPunctureFILES = TwoPunctureABE.o TwoPunctures.o 
-
-CUDAFILES = bssn_gpu.o bssn_gpu_rhs_ss.o
-
-# file dependences
-$(C++FILES) $(C++FILES_GPU) $(F90FILES) $(CFILES) $(AHFDOBJS) $(CUDAFILES): macrodef.fh
-
-$(C++FILES): Block.h enforce_algebra.h fmisc.h initial_puncture.h macrodef.h\
-	     misc.h monitor.h MyList.h Parallel.h MPatch.h prolongrestrict.h\
-	     rungekutta4_rout.h var.h bssn_class.h bssn_rhs.h sommerfeld_rout.h\
-	     cgh.h surface_integral.h ShellPatch.h shellfunctions.h perf.h\
-             fadmquantites_bssn.h cpbc.h getnp4.h initial_null.h NullEvol.h\
-	     NullShellPatch.h initial_maxwell.h bssnEM_class.h getnpem2.h\
-	     empart.h NullNews.h kodiss.h Parallel_bam.h ricci_gamma.h\
-             initial_null2.h NullShellPatch2.h 
-             
-$(C++FILES_GPU): Block.h enforce_algebra.h fmisc.h initial_puncture.h macrodef.h\
-	     misc.h monitor.h MyList.h Parallel.h MPatch.h prolongrestrict.h\
-	     rungekutta4_rout.h var.h bssn_rhs.h sommerfeld_rout.h\
-	     cgh.h surface_integral.h ShellPatch.h shellfunctions.h perf.h\
-             fadmquantites_bssn.h cpbc.h getnp4.h initial_null.h NullEvol.h\
-	     NullShellPatch.h initial_maxwell.h bssnEM_class.h getnpem2.h\
-	     empart.h NullNews.h kodiss.h Parallel_bam.h ricci_gamma.h\
-             initial_null2.h NullShellPatch2.h \
-             bssn_gpu_class.h bssn_macro.h
-             
-$(AHFDOBJS): cctk.h cctk_Config.h cctk_Types.h cctk_Constants.h myglobal.h
-
-$(C++FILES) $(C++FILES_GPU) $(CFILES) $(AHFDOBJS) $(CUDAFILES): macrodef.h
-
-TwoPunctureFILES: TwoPunctures.h
-
-$(CUDAFILES): bssn_gpu.h gpu_mem.h gpu_rhsSS_mem.h
-
-misc.o : zbesh.o
-
-# projects
-ABE: $(C++FILES) $(CFILES) $(F90FILES) $(F77FILES) $(AHFDOBJS)
-	$(CLINKER) $(CXXAPPFLAGS) -o $@ $(C++FILES) $(CFILES) $(F90FILES) $(F77FILES) $(AHFDOBJS) $(LDLIBS)
-	
-ABEGPU: $(C++FILES_GPU) $(CFILES) $(F90FILES) $(F77FILES) $(AHFDOBJS) $(CUDAFILES)
-	$(CLINKER) $(CXXAPPFLAGS) -o $@ $(C++FILES_GPU) $(CFILES) $(F90FILES) $(F77FILES) $(AHFDOBJS) $(CUDAFILES) $(LDLIBS)
-
-TwoPunctureABE: $(TwoPunctureFILES)
-	$(CLINKER) $(TP_OPTFLAGS) -qopenmp -o $@ $(TwoPunctureFILES) $(LDLIBS)
-
-clean:
-	rm *.o ABE ABEGPU TwoPunctureABE make.log -f
+           fadmquantites_bssn.o Z4c_rhs.o Z4c_rhs_ss.o point_diff_new_sh.o\
+	   cpbc.o getnp4old.o NullEvol.o initial_null.o initial_maxwell.o\
+	   getnpem2.o empart.o NullNews.o fourdcurvature.o\
+	   bssn2adm.o adm_constraint.o adm_ricci_gamma.o\
+	   scalar_rhs.o initial_scalar.o NullEvol2.o initial_null2.o\
+	   NullNews2.o tool_f.o
+
+ifeq ($(USE_CXX_KERNELS),0)
+# Fortran mode: include original bssn_rhs.o
+F90FILES = $(F90FILES_BASE) bssn_rhs.o
+else
+# C++ mode (default): bssn_rhs.o replaced by C++ kernel
+F90FILES = $(F90FILES_BASE)
+endif
+
+F77FILES = zbesh.o
+
+AHFDOBJS = expansion.o expansion_Jacobian.o patch.o coords.o patch_info.o patch_interp.o patch_system.o \
+tgrid.o fd_grid.o ghost_zone.o array.o round.o norm.o fuzzy.o error_exit.o miscfp.o \
+linear_map.o cpm_map.o BH_diagnostics.o setup.o horizon_sequence.o find_horizons.o \
+initial_guess.o Newton.o Jacobian.o ilucg.o IntPnts0.o IntPnts.o
+
+TwoPunctureFILES = TwoPunctureABE.o TwoPunctures.o 
+
+CUDAFILES = bssn_gpu.o bssn_gpu_rhs_ss.o
+
+# file dependences
+$(C++FILES) $(C++FILES_GPU) $(F90FILES) $(CFILES) $(AHFDOBJS) $(CUDAFILES): macrodef.fh
+
+$(C++FILES): Block.h enforce_algebra.h fmisc.h initial_puncture.h macrodef.h\
+	     misc.h monitor.h MyList.h Parallel.h MPatch.h prolongrestrict.h\
+	     rungekutta4_rout.h var.h bssn_class.h bssn_rhs.h sommerfeld_rout.h\
+	     cgh.h surface_integral.h ShellPatch.h shellfunctions.h perf.h\
+             fadmquantites_bssn.h cpbc.h getnp4.h initial_null.h NullEvol.h\
+	     NullShellPatch.h initial_maxwell.h bssnEM_class.h getnpem2.h\
+	     empart.h NullNews.h kodiss.h Parallel_bam.h ricci_gamma.h\
+             initial_null2.h NullShellPatch2.h 
+             
+$(C++FILES_GPU): Block.h enforce_algebra.h fmisc.h initial_puncture.h macrodef.h\
+	     misc.h monitor.h MyList.h Parallel.h MPatch.h prolongrestrict.h\
+	     rungekutta4_rout.h var.h bssn_rhs.h sommerfeld_rout.h\
+	     cgh.h surface_integral.h ShellPatch.h shellfunctions.h perf.h\
+             fadmquantites_bssn.h cpbc.h getnp4.h initial_null.h NullEvol.h\
+	     NullShellPatch.h initial_maxwell.h bssnEM_class.h getnpem2.h\
+	     empart.h NullNews.h kodiss.h Parallel_bam.h ricci_gamma.h\
+             initial_null2.h NullShellPatch2.h \
+             bssn_gpu_class.h bssn_macro.h
+             
+$(AHFDOBJS): cctk.h cctk_Config.h cctk_Types.h cctk_Constants.h myglobal.h
+
+$(C++FILES) $(C++FILES_GPU) $(CFILES) $(AHFDOBJS) $(CUDAFILES): macrodef.h
+
+TwoPunctureFILES: TwoPunctures.h
+
+$(CUDAFILES): bssn_gpu.h gpu_mem.h gpu_rhsSS_mem.h
+
+misc.o : zbesh.o
+
+# projects
+ABE: $(C++FILES) $(CFILES) $(F90FILES) $(F77FILES) $(AHFDOBJS)
+	$(CLINKER) $(CXXAPPFLAGS) -o $@ $(C++FILES) $(CFILES) $(F90FILES) $(F77FILES) $(AHFDOBJS) $(LDLIBS)
+	
+ABEGPU: $(C++FILES_GPU) $(CFILES) $(F90FILES) $(F77FILES) $(AHFDOBJS) $(CUDAFILES)
+	$(CLINKER) $(CXXAPPFLAGS) -o $@ $(C++FILES_GPU) $(CFILES) $(F90FILES) $(F77FILES) $(AHFDOBJS) $(CUDAFILES) $(LDLIBS)
+
+TwoPunctureABE: $(TwoPunctureFILES)
+	$(CLINKER) $(TP_OPTFLAGS) -fopenmp -o $@ $(TwoPunctureFILES) $(LDLIBS)
+
+clean:
+	rm *.o ABE ABEGPU TwoPunctureABE make.log -f

AMSS_NCKU_source/makefile.inc

@@ -1,33 +1,17 @@
-## GCC version (commented out)
-## filein  = -I/usr/include -I/usr/lib/x86_64-linux-gnu/mpich/include -I/usr/lib/x86_64-linux-gnu/openmpi/lib/ -I/usr/lib/gcc/x86_64-linux-gnu/11/ -I/usr/include/c++/11/
-## filein  = -I/usr/include/ -I/usr/include/openmpi-x86_64/ -I/usr/lib/x86_64-linux-gnu/openmpi/include/ -I/usr/lib/x86_64-linux-gnu/openmpi/lib/ -I/usr/lib/gcc/x86_64-linux-gnu/11/ -I/usr/include/c++/11/
-## LDLIBS  = -L/usr/lib/x86_64-linux-gnu -L/usr/lib64 -L/usr/lib/gcc/x86_64-linux-gnu/11 -lgfortran -lmpi -lgfortran
+## AMD AOCC version with AOCL (Optimized for AMD EPYC Zen 4)
 
-## Intel oneAPI version with oneMKL (Optimized for performance)
-filein  = -I/usr/include/ -I${MKLROOT}/include
+## AOCL root path for includes and libraries
+AOCL_ROOT ?= /home/gh0s7/AOCC/aocl/5.2.0/aocc
 
-## Using sequential MKL (OpenMP disabled for better single-threaded performance)
-## Added -lifcore for Intel Fortran runtime and -limf for Intel math library
-LDLIBS  = -L${MKLROOT}/lib -lmkl_intel_lp64 -lmkl_sequential -lmkl_core -lifcore -limf -lpthread -lm -ldl -liomp5
+## AOCC-built OpenMPI prefix
+OMPI_PREFIX ?= /home/gh0s7/AOCC/aocc-openmpi
 
-## Memory allocator switch
-##   1 (default) : link Intel oneTBB allocator (libtbbmalloc)
-##   0           : use system default allocator (ptmalloc)
-USE_TBBMALLOC ?= 1
-TBBMALLOC_SO ?= /home/intel/oneapi/2025.3/lib/libtbbmalloc.so
-ifneq ($(wildcard $(TBBMALLOC_SO)),)
-TBBMALLOC_LIBS = -Wl,--no-as-needed $(TBBMALLOC_SO) -Wl,--as-needed
-else
-TBBMALLOC_LIBS = -Wl,--no-as-needed -ltbbmalloc -Wl,--as-needed
-endif
-ifeq ($(USE_TBBMALLOC),1)
-LDLIBS := $(TBBMALLOC_LIBS) $(LDLIBS)
-endif
+filein  = -I/usr/include/ -I$(AOCL_ROOT)/include
 
-## PGO build mode switch (ABE only; TwoPunctureABE always uses opt flags)
-##   opt        : (default) maximum performance with PGO profile-guided optimization
-##   instrument : PGO Phase 1 instrumentation to collect fresh profile data
-PGO_MODE ?= opt
+## Using AOCL BLIS + libFLAME for BLAS/LAPACK
+## AOCC Fortran runtime: -lflang (includes FortranRuntime)
+## AOCC OpenMP runtime: -lomp (LLVM OpenMP)
+LDLIBS  = -L$(AOCL_ROOT)/lib -lblis -lflame -lamdlibm -lflang -lpgmath -lpthread -lm -ldl -lomp
 
 ## Interp_Points load balance profiling mode
 ##   off        : (default) no load balance instrumentation
@@ -48,16 +32,28 @@ endif
 ##   0           : fall back to original Fortran kernels
 USE_CXX_KERNELS ?= 1
 
+## BSSN-EScalar RHS switch
+##   1 (default) : use BSSN-EScalar C wrapper on the normal patch path
+##   0           : keep the original Fortran BSSN-EScalar RHS for precision-safe runs
+## Note: this requires USE_CXX_KERNELS=1 because the wrapper reuses the C BSSN kernel.
+USE_CXX_ESCALAR_KERNEL ?= 1
+
+## Cached transfer switch
+##   auto (default): enable for BSSN vacuum, keep other paths on the safe uncached path
+##   1             : force cached Sync/Restrict/OutBd transfer on evolution hot paths
+##   0             : force the original uncached transfer path
+USE_TRANSFER_CACHE ?= auto
+
 ## RK4 kernel implementation switch
 ##   1 (default) : use C/C++ rewrite of rungekutta4_rout (for optimization experiments)
 ##   0           : use original Fortran rungekutta4_rout.o
 USE_CXX_RK4 ?= 1
 
-f90          = ifx
-f77          = ifx
-CXX          = icpx
-CC           = icx
-CLINKER      = mpiicpx
+f90          = flang
+f77          = flang
+CXX          = clang++
+CC           = clang
+CLINKER      = $(OMPI_PREFIX)/bin/mpicxx
 
 Cu = nvcc
 CUDA_LIB_PATH = -L/usr/lib/cuda/lib64 -I/usr/include -I/usr/lib/cuda/include

BSSN_BUILD_CONFIG_MIGRATION.md

@@ -0,0 +1,211 @@
+# BSSN Build Config Migration
+
+This note records the build-configuration fix needed when replacing
+`AMSS_NCKU_Input.py` or `generate_macrodef.py` with a newer upstream version.
+
+## Problem
+
+`AMSS_NCKU_source/macrodef.h` is not the authoritative file used by normal
+runs. `AMSS_NCKU_Program.py` first generates macro files under
+`input_data.File_directory`, copies `AMSS_NCKU_source` to
+`<File_directory>/AMSS_NCKU_source_copy`, then copies the generated macro files
+into that copied source tree and compiles there.
+
+Therefore, makefile logic must not depend only on the stale
+`AMSS_NCKU_source/macrodef.h`. The actual equation path must be passed to the
+copied build tree from the same generation step that creates `macrodef.h`.
+
+The performance regression was caused by compiling/linking the
+`BSSN-EScalar` C wrapper into BSSN vacuum builds. For BSSN vacuum (`ABEtype=0`),
+the build must use:
+
+```make
+BSSN_USE_TRANSFER_CACHE=1
+BSSN_USE_ESCALAR_C_KERNEL=0
+```
+
+and must not link `bssn_escalar_rhs_c.o`.
+
+## Required Migration Steps
+
+### 1. Add an ABE type helper in `generate_macrodef.py`
+
+Add a helper that maps `input_data.Equation_Class` to the numeric `ABEtype`.
+Use the same mapping as `macrodef.h`:
+
+```python
+def get_abe_type():
+    if ( input_data.Equation_Class == "BSSN" ):
+        return 0
+    elif ( input_data.Equation_Class == "BSSN-EScalar" ):
+        return 1
+    elif ( input_data.Equation_Class == "BSSN-EM" ):
+        return 3
+    elif ( input_data.Equation_Class == "Z4C" ):
+        return 2
+    else:
+        raise ValueError("Equation_Class setting error!!!")
+```
+
+Update `generate_macrodef_h()` to print `#define ABEtype {get_abe_type()}`
+instead of duplicating the if/elif mapping.
+
+### 2. Generate a makefile fragment
+
+In `generate_macrodef.py`, add:
+
+```python
+def generate_build_config():
+    file1 = open(os.path.join(input_data.File_directory, "AMSS_NCKU_build.mk"), "w")
+    print("# Generated by generate_macrodef.py; do not edit manually.", file=file1)
+    print(f"ABE_TYPE := {get_abe_type()}", file=file1)
+    file1.close()
+```
+
+This file is the build-time authority for the equation path.
+
+### 3. Call and copy the generated build config
+
+In `AMSS_NCKU_Program.py`, after generating `macrodef.h` and `macrodef.fh`, call:
+
+```python
+generate_macrodef.generate_build_config()
+print(" AMSS-NCKU build config AMSS_NCKU_build.mk has been generated. ")
+```
+
+When copying generated files into `AMSS_NCKU_source_copy`, also copy:
+
+```python
+build_config_path = os.path.join(File_directory, "AMSS_NCKU_build.mk")
+shutil.copy2(build_config_path, AMSS_NCKU_source_copy)
+```
+
+### 4. Make the source makefile consume the generated config
+
+At the top of `AMSS_NCKU_source/makefile`, after `include makefile.inc`, add:
+
+```make
+-include AMSS_NCKU_build.mk
+
+ABE_TYPE ?= $(shell awk '/^[[:space:]]*\#define[[:space:]]+ABEtype/ {print $$3; exit}' macrodef.h 2>/dev/null)
+```
+
+The generated `AMSS_NCKU_build.mk` is used during normal Python-driven builds.
+The fallback keeps manual source-tree builds usable.
+
+### 5. Gate path-specific build options by `ABE_TYPE`
+
+Use effective build switches:
+
+```make
+ifeq ($(USE_TRANSFER_CACHE),auto)
+ifeq ($(ABE_TYPE),0)
+EFFECTIVE_USE_TRANSFER_CACHE = 1
+else
+EFFECTIVE_USE_TRANSFER_CACHE = 0
+endif
+else
+EFFECTIVE_USE_TRANSFER_CACHE = $(USE_TRANSFER_CACHE)
+endif
+
+ifeq ($(USE_CXX_ESCALAR_KERNEL),1)
+ifeq ($(ABE_TYPE),1)
+EFFECTIVE_USE_CXX_ESCALAR_KERNEL = 1
+else
+EFFECTIVE_USE_CXX_ESCALAR_KERNEL = 0
+endif
+else
+EFFECTIVE_USE_CXX_ESCALAR_KERNEL = 0
+endif
+
+TRANSFER_CACHE_FLAG = -DBSSN_USE_TRANSFER_CACHE=$(EFFECTIVE_USE_TRANSFER_CACHE)
+ESCALAR_KERNEL_FLAG = -DBSSN_USE_ESCALAR_C_KERNEL=$(EFFECTIVE_USE_CXX_ESCALAR_KERNEL)
+```
+
+Only add `bssn_escalar_rhs_c.o` when the effective EScalar C kernel switch is
+enabled:
+
+```make
+ifeq ($(EFFECTIVE_USE_CXX_ESCALAR_KERNEL),1)
+CFILES += bssn_escalar_rhs_c.o
+endif
+```
+
+### 6. Use safe transfer-cache default
+
+In `AMSS_NCKU_source/makefile.inc`, keep:
+
+```make
+USE_TRANSFER_CACHE ?= auto
+```
+
+With the effective switch logic above, this enables cached transfer for BSSN
+vacuum while keeping non-BSSN paths on the uncached path by default.
+
+## Verification Checklist
+
+Run these checks after migrating:
+
+```bash
+python3 -c "import generate_macrodef; generate_macrodef.generate_build_config()"
+cat GW150914/AMSS_NCKU_build.mk
+```
+
+For BSSN, the generated file should contain:
+
+```make
+ABE_TYPE := 0
+```
+
+Dry-run the copied or source makefile:
+
+```bash
+make -n -B INTERP_LB_MODE=off ABE | grep -E 'BSSN_USE_TRANSFER_CACHE|BSSN_USE_ESCALAR_C_KERNEL|bssn_escalar_rhs_c'
+```
+
+Expected BSSN result:
+
+```text
+-DBSSN_USE_TRANSFER_CACHE=1 -DBSSN_USE_ESCALAR_C_KERNEL=0
+```
+
+and no `bssn_escalar_rhs_c.o` in the final link command.
+
+Run the full workflow:
+
+```bash
+python3 AMSS_NCKU_Program.py
+```
+
+For the 10-step BSSN test, compare coordinate output:
+
+```bash
+python3 - <<'PY'
+from pathlib import Path
+old = Path('../GW150914-06457/AMSS_NCKU_output/bssn_BH.dat')
+new = Path('GW150914/AMSS_NCKU_output/bssn_BH.dat')
+
+def rows(path):
+    out = []
+    for line in path.read_text().splitlines():
+        if not line.strip() or line.lstrip().startswith('#'):
+            continue
+        out.append([float(x) for x in line.split()])
+    return out
+
+ro, rn = rows(old), rows(new)
+n = min(len(ro), len(rn))
+max_abs = 0.0
+for i in range(n):
+    for a, b in zip(ro[i], rn[i]):
+        max_abs = max(max_abs, abs(a - b))
+print(f"old_rows={len(ro)} new_rows={len(rn)} compared_rows={n}")
+print(f"max_abs_diff={max_abs:.17g}")
+PY
+```
+
+For the validated migration, the first 10 rows matched exactly:
+
+```text
+max_abs_diff=0
+```

generate_macrodef.py

@@ -12,6 +12,37 @@ import os
 import AMSS_NCKU_Input as input_data          ## import program input file
 
 
+##################################################################
+
+def get_abe_type():
+    if ( input_data.Equation_Class == "BSSN" ):
+        return 0
+    elif ( input_data.Equation_Class == "BSSN-EScalar" ):
+        return 1
+    elif ( input_data.Equation_Class == "BSSN-EM" ):
+        return 3
+    elif ( input_data.Equation_Class == "Z4C" ):
+        return 2
+    else:
+        raise ValueError("Equation_Class setting error!!!")
+
+
+##################################################################
+
+## Generate the makefile fragment used by the copied source tree.
+## The source-tree macrodef.h is not authoritative because macro files
+## are regenerated under File_directory for each run.
+
+def generate_build_config():
+
+    file1 = open( os.path.join(input_data.File_directory, "AMSS_NCKU_build.mk"), "w")
+
+    print( "# Generated by generate_macrodef.py; do not edit manually.", file=file1 )
+    print( f"ABE_TYPE := {get_abe_type()}",                             file=file1 )
+
+    file1.close()
+
+
 ##################################################################
 
 ## Generate the macro file macrodef.h according to user settings
@@ -58,19 +89,10 @@ def generate_macrodef_h():
     # 2: Z4c vacuum
     # 3: coupled to Maxwell field
 
-    if ( input_data.Equation_Class == "BSSN" ):
-        print( "#define ABEtype 0", file=file1 )
-        print(                      file=file1 )
-    elif ( input_data.Equation_Class == "BSSN-EScalar" ):
-        print( "#define ABEtype 1", file=file1 )
-        print(                      file=file1 )
-    elif ( input_data.Equation_Class == "BSSN-EM" ):
-        print( "#define ABEtype 3", file=file1 )
-        print(                      file=file1 )
-    elif ( input_data.Equation_Class == "Z4C" ):
-        print( "#define ABEtype 2", file=file1 )
-        print(                      file=file1 )
-    else:
+    try:
+        print( f"#define ABEtype {get_abe_type()}", file=file1 )
+        print(                                      file=file1 )
+    except ValueError:
         print( "Equation_Class setting error!!!"                )
         print()
         print( "# Equation type #define ABEtype setting error!!!", file=file1 )