Fixups for M4 macs

config.py

@@ -382,7 +382,7 @@ def setup_extension_modules(self):
         # If user wishes to specify number of groups at compile time
         if self.num_groups:
             for k in self.compiler_flags:
-                self.compiler_flags[k].append('-DNGROUPS=' +
+                self.compiler_flags[k].append('-DNRGROUPS=' +
                                               str(self.num_groups))
 
         # If the user wishes to compile using debug mode, append the debugging
@@ -432,6 +432,13 @@ def setup_extension_modules(self):
         for cc in self.include_directories.keys():
             self.include_directories[cc].append(numpy_include)
 
+        # Add the python dev libraries
+        import sysconfig
+        python_include = sysconfig.get_path('include')
+        print(python_include)
+        for cc in self.include_directories.keys():
+            self.include_directories[cc].append(python_include)
+
         # Add the mpi4py module directory
         try:
             import mpi4py
@@ -451,6 +458,9 @@ def setup_extension_modules(self):
         if self.cc == 'mpicc':
             self.swig_flags += ['-DMPIx']
 
+        # Dedup rpath
+        self.linker_flags[self.cc] = list(dict.fromkeys(self.linker_flags[self.cc]))
+
         self.extensions.append(
              Extension(name = '_openmoc',
                        sources = copy.deepcopy(self.sources[self.cc]),

openmoc/swig/openmoc.i

@@ -92,7 +92,7 @@
 %ignore twiddleRead(boundaryType* ptr, size_t size, size_t nmemb, FILE* stream);
 %ignore twiddleRead(double* ptr, size_t size, size_t nmemb, FILE* stream);
 %ignore twiddleRead(long* ptr, size_t size, size_t nmemb, FILE* stream);
-%ignore setRuntimeParameters(RuntimeParameters &RP, int argc, char *argv[]); 
+%ignore setRuntimeParameters(RuntimeParameters &RP, int argc, char *argv[]);
 
 /* Instruct SWIG to ignore methods used in getting CSR Matrix format and Vector
  * attributes. These attributes should be used internally only by the Matrix and
@@ -105,7 +105,7 @@
 
 %exception {
   try {
-    $function
+    $action
   } catch (const std::exception &e) {
       SWIG_exception(SWIG_RuntimeError, e.what());
   }

profile/Makefile

@@ -204,7 +204,7 @@ ifeq ($(OPTIMIZE),yes)
     CFLAGS += -O3 -ffast-math -fvectorize -fpic
   endif
   # Set the number of groups at compile time
-  #CFLAGS += -DNGROUPS=70
+  #CFLAGS += -DNRGROUPS=70
   # Set the source to linear source at compile time
   #CFLAGS += -DLINEARSOURCE
   # Set the dimension to 3D at compile time

src/CPUSolver.h

@@ -21,8 +21,8 @@
 
 #undef track_flux
 /** Optimization macro to facilitate SIMD vectorization */
-#ifdef NGROUPS
-#define _NUM_GROUPS (NGROUPS)
+#ifdef NRGROUPS
+#define _NUM_GROUPS (NRGROUPS)
 #else
 #define _NUM_GROUPS (_num_groups)
 #endif

src/MOCKernel.cpp

@@ -10,7 +10,7 @@
 MOCKernel::MOCKernel(TrackGenerator* track_generator) {
   _count = 0;
   _max_tau = track_generator->retrieveMaxOpticalLength();
-#ifndef NGROUPS
+#ifndef NRGROUPS
   _num_groups = track_generator->getGeometry()->getNumEnergyGroups();
 #endif
 }

src/Solver.cpp

@@ -829,12 +829,12 @@ void Solver::initializeFSRs() {
   _track_generator->initializeFSRVolumesBuffer();
   _FSR_volumes = _track_generator->getFSRVolumes();
 
-#ifdef NGROUPS
-  if (_geometry->getNumEnergyGroups() != NGROUPS)
+#ifdef NRGROUPS
+  if (_geometry->getNumEnergyGroups() != NRGROUPS)
     log_printf(ERROR, "OpenMOC has been compiled for %d groups, and the "
                "current case is in %d groups, please re-compile with the right "
-               "number of groups for the -DNGROUPS flag or without that flag.",
-               NGROUPS, _geometry->getNumEnergyGroups());
+               "number of groups for the -DNRGROUPS flag or without that flag.",
+               NRGROUPS, _geometry->getNumEnergyGroups());
 #endif
 
   /* Retrieve simulation parameters from the Geometry */

src/TrackGenerator.h

@@ -27,6 +27,7 @@
 
 /* Modified from so/33745364/sched-getcpu-equivalent-for-os-x */
 #ifndef __linux__
+#if defined(__x86_64__) || defined(_M_X64) || defined(__i386__) || defined(_M_IX86)
 #include <cpuid.h>
 
 inline int sched_getcpu() {
@@ -41,6 +42,23 @@ inline int sched_getcpu() {
   if (CPU < 0) CPU = 0;
   return CPU;
 }
+#else
+#include <iostream>
+#include <sys/types.h>
+#include <sys/sysctl.h>
+inline int sched_getcpu() {
+    char buffer[128];
+    size_t bufferlen = sizeof(buffer);
+
+    if (sysctlbyname("machdep.cpu.brand_string", &buffer, &bufferlen, NULL, 0) == 0) {
+        return atoi(buffer);
+    }
+    else
+        std::cerr << "Failed to get CPU id" << std::endl;
+
+    return 0;
+  }
+#endif
 #endif
 
 /**

src/TrackTraversingAlgorithms.cpp

@@ -476,7 +476,7 @@ LinearExpansionGenerator::LinearExpansionGenerator(CPULSSolver* solver)
   _FSR_volumes = track_generator->getFSRVolumesBuffer();
   _FSR_locks = track_generator->getFSRLocks();
   _quadrature = track_generator->getQuadrature();
-#ifndef NGROUPS
+#ifndef NRGROUPS
   _NUM_GROUPS = solver->getNumEnergyGroups();
 #endif
 
@@ -528,7 +528,7 @@ LinearExpansionGenerator::~LinearExpansionGenerator() {
 
 
 /**
- * @brief When executed, the LinearExpansionGenerator Kernel loops over all 
+ * @brief When executed, the LinearExpansionGenerator Kernel loops over all
  *        Tracks to compute constant terms used to compute the linear source.
  */
 void LinearExpansionGenerator::execute() {
@@ -843,7 +843,7 @@ TransportSweep::TransportSweep(CPUSolver* cpu_solver)
   _ls_solver = dynamic_cast<CPULSSolver*>(cpu_solver);
   TrackGenerator* track_generator = cpu_solver->getTrackGenerator();
   _geometry = _track_generator->getGeometry();
-#ifndef NGROUPS
+#ifndef NRGROUPS
   _NUM_GROUPS = _geometry->getNumEnergyGroups();
 #endif
 }
@@ -936,7 +936,7 @@ void TransportSweep::onTrack(Track* track, segment* segments) {
 
   /* Allocate a temporary flux buffer on the stack (free) and initialize it */
   /* Select right size for buffer */
-#ifndef NGROUPS
+#ifndef NRGROUPS
   int _NUM_GROUPS = _cpu_solver->getNumEnergyGroups();
 #endif
 #ifndef LINEARSOURCE
@@ -975,7 +975,7 @@ void TransportSweep::onTrack(Track* track, segment* segments) {
     else
 #endif
       _ls_solver->tallyLSScalarFlux(curr_segment, azim_index, polar_index,
-                                    fsr_flux, fsr_flux_x, fsr_flux_y, 
+                                    fsr_flux, fsr_flux_x, fsr_flux_y,
                                     fsr_flux_z, track_flux, direction);
 
     /* Accumulate contribution of segments to scalar flux before changing fsr */
@@ -1023,7 +1023,7 @@ void TransportSweep::onTrack(Track* track, segment* segments) {
     else
 #endif
       _ls_solver->tallyLSScalarFlux(curr_segment, azim_index, polar_index,
-                                    fsr_flux, fsr_flux_x, fsr_flux_y, 
+                                    fsr_flux, fsr_flux_x, fsr_flux_y,
                                     fsr_flux_z, track_flux, direction);
 
     /* Accumulate contribution of segments to scalar flux before changing fsr */
@@ -1240,7 +1240,7 @@ void ReadSegments::onTrack(Track* track, segment* segments) {
 
 
 /**
- * @brief Constructor for TransportSweepOTF calls the TraverseSegments 
+ * @brief Constructor for TransportSweepOTF calls the TraverseSegments
  *        constructor.
  * @param track_generator Track generator to generate the tracks
  */

src/accel/cuda/GPUSolver.cu

@@ -3,7 +3,7 @@
 /** The number of FSRs */
 __constant__ long num_FSRs;
 
-#ifndef NGROUPS
+#ifndef NRGROUPS
 /** The number of energy groups */
 __constant__ int NUM_GROUPS;
 #endif
@@ -1263,7 +1263,7 @@ void GPUSolver::initializeMaterials(solverMode mode) {
     log_printf(ERROR, "Attempt to initialize GPU XS data with zero or less energy groups.");
 
   /* Copy the number of energy groups to constant memory on the GPU */
-#ifndef NGROUPS
+#ifndef NRGROUPS
   cudaMemcpyToSymbol(NUM_GROUPS, &_NUM_GROUPS, sizeof(int));
   getLastCudaError();
 #endif
@@ -1292,7 +1292,7 @@ void GPUSolver::initializeMaterials(solverMode mode) {
      * on the device */
     cudaMalloc(&_materials, _num_materials * sizeof(dev_material));
     getLastCudaError();
-    
+
     for (iter=host_materials.begin(); iter != host_materials.end(); ++iter) {
       clone_material(iter->second, &_materials[material_index]);
 

src/accel/cuda/GPUSolver.h

@@ -36,9 +36,9 @@
 #include "GPUQuery.h"
 
 /** If number of groups is known at compile time */
-#ifdef NGROUPS
-#define NUM_GROUPS (NGROUPS)
-#define _NUM_GROUPS (NGROUPS)
+#ifdef NRGROUPS
+#define NUM_GROUPS (NRGROUPS)
+#define _NUM_GROUPS (NRGROUPS)
 #else
 #define _NUM_GROUPS (_num_groups)
 #endif
@@ -170,7 +170,7 @@ class GPUSolver : public Solver {
   double computeResidual(residualType res_type);
 
   void computeFSRFissionRates(double* fission_rates, long num_FSRs, bool nu = false);
-  
+
   /// Missing implementations
   void resetFixedSources() {};
   void printCycle(long track_start, int domain_start, int length) {};

src/log.cpp

@@ -272,7 +272,7 @@ void set_log_level(const char* new_level) {
 
 /**
  * @brief Sets the minimum log message level which will be printed to the
- *        console and to the log file. This is an overloaded version to handle 
+ *        console and to the log file. This is an overloaded version to handle
  *        a logLevel type input.
  * @param new_level the minimum logging level as an int (or enum type logLevel)
  */
@@ -584,6 +584,10 @@ void log_printf(logLevel level, const char* format, ...) {
     log_file << msg_string;
     log_file.close();
 
+    // Need to make sure this gets initialized
+    static std::once_flag init_flag;
+    std::call_once(init_flag, []{ omp_init_lock(&log_error_lock); });
+
     /* Write the log message to the shell */
     if (level == ERROR) {
       omp_set_lock(&log_error_lock);