parallelize synapse generation using OpenMP

- only tested on Debian Linux with a 16-thread amd64 CPU
- Windows may need DLLs to be shipped with the EXE
- using `-fopenmp` should be made optional via qmake somehow
  (without the flag, parallelization pragmas are ignored
  and it reverts to serial operation)
- assumes `status::update()` among other code is thread-safe,
  it seems to work on my machine without issues...

brain/src/brain.cpp

@@ -16,6 +16,7 @@
 
 #include <iostream>
 #include <algorithm>
+#include <atomic>
 #include <sndfile.h>
 #include <float.h>
 #include <spiralcore/audio.h>
@@ -242,10 +243,12 @@ void brain::build_synapses_thresh(search_params &params, double thresh) {
   m_average_error = calc_average_diff(params)*thresh;
   double err = m_average_error*thresh;
   u32 brain_size = m_blocks.size();
-  u32 outer_index = 0;
-  for (auto &i : m_blocks) {
+  std::atomic<u32> progress{0};
+  #pragma omp parallel for
+  for (u32 outer_index = 0; outer_index < brain_size; ++outer_index) {
+    auto &i = m_blocks[outer_index];
     u32 index = 0;
-    status::update("building synapses %d%%",(int)(outer_index/(float)brain_size*100));
+    status::update("building synapses %d%%",(int)(progress/(float)brain_size*100));
     for (auto &j : m_blocks) {
       if (index!=outer_index) {
         // collect connections that are under threshold in closeness
@@ -256,30 +259,32 @@ void brain::build_synapses_thresh(search_params &params, double thresh) {
       }
       ++index;
     }
-    ++outer_index;
+    ++progress;
   }  
 }
 
 void brain::build_synapses_fixed(search_params &params) {
   //m_average_error = calc_average_diff(params)*thresh;
   u32 brain_size = m_blocks.size();
-  u32 outer_index = 0;
   u32 num_synapses = NUM_FIXED_SYNAPSES;
   if (num_synapses>=m_blocks.size()) num_synapses=m_blocks.size()-1;
 
   // need to stop the progress updates flooding osc
   u32 update_period = 100;
-  u32 update_tick = 0;
-
-  for (auto &i:m_blocks) {
+  std::atomic<u32> update_tick{0};
+  std::atomic<u32> progress{0};
+  #pragma omp parallel for
+  for (u32 outer_index = 0; outer_index < brain_size; ++outer_index) {
+    auto &i = m_blocks[outer_index];
     if (update_tick>update_period) {
-      status::update("building synapses %d%%",(int)(outer_index/(float)brain_size*100));     
+      status::update("building synapses %d%%",(int)(progress/(float)brain_size*100));
       update_tick=0;
     }
     update_tick++;
 
     u32 index = 0;
     vector<pair<u32,double>> collect;
+    collect.reserve(brain_size);
 
     // collect comparisons to all other blocks
     for (auto &j:m_blocks) {
@@ -304,7 +309,7 @@ void brain::build_synapses_fixed(search_params &params) {
       i.get_synapse().push_back(collect[n].first);
     }
 
-    ++outer_index;
+    ++progress;
   }
   status::update("Done: %d synapses grown for %d blocks",num_synapses*brain_size,brain_size);
 }

samplebrain.pro

@@ -43,9 +43,9 @@ SOURCES += app/MainWindow.cpp \
 INCLUDEPATH += brain/src
 INCLUDEPATH += /usr/local/include
 INCLUDEPATH += /opt/local/include
-LIBS += -L.. -L/usr/local/lib -L/opt/local/lib -lportaudio -lfftw3 -lsndfile -llo -ldl -lpthread -lm
+LIBS += -L.. -L/usr/local/lib -L/opt/local/lib -lportaudio -lfftw3 -lsndfile -llo -ldl -lpthread -lm -fopenmp
 
-QMAKE_CXXFLAGS += -O3 -march=core2 -Wall -Wno-unused -std=c++11
+QMAKE_CXXFLAGS += -O3 -march=core2 -fopenmp -Wall -Wno-unused -std=c++11
 
 # assets
 RESOURCES     = app/samplebrain.qrc