Add memory measures to benchmarks.

app/benchmark_fft/main.cpp

@@ -38,6 +38,7 @@ int main(int argc, char **argv) {
   string test_name = to_string(num_threads) + ".csv";
   string full_directory = directory + "/PLS_v3/";
   benchmark_runner runner{full_directory, test_name};
+  runner.enable_memory_stats();
 
   fft::complex_vector data(fft::SIZE);
   fft::complex_vector swap_array(fft::SIZE);

app/benchmark_fib/main.cpp

@@ -36,6 +36,7 @@ int main(int argc, char **argv) {
   string test_name = to_string(num_threads) + ".csv";
   string full_directory = directory + "/PLS_v3/";
   benchmark_runner runner{full_directory, test_name};
+  runner.enable_memory_stats();
 
   pls::scheduler scheduler{(unsigned) num_threads, MAX_NUM_TASKS, MAX_STACK_SIZE};
 

app/benchmark_matrix/main.cpp

@@ -31,6 +31,7 @@ int main(int argc, char **argv) {
   string test_name = to_string(num_threads) + ".csv";
   string full_directory = directory + "/PLS_v3/";
   benchmark_runner runner{full_directory, test_name};
+  runner.enable_memory_stats();
 
   pls_matrix<double, matrix::MATRIX_SIZE> a;
   pls_matrix<double, matrix::MATRIX_SIZE> b;

app/benchmark_matrix_div_conquer/main.cpp

@@ -112,6 +112,7 @@ int main(int argc, char **argv) {
   string test_name = to_string(num_threads) + ".csv";
   string full_directory = directory + "/PLS_v3/";
   benchmark_runner runner{full_directory, test_name};
+  runner.enable_memory_stats();
 
   // Only run on one version to avoid copy
   std::unique_ptr<double[]> result_data{new double[size * size]};

app/benchmark_unbalanced/main.cpp

@@ -42,6 +42,7 @@ int main(int argc, char **argv) {
   string test_name = to_string(num_threads) + ".csv";
   string full_directory = directory + "/PLS_v3/";
   benchmark_runner runner{full_directory, test_name};
+  runner.enable_memory_stats();
 
   scheduler scheduler{(unsigned) num_threads, MAX_NUM_TASKS, MAX_STACK_SIZE};
 

extern/benchmark_runner/benchmark_runner.h

@@ -29,6 +29,12 @@ class benchmark_runner {
   chrono::steady_clock::time_point last_start_time_;
   vector<long> times_;
 
+  bool memory_stats_enabled_{false};
+  const string MEMORY_PRE_RUN = "memory_pages_pre_run";
+  const string MEMORY_POST_RUN = "memory_pages_post_run";
+  unsigned long memory_pre_run_;
+  unsigned long memory_post_run_;
+
   map<string, vector<long>> custom_stats_;
 
   void print_statistics() {
@@ -96,6 +102,12 @@ class benchmark_runner {
     custom_stats_.insert({name, {}});
   }
 
+  void enable_memory_stats() {
+    memory_stats_enabled_ = true;
+    add_custom_stats_field(MEMORY_PRE_RUN);
+    add_custom_stats_field(MEMORY_POST_RUN);
+  }
+
   static void read_args(int argc, char **argv, int &num_threads, string &path) {
     if (argc < 3) {
       cout << "Must Specifiy concurrency and output directory! (usage: `benchmark <output_directory> <num_threads>`)"
@@ -109,17 +121,31 @@ class benchmark_runner {
   }
 
   void start_iteration() {
+    if (memory_stats_enabled_) {
+      auto memory_stats = query_process_memory_pages();
+      memory_pre_run_ = memory_stats.first;
+    }
+
     last_start_time_ = chrono::steady_clock::now();
   }
 
   void end_iteration() {
+    size_t iteration_index = times_.size();
+
     auto end_time = chrono::steady_clock::now();
     long time = chrono::duration_cast<chrono::microseconds>(end_time - last_start_time_).count();
-    times_.emplace_back(time);
 
+    times_.emplace_back(time);
     for (auto &iter : custom_stats_) {
       iter.second.emplace_back(0);
     }
+    if (memory_stats_enabled_) {
+      auto memory_stats = query_process_memory_pages();
+      memory_post_run_ = memory_stats.first;
+
+      custom_stats_[MEMORY_PRE_RUN][iteration_index] = memory_pre_run_;
+      custom_stats_[MEMORY_POST_RUN][iteration_index] = memory_post_run_;
+    }
   }
 
   void store_custom_stat(const string &name, long value) {