Instrumentation code to determine number of buffers in live and free

lists of GenomicsDBColumnarField

src/main/cpp/include/genomicsdb/genomicsdb_columnar_field.h

@@ -356,6 +356,9 @@ class GenomicsDBColumnarField
      */
     void print_data_in_buffer_at_index_as_csv(std::ostream& fptr,
         const GenomicsDBBuffer* buffer_ptr, const size_t index) const;
+    //Get list lengths
+    size_t get_free_buffer_list_length() const { return m_free_buffer_list_length; }
+    size_t get_live_buffer_list_length() const { return m_live_buffer_list_length; }
   private:
     void copy_simple_members(const GenomicsDBColumnarField& other);
     void assign_function_pointers();
@@ -371,6 +374,7 @@ class GenomicsDBColumnarField
         buffer_ptr->set_next_buffer(m_free_buffer_list_head_ptr);
       }
       m_free_buffer_list_head_ptr = buffer_ptr;
+      ++m_free_buffer_list_length;
     }
     GenomicsDBBuffer* create_new_buffer() const
     {
@@ -400,6 +404,9 @@ class GenomicsDBColumnarField
     GenomicsDBBuffer* m_live_buffer_list_tail_ptr;
     //Index of the element in the live buffer list head that must be read next
     size_t m_curr_index_in_live_buffer_list_tail;
+    //Length of live and free lists
+    size_t m_live_buffer_list_length;
+    size_t m_free_buffer_list_length;
 };
 
 #endif

src/main/cpp/src/genomicsdb/genomicsdb_columnar_field.cc

@@ -26,7 +26,8 @@
 GenomicsDBColumnarField::GenomicsDBColumnarField(const std::type_index element_type, const int length_descriptor,
     const unsigned fixed_length_field_length, const size_t num_bytes)
 : m_element_type(element_type), m_length_descriptor(length_descriptor),
-  m_fixed_length_field_num_elements(fixed_length_field_length)
+  m_fixed_length_field_num_elements(fixed_length_field_length),
+  m_free_buffer_list_length(0ull), m_live_buffer_list_length(0ull)
 {
   m_element_size = (element_type == std::type_index(typeid(bool)))
     ? sizeof(char)
@@ -58,6 +59,8 @@ void GenomicsDBColumnarField::copy_simple_members(const GenomicsDBColumnarField&
   m_print_csv = other.m_print_csv;
   m_buffer_size = other.m_buffer_size;
   m_curr_index_in_live_buffer_list_tail = other.m_curr_index_in_live_buffer_list_tail;
+  m_free_buffer_list_length = other.m_free_buffer_list_length;
+  m_live_buffer_list_length = other.m_live_buffer_list_length;
 }
 
 GenomicsDBColumnarField::GenomicsDBColumnarField(GenomicsDBColumnarField&& other)
@@ -311,6 +314,10 @@ void GenomicsDBColumnarField::move_buffer_to_live_list(GenomicsDBBuffer* buffer)
     m_free_buffer_list_head_ptr = next_in_free_list;
   //Reset idx pointed to in tail to 0
   m_curr_index_in_live_buffer_list_tail = 0u;
+  //Modify lengths
+  assert(m_free_buffer_list_length > 0u);
+  --m_free_buffer_list_length;
+  ++m_live_buffer_list_length;
 }
 
 void GenomicsDBColumnarField::move_buffer_to_free_list(GenomicsDBBuffer* buffer)
@@ -338,6 +345,9 @@ void GenomicsDBColumnarField::move_buffer_to_free_list(GenomicsDBBuffer* buffer)
   //If this was the head, update to next
   if(buffer == m_live_buffer_list_head_ptr)
     m_live_buffer_list_head_ptr = next_in_live_list;
+  //Modify lengths
+  ++m_free_buffer_list_length;
+  --m_live_buffer_list_length;
 }
 
 void GenomicsDBColumnarField::move_all_buffers_from_live_list_to_free_list()

src/main/cpp/src/genomicsdb/genomicsdb_iterators.cc

@@ -128,6 +128,15 @@ SingleCellTileDBIterator::~SingleCellTileDBIterator()
   for(auto i=0u;i<GenomicsDBIteratorStatsEnum::NUM_STATS;++i)
     std::cerr << "," << m_num_cells_traversed_stats[i];
   std::cerr << "\n";
+  auto num_live_list_entries = 0ull;
+  auto num_free_list_entries = 0ull;
+  for(const auto& field : m_fields)
+  {
+    num_free_list_entries += field.get_free_buffer_list_length();
+    num_live_list_entries += field.get_live_buffer_list_length();
+  }
+  std::cerr << "Buffer_lists_lengths final "<<num_free_list_entries
+    <<" "<<num_live_list_entries <<"\n";
   std::cerr << "Histogram:\n";
   for(auto val : m_num_cells_traversed_in_find_intersecting_intervals_mode_histogram)
     std::cerr << val << "\n";
@@ -329,6 +338,17 @@ void SingleCellTileDBIterator::read_from_TileDB(TileDB_CTX* tiledb_ctx, const ch
           break;
       }
       m_in_find_intersecting_intervals_mode = false;
+#ifdef DO_PROFILING
+      auto num_live_list_entries = 0ull;
+      auto num_free_list_entries = 0ull;
+      for(const auto& field : m_fields)
+      {
+        num_free_list_entries += field.get_free_buffer_list_length();
+        num_live_list_entries += field.get_live_buffer_list_length();
+      }
+      std::cerr << "Buffer_lists_lengths end_of_intersecting_intervals_mode "<<num_free_list_entries
+        <<" "<<num_live_list_entries <<"\n";
+#endif
     }
     //If done reading current column interval, move to next query column interval, if any remaining OR
     //this stack was performing intersecting intervals search