test xindex size inaccurately

Author: olemon111

include/alex/alex.h

@@ -121,13 +121,13 @@ namespace alex
       int num_sideways_splits = 0;
       int num_model_node_expansions = 0;
       int num_model_node_splits = 0;
-      long long num_downward_split_keys = 0;
-      long long num_sideways_split_keys = 0;
-      long long num_model_node_expansion_pointers = 0;
-      long long num_model_node_split_pointers = 0;
-      mutable long long num_node_lookups = 0;
-      mutable long long num_lookups = 0;
-      long long num_inserts = 0;
+      double num_downward_split_keys = 0;
+      double num_sideways_split_keys = 0;
+      double num_model_node_expansion_pointers = 0;
+      double num_model_node_split_pointers = 0;
+      mutable double num_node_lookups = 0;
+      mutable double num_lookups = 0;
+      double num_inserts = 0;
       double splitting_time = 0;
       double cost_computation_time = 0;
     };
@@ -2697,9 +2697,9 @@ namespace alex
     size_t max_size() const { return size_t(-1); }
 
     // Size in bytes of all the keys, payloads, and bitmaps stored in this index
-    long long data_size() const
+    double data_size() const
     {
-      long long size = 0;
+      double size = 0;
       for (NodeIterator node_it = NodeIterator(this); !node_it.is_end();
            node_it.next())
       {
@@ -2714,9 +2714,9 @@ namespace alex
 
     // Size in bytes of all the model nodes (including pointers) and metadata in
     // data nodes
-    long long model_size() const
+    double model_size() const
     {
-      long long size = 0;
+      double size = 0;
       for (NodeIterator node_it = NodeIterator(this); !node_it.is_end();
            node_it.next())
       {

include/alex/alex_base.h

@@ -255,7 +255,7 @@ class ExpectedShiftsAccumulator : public StatAccumulator {
   // Therefore, we track n^2/4.
   void accumulate(int actual_position, int) override {
     if (actual_position > last_position_ + 1) {
-      long long dense_region_length = last_position_ - dense_region_start_idx_ + 1;
+      double dense_region_length = last_position_ - dense_region_start_idx_ + 1;
       num_expected_shifts_ += (dense_region_length * dense_region_length) / 4;
       dense_region_start_idx_ = actual_position;
     }
@@ -266,8 +266,8 @@ class ExpectedShiftsAccumulator : public StatAccumulator {
   double get_stat() override {
     if (count_ == 0) return 0;
     // first need to accumulate statistics for current packed region
-    long long dense_region_length = last_position_ - dense_region_start_idx_ + 1;
-    long long cur_num_expected_shifts =
+    double dense_region_length = last_position_ - dense_region_start_idx_ + 1;
+    double cur_num_expected_shifts =
         num_expected_shifts_ + (dense_region_length * dense_region_length) / 4;
     return cur_num_expected_shifts / static_cast<double>(count_);
   }
@@ -282,7 +282,7 @@ class ExpectedShiftsAccumulator : public StatAccumulator {
  public:
   int last_position_ = -1;
   int dense_region_start_idx_ = 0;
-  long long num_expected_shifts_ = 0;
+  double num_expected_shifts_ = 0;
   int count_ = 0;
   int data_capacity_ = -1;  // capacity of node
 };
@@ -299,7 +299,7 @@ class ExpectedIterationsAndShiftsAccumulator : public StatAccumulator {
         std::log2(std::abs(predicted_position - actual_position) + 1);
 
     if (actual_position > last_position_ + 1) {
-      long long dense_region_length = last_position_ - dense_region_start_idx_ + 1;
+      double dense_region_length = last_position_ - dense_region_start_idx_ + 1;
       num_expected_shifts_ += (dense_region_length * dense_region_length) / 4;
       dense_region_start_idx_ = actual_position;
     }
@@ -320,8 +320,8 @@ class ExpectedIterationsAndShiftsAccumulator : public StatAccumulator {
 
   double get_expected_num_shifts() {
     if (count_ == 0) return 0;
-    long long dense_region_length = last_position_ - dense_region_start_idx_ + 1;
-    long long cur_num_expected_shifts =
+    double dense_region_length = last_position_ - dense_region_start_idx_ + 1;
+    double cur_num_expected_shifts =
         num_expected_shifts_ + (dense_region_length * dense_region_length) / 4;
     return cur_num_expected_shifts / static_cast<double>(count_);
   }
@@ -338,7 +338,7 @@ class ExpectedIterationsAndShiftsAccumulator : public StatAccumulator {
   double cumulative_log_error_ = 0;
   int last_position_ = -1;
   int dense_region_start_idx_ = 0;
-  long long num_expected_shifts_ = 0;
+  double num_expected_shifts_ = 0;
   int count_ = 0;
   int data_capacity_ = -1;  // capacity of node
 };

include/alex/alex_nodes.h

@@ -61,7 +61,7 @@ class AlexNode {
   virtual ~AlexNode() = default;
 
   // The size in bytes of all member variables in this class
-  virtual long long node_size() const = 0;
+  virtual double node_size() const = 0;
 };
 
 template <class T, class P, class Alloc = std::allocator<std::pair<T, P>>>
@@ -143,8 +143,8 @@ class AlexModelNode : public AlexNode<T, P> {
     return pointer_alloc_type(allocator_);
   }
 
-  long long node_size() const override {
-    long long size = sizeof(self_type);
+  double node_size() const override {
+    double size = sizeof(self_type);
     size += num_children_ * sizeof(AlexNode<T, P>*);  // pointers to children
     return size;
   }
@@ -349,8 +349,8 @@ class AlexDataNode : public AlexNode<T, P> {
       sizeof(V);  // cannot expand beyond this number of key/data slots
 
   // Counters used in cost models
-  long long num_shifts_ = 0;                 // does not reset after resizing
-  long long num_exp_search_iterations_ = 0;  // does not reset after resizing
+  double num_shifts_ = 0;                 // does not reset after resizing
+  double num_exp_search_iterations_ = 0;  // does not reset after resizing
   int num_lookups_ = 0;                      // does not reset after resizing
   int num_inserts_ = 0;                      // does not reset after resizing
   int num_resizes_ = 0;  // technically not required, but nice to have
@@ -2221,11 +2221,11 @@ class AlexDataNode : public AlexNode<T, P> {
   /*** Stats ***/
 
   // Total size of node metadata
-  long long node_size() const override { return sizeof(self_type); }
+  double node_size() const override { return sizeof(self_type); }
 
   // Total size in bytes of key/payload/data_slots and bitmap
-  long long data_size() const {
-    long long data_size = data_capacity_ * sizeof(T);
+  double data_size() const {
+    double data_size = data_capacity_ * sizeof(T);
     data_size += data_capacity_ * sizeof(P);
     data_size += bitmap_size_ * sizeof(uint64_t);
     return data_size;

include/xindex/xindex.h

@@ -52,6 +52,7 @@ class XIndex {
   size_t range_scan(const key_t &begin, const key_t &end,
                     std::vector<std::pair<key_t, val_t>> &result,
                     const uint32_t worker_id);
+  double index_size();
  private:
   void start_bg();
   void terminate_bg();

include/xindex/xindex_group.h

@@ -113,6 +113,7 @@ namespace xindex
 
     void free_data();
     void free_buffer();
+    double group_size();
 
   private:
     inline size_t locate_model(const key_t &key);