Morton filter: calculate memory usage (wasn't calculated before)

Author: thomasmueller

benchmarks/bulk-insert-and-query.cc

@@ -318,6 +318,7 @@ class MortonFilter {
     size_t size;
 public:
     MortonFilter(const size_t size) {
+        // filter = new Morton3_8((size_t) (size / 0.50) + 64);
         filter = new Morton3_8((size_t) (2.1 * size) + 64);
         this->size = size;
     }
@@ -331,8 +332,14 @@ class MortonFilter {
         return filter->likely_contains(item);
     };
     size_t SizeInBytes() const {
-        // TODO don't know how to get / calculate it
-        return size;
+        // according to morton_sample_configs.h:
+        // Morton3_8 - 3-slot buckets with 8-bit fingerprints: 11.7 bits/item
+        // (load factor = 0.95)
+        // so in theory we could just hardcode the size here,
+        // and don't measure it
+        // return (size_t)((size * 11.7) / 8);
+
+        return filter->SizeInBytes();
     }
 };
 

src/morton/compressed_cuckoo_filter.h

@@ -258,6 +258,8 @@ namespace CompressedCuckoo{
     // The number of times that we've doubled the filter's capacity
     uint_fast16_t _resize_count;
 
+    size_t sizeInBytes;
+
     friend Tester; // Class with a bunch of test functions in test.cc
 
   public:
@@ -275,6 +277,11 @@ namespace CompressedCuckoo{
     _resize_count(0)
   {
 
+    sizeInBytes = 0;
+    sizeInBytes += sizeof(bool) * (_block_fullness_array_enabled ? _total_blocks : 0);
+    sizeInBytes += sizeof(counter_t) * (_buckets_per_block + 1);
+    sizeInBytes += sizeof(block_t) * _total_blocks;
+
     // Supporting dual use as a compressed cuckoo filter and Morton filter
     static_assert((_morton_filter_functionality_enabled ^ (_ota_len_bits == 0)),
     "ERROR: If Morton filter functionality is enabled, then the overflow tracking array must be at least one bit in length.");
@@ -335,6 +342,9 @@ namespace CompressedCuckoo{
     }
   }
 
+  size_t SizeInBytes() {
+    return sizeInBytes;
+  }
 
   INLINE atom_t fingerprint_function(const hash_t raw_hash) const{
     atom_t fingerprint =