reserve estimated cardinality up front

duckdb · Apr 5, 2024 · 692913f · 692913f
1 parent f03fe75
commit 692913f
Show file tree

Hide file tree

Showing 3 changed files with 7 additions and 5 deletions.
diff --git a/src/hnsw/hnsw_index.cpp b/src/hnsw/hnsw_index.cpp
@@ -120,7 +120,8 @@ class LinkedBlockWriter {
 // Constructor
 HNSWIndex::HNSWIndex(const string &name, IndexConstraintType index_constraint_type, const vector<column_t> &column_ids,
                      TableIOManager &table_io_manager, const vector<unique_ptr<Expression>> &unbound_expressions,
-                     AttachedDatabase &db, const case_insensitive_map_t<Value> &options, const IndexStorageInfo &info)
+                     AttachedDatabase &db, const case_insensitive_map_t<Value> &options, const IndexStorageInfo &info,
+                     idx_t estimated_cardinality)
     : Index(name, TYPE_NAME, index_constraint_type, column_ids, table_io_manager, unbound_expressions, db) {
 
 	if (index_constraint_type != IndexConstraintType::NONE) {
@@ -173,7 +174,7 @@ HNSWIndex::HNSWIndex(const string &name, IndexConstraintType index_constraint_ty
 		index.load_from_stream(
 		    [&](void *data, size_t size) { return size == reader.ReadData(static_cast<data_ptr_t>(data), size); });
 	} else {
-		index.reserve(32);
+		index.reserve(MinValue(static_cast<idx_t>(32), estimated_cardinality));
 	}
 	index_size = index.size();
 }

diff --git a/src/hnsw/hnsw_index_physical_create.cpp b/src/hnsw/hnsw_index_physical_create.cpp
@@ -46,8 +46,9 @@ unique_ptr<GlobalSinkState> PhysicalCreateHNSWIndex::GetGlobalSinkState(ClientCo
 	auto &constraint_type = info->constraint_type;
 	auto &db = storage.db;
 
-	gstate->global_index = make_uniq<HNSWIndex>(info->index_name, constraint_type, storage_ids, table_manager,
-	                                            unbound_expressions, db, info->options);
+	gstate->global_index =
+	    make_uniq<HNSWIndex>(info->index_name, constraint_type, storage_ids, table_manager, unbound_expressions, db,
+	                         info->options, IndexStorageInfo(), estimated_cardinality);
 
 	return std::move(gstate);
 }

diff --git a/src/include/hnsw/hnsw_index.hpp b/src/include/hnsw/hnsw_index.hpp
@@ -30,7 +30,7 @@ class HNSWIndex : public Index {
 	HNSWIndex(const string &name, IndexConstraintType index_constraint_type, const vector<column_t> &column_ids,
 	          TableIOManager &table_io_manager, const vector<unique_ptr<Expression>> &unbound_expressions,
 	          AttachedDatabase &db, const case_insensitive_map_t<Value> &options,
-	          const IndexStorageInfo &info = IndexStorageInfo());
+	          const IndexStorageInfo &info = IndexStorageInfo(), idx_t estimated_cardinality = 0);
 
 	//! The actual usearch index
 	unum::usearch::index_dense_t index;