PostgreSQL extension claims vector search index faster than pgvector's HNSW
Pg_vectorsearch: An index that is faster than pgvector's IVFFlat and HNSW
Timescale's pg_vectorsearch introduces PRISM, a PostgreSQL-native index for approximate nearest neighbor search. It uses RaBitQ quantization with error bounds and SIMD-optimized distance computation to outperform pgvector's IVFFlat and HNSW. On a 100-million-vector dataset, PRISM achieves equivalent recall faster than HNSW while using a quarter of the memory. The extension is in alpha, compatible with existing pgvector columns, and supports dynamic updates.
Compared to graph-based HNSW, PRISM is faster at equivalent recall on a 100-million-vector dataset, on an instance with a quarter of the memory.