High-performance ANN vector search engine in modern C++ implementing HNSW, IVF, PQ, SQ8, IVF-SQ, and IVF-PQ with SIFT1M benchmarking.

A comprehensive, multiprocessing-isolated benchmark for evaluating vector database performance and quality. Measures insertion speed, search latency (diverse, sequential, filtered, and bulk), recall accuracy, and memory usage across standard (ChromaDB, LanceDB, Qdrant, FAISS, USearch) and custom engine implementations.

Comparison of Hierarchical K-Means and HNSW for approximate nearest neighbor search on SIFT1M