Scaling pgvector: Memory, Quantization, and Index Build Strategies (opens in new tab)

<h1> Scaling pgvector: Memory, Quantization, and Index Build Strategies </h1> <p>pgvector handles small-scale vector search effortlessly. A few hundred thousand embeddings with an HNSW index, and similarity queries return in milliseconds. But once you push past a million vectors, three problems converge -- and if you haven't planned for them, they hit at the same time.</p> <h2> Three Walls at Scale </h2> <h3> Wall 1: HNSW Index Builds Need Massive Memory </h3> <p>B...