AIGoogle Cloud Blogabout 4 hours ago
AlloyDB accelerates pgvector HNSW 4x
6 min read
AlloyDB's columnar engine accelerated HNSW, now in preview, delivers up to 4x higher queries per second for vector search compared to standard PostgreSQL HNSW. The feature pins the pgvector HNSW index in a specialized in-memory cache, bypassing buffer manager overhead. At a fixed QPS of ~350, recall improves from 0.78 to over 0.94.
Level
Hype check
Tap to vote and see what everyone thinks.
Summary by ByteBrief
