PlanetScale released
TIN, a full-text search extension for Postgres. It supports boolean/phrase/fuzzy/regex search, BM25 ranking, COUNT(*), and real-time updates. The core idea is to use Postgres's ctid directly as the index posting ID instead of a separate document ID. This removes the ID→ctid translation cost, and postings are compressed as page/offset bitmaps. These bitmaps are processed with AVX2/AVX-512 SIMD AND/OR/POPCNT, making conjunction, disjunction, and count queries extremely fast. In their own benchmarks, they claim roughly 10–57× higher QPS than ParadeDB and, on some workloads, hundreds of times faster than Postgres GIN, with little degradation under concurrent writes. The key point is that the search engine was designed from scratch around Postgres internals (ctid, visibility map), eliminating the abstraction cost of bolting on an external search engine.Introducing TIN: full-text search for Postgres — PlanetScale
TIN is a fast, full-featured, full-text search index for Postgres
https://planetscale.com/blog/introducing-tin


Seonglae Cho