Tiger Data
A data infrastructure company in the Postgres ecosystem, known for the time-series database TimescaleDB, focused on PostgreSQL extension development and large-scale data processing.
Timeline (last 90 days)
Tiger Data (the company behind TimescaleDB) has released an open-source Postgres extension named pg_vectorsearch, using a new index, PRISM.
At a scale of 1 million vectors, PRISM achieves 99% recall and 15 milliseconds median latency on a 2GB c8g.medium; HNSW on a 4GB c8g.large is 392 milliseconds, with building taking about 21.7 hours and exhibiting memory thrashing.
Tiger Data announced that pgvectorscale has entered maintenance mode; it remains stable and able to run production workloads, but its DiskANN-inspired graph index may lose recall on low-dimensional datasets.
The official announcement of pg_vectorsearch did not give the dimensionality of the benchmark dataset, the throughput (QPS), or a direct IVFFlat comparison, and multi-client throughput testing is listed as a to-be-completed item.
All Facts (4)
At a scale of 1 million vectors, PRISM achieves 99% recall and 15 milliseconds median latency on a 2GB c8g.medium; HNSW on a 4GB c8g.large is 392 milliseconds, with building taking about 21.7 hours and exhibiting memory thrashing.
50%UnverifiedThe official announcement of pg_vectorsearch did not give the dimensionality of the benchmark dataset, the throughput (QPS), or a direct IVFFlat comparison, and multi-client throughput testing is listed as a to-be-completed item.
50%UnverifiedTiger Data (the company behind TimescaleDB) has released an open-source Postgres extension named pg_vectorsearch, using a new index, PRISM.
50%UnverifiedTiger Data announced that pgvectorscale has entered maintenance mode; it remains stable and able to run production workloads, but its DiskANN-inspired graph index may lose recall on low-dimensional datasets.
50%