Rebuilding Postgres for 300x faster analytics: batching, operator fusion, and SIMD
Summary
A blog post by malisper.me describes rebuilding Postgres analytics performance with pgrust, claiming up to 300x faster analytics than Postgres via batching, operator fusion, and SIMD. The article walks through a miniature Postgres-like query engine (Volcano model), introduces batching to reduce per-row overhead, demonstrates operator fusion to inline work, discusses potential future use of JIT and SIMD, and provides benchmark results and setup.