Tag

#polars

2 posts ·all posts

150M rows under a 4.5 GiB cap: SQE next to DuckDB, Polars and Spark

A response to "DuckDB 2.0 vs Polars vs Spark on one machine": we reran its engine-bench workload unchanged, with SQE added. Every answer matches DuckDB's and SQE finishes every run under the cap. One of its leads over DuckDB held up under a closer look and one did not. The benchmark also found two problems in SQE, and both are now fixed.

Read