Polars 2.0 ships: streaming by default, SQL first-class, beats DuckDB on TPC-H
banteg · x · 2026-10-06
Polars 2.0 officially released
Announced by Ritchie Vink, Polars 2.0 removes many legacy decisions. Highlights:
- Streaming engine is now the default, with initial out-of-core (spill-to-disk) support for datasets larger than memory;
- SQL promoted to a first-class citizen, with dramatically expanded SQL coverage plus optimizer improvements (join reordering, much better common-subplan elimination, dynamic predicates/bloom filters);
- A new Map dtype, and stricter dtype/explicitness rules for faster feedback and faster AI iteration;
- Many core performance improvements.
Benchmarks
On TPC-H and TPC-DS1-derived data, the team benchmarked against DuckDB 1.5.6, DuckDB 2.0 alpha, and DataFusion 54.0 (c7a.4xlarge and c7a.metal, 5 hot runs per query, separate process, 60s timeout). Polars SQL leads both DuckDB and DataFusion, and the team claims it is now among the fastest single-node analytical SQL engines.
More from Infra
- Google Buys 890 MW of Nuclear Without Building a Single New Reactor — MicahBerkley · 2026-10-06
- Cycle.io Launches DevOps MCP: 3-Node Mongo Replica Set Across 3 Clouds in 15 Minutes — AlexMattoni · 2026-10-06
- Weaviate ships query profiling: a 48ms slow query turned out to be disk reads, not HNSW — CShorten30 · 2026-10-06
- HN Debate: Did Oracle Just Trigger the Implosion of the AI Bubble? — mpweiher · 2026-10-06
- Lambda adopts NVIDIA's AIPerf for model cards showing real-workload inference benchmarks — TheZachMueller · 2026-10-06
- Nvidia nears $6 trillion market value as AI frenzy keeps pushing stocks higher — AryHHAry · 2026-10-06