Free-threaded Python 3.14t plus fastsimd parses JSON at 65.7 GB/s on 64 cores
lemire · x · 2026-10-12
Daniel Lemire benchmarked JSON parsing in free-threaded Python 3.14 (python3.14t, no GIL) on a realistic workload: 8 hours of GH Archive events (7.6 GB decompressed, 2.2M NDJSON records), counting events and pull requests by language.
- Single core: fastsimd mapmany with lazy views hits 2.13 GB/s — 12x the stdlib (0.18 GB/s); msgspec reaches 1.29 GB/s.
- 64 cores, one process, 64 threads: lazy-view mapmany reaches 65.7 GB/s, plain mapmany 29.8 GB/s, msgspec 24.6 GB/s, stdlib just 3.17 GB/s.
Takeaway: with the GIL gone and lazy views avoiding intermediate object materialization, multi-GB-per-second JSON processing is now practical in pure Python workflows.
More from Research
- Cameron Wolfe to talk at AIE NYC: a roadmap for evaluating highly-subjective LLM tasks — cwolferesearch · 2026-10-12
- Gaël Varoquaux compiles his AI blog series: a researcher's take on AI and society — GaelVaroquaux · 2026-10-12
- OpenAI's AI solves famous math problems—researchers recognize their own ideas in them — GaelVaroquaux · 2026-10-12
- LLMs predict edit sequences to optimize molecular synthesizability, new Nature MI paper — lulzxdxdxd · 2026-10-12
- Dexterity Is the Frontier: Chris Paxton's Field Notes from IROS 2026 — chris_j_paxton · 2026-10-12
- Decision Model, the typed-output alternative to free-form LLM text, lands on Papers with Code — NielsRogge · 2026-10-12