uv cuts peak RSS 15-20% on large solves via AI-assisted PRs, arena checkpointing trick
charliermarsh · x · 2026-10-23
uv author Charlie Marsh shares a side quest: with AI (Astra) he landed a few small PRs cutting uv's peak RSS on large dependency solves by 15-20%. On a 7,777-package project, fresh resolve dropped from 950.1 MiB to 766.4 MiB (-19.4%) and upgrades from 1,199.7 MiB to 1,014.9 MiB (-15.3%), via deferred forks (#22769), an arena checkpointing trick in the resolver (#22784), and incremental results (#22770).
Related event: uv Author Uses AI-Driven Small PRs to Cut Peak Memory 15-20%(3 posts)→
More from coding & agent
- PL researcher admits Opus 5.5 is already a better programmer than him — tweetsatpreet · 2026-10-25
- Free LLM Dojo launches: 50+ interactive labs and 490 AI-engineering interview questions — sharpeye_wnl · 2026-10-25
- mlx-decision runs decision models locally on Apple Silicon, up to 6x faster than PyTorch — armbues · 2026-10-25
- How do teams share memory across multiple AI agents? — belhelha · 2026-10-25
- Vibe Monitor ships after 41 beta iterations: local stdio MCP adds architecture checkpoints to coding agents — Either-Alternative32 · 2026-10-25
- KNOWS benchmark: best computer-use agent fully solves just 3.6% of tasks — anmarasovic · 2026-10-25