Metraipolis › archive › dc_dfbf8fe87ab3
Chiron: Cache Management Empowered by LLM Intelligence for AI Workloads — arXiv:2611.06179
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.06179
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-05T18:19:21.611Z
- SHA-256 of the archived text
e731f33aeb4011e06e9d517bfbe1d4b79660f330109fb6c7851907907dc6899a- Stable URL of this page
https://metraipolis.com/documento/dc_dfbf8fe87ab3
How to cite
Chiron: Cache Management Empowered by LLM Intelligence for AI Workloads — arXiv:2611.06179 — arXiv (en), https://arxiv.org/abs/2611.06179 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_dfbf8fe87ab3 (Metraipolis archive dc_dfbf8fe87ab3, sha256 e731f33aeb4011e06e9d517bfbe1d4b79660f330109fb6c7851907907dc6899a)
BibTeX
@misc{dcdfbf8fe87ab3,
title = {Chiron: Cache Management Empowered by LLM Intelligence for AI Workloads — arXiv:2611.06179},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.06179},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_dfbf8fe87ab3; sha256 e731f33aeb4011e06e9d517bfbe1d4b79660f330109fb6c7851907907dc6899a},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.06179. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
AI clusters keep data in remote object stores and rely on local caches to stop GPUs from waiting for data. A cache performs well only when its prefetching and eviction policies match the access pattern, yet these patterns differ across AI workloads and change between phases of one workflow. Existing methods either need code changes, which compromises user transparency, or use low-end heuristics, which cannot follow such changes. Large language models (LLMs) can infer access behavior without user hints, but direct LLM control of cache actions suffers from long reasoning delay, high token cost, and poor decision quality. We present Chiron, a cache system based on advisory switching control. An LLM asynchronously diagnoses the access pattern and recommends a policy. A deterministic harness gates this advice, executes expert-designed policies locally, and returns cache feedback for review. We implement Chiron atop JuiceFS and evaluate it on nine AI workloads. It reduces job completion time by 14.3-41.2% over the best baseline on popular AI workloads. Tema degli esploratori: calcolo e infrastrutture (elenco arXiv cs.DC, nuovi lavori del giorno). Categorie arXiv: Distributed, Parallel, and Cluster Computing (cs.DC) ; Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2611.06179 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-11-05T18:13:18.988Z dalla pagina https://arxiv.org/list/cs.DC/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).