Metraipolis › archive › dc_570d171e6bb2
ePACT: Energy-Performance-Aware Commitment Tracking for LLM Serving — arXiv:2610.01784
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.01784
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-06T14:02:54.297Z
- SHA-256 of the archived text
f217bb06afe5e5ae05777b6dd1955e11cf167eb25726b365562c70cb8fe7e86d- Stable URL of this page
https://metraipolis.com/documento/dc_570d171e6bb2
How to cite
ePACT: Energy-Performance-Aware Commitment Tracking for LLM Serving — arXiv:2610.01784 — arXiv (en), https://arxiv.org/abs/2610.01784 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_570d171e6bb2 (Metraipolis archive dc_570d171e6bb2, sha256 f217bb06afe5e5ae05777b6dd1955e11cf167eb25726b365562c70cb8fe7e86d)
BibTeX
@misc{dc570d171e6bb2,
title = {ePACT: Energy-Performance-Aware Commitment Tracking for LLM Serving — arXiv:2610.01784},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.01784},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_570d171e6bb2; sha256 f217bb06afe5e5ae05777b6dd1955e11cf167eb25726b365562c70cb8fe7e86d},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.01784. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Reducing LLM serving energy does not by itself guarantee lower deployment cost when electricity procurement exposes operators to unfavorable deviations from preset commitments. We study hourly commitments with positive, potentially asymmetric costs for overuse and underuse, and formulate energy-Performance-Aware Commitment Tracking: minimize deviation costs subject to request-level service requirements. We implement ePACT, a two-level controller that adjusts serving capacity and GPU clocks as requests arrive. A global planner updates interval energy targets from measured consumption and the remaining hourly commitment. A local decision maker predicts candidate configurations' energy and completion times, checks predicted deadline misses, and selects among admitted configurations by asymmetric target-deviation cost, with a service-first fallback. Coarse-to-fine action search runs asynchronously with serving. We evaluate ePACT through single-hour comparisons, controller ablations, and full-day trace simulations for H20 and H200 GPU pools. In the 24-hour simulations, ePACT reduces the asymmetric deviation cost by $73.8\%$ and $75.7\%$ relative to vLLM while retaining near-vLLM SLO attainment. Mean absolute hourly deviations are $2.16\%$ and $2.31\%$, respectively. Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-10-02 in cs.DC). Categorie arXiv: Distributed, Parallel, and Cluster Computing (cs.DC). Pagina del lavoro: https://arxiv.org/abs/2610.01784 — 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 2026-10-06T12:22:22.563Z dalla pagina https://arxiv.org/catchup/cs.DC/2026-10-02?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).