Metraipolis › archive › dc_024d521ad140
Beacon: LLM Multi-Agent Driven Hardware Design Space Exploration for Heterogeneous Multi-Chiplet Deep Learning Accelerators — arXiv:2608.30932
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.30932
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-17T08:19:11.358Z
- SHA-256 of the archived text
8bf7ed338be04ec7d7ef6a473a18db708c86459afa3916938d383ed1e1d933b2- Stable URL of this page
https://metraipolis.com/documento/dc_024d521ad140
How to cite
Beacon: LLM Multi-Agent Driven Hardware Design Space Exploration for Heterogeneous Multi-Chiplet Deep Learning Accelerators — arXiv:2608.30932 — arXiv (en), https://arxiv.org/abs/2608.30932 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_024d521ad140 (Metraipolis archive dc_024d521ad140, sha256 8bf7ed338be04ec7d7ef6a473a18db708c86459afa3916938d383ed1e1d933b2)
BibTeX
@misc{dc024d521ad140,
title = {Beacon: LLM Multi-Agent Driven Hardware Design Space Exploration for Heterogeneous Multi-Chiplet Deep Learning Accelerators — arXiv:2608.30932},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.30932},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_024d521ad140; sha256 8bf7ed338be04ec7d7ef6a473a18db708c86459afa3916938d383ed1e1d933b2},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.30932. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Heterogeneous multi-chiplet accelerators allow chiplets to be configured independently to better match different operator characteristics and improve inference efficiency. However, heterogeneity makes simulator evaluation expensive, limiting the number of iterations affordable for hardware design space exploration (HW-DSE). Mainstream data-driven methods rely mainly on final metrics and a few predefined states, and require many search iterations to implicitly learn the relationships between input parameters and optimization objectives, making them less effective in this setting. In practice, evaluators also generate detailed reports on execution timelines, resource utilization, memory accesses, and communication behavior. Large language models (LLMs) can combine domain knowledge with these reports to explicitly identify bottleneck locations, degradation causes, and parameter adjustment directions, thereby improving each design decision under limited iteration budgets. Based on this observation, we propose Beacon, a report-driven LLM multi-agent framework for heterogeneous multi-chiplet HW-DSE. Beacon employs hierarchical agents for bottleneck localization, root-cause diagnosis, and hardware candidate generation, together with an Analysis Toolbox and RAG memory for closed-loop search. Under the same limited iteration budget, Beacon reduces the composite latency-energy-monetary-cost objective by 25.1\%--93.5\% compared with random search, Bayesian optimization, and reinforcement learning. Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-09-01 in cs.AR). Categorie arXiv: Hardware Architecture (cs.AR). Pagina del lavoro: https://arxiv.org/abs/2608.30932 — 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 2026-10-17T08:07:25.704Z dalla pagina https://arxiv.org/catchup/cs.AR/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).