Metraipolis › archive › dc_622329f132fa
Fengshui: Demystifying Chiplet Ecosystem and Bespoke Neural Network Accelerator Codesign — arXiv:2609.10970
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.10970
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T05:44:55.516Z
- SHA-256 of the archived text
8893e535cc7ae282b2243e4a6bf4a23903fbc6a7fb094fb13b1d77282d37e2be- Stable URL of this page
https://metraipolis.com/documento/dc_622329f132fa
How to cite
Fengshui: Demystifying Chiplet Ecosystem and Bespoke Neural Network Accelerator Codesign — arXiv:2609.10970 — arXiv (en), https://arxiv.org/abs/2609.10970 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_622329f132fa (Metraipolis archive dc_622329f132fa, sha256 8893e535cc7ae282b2243e4a6bf4a23903fbc6a7fb094fb13b1d77282d37e2be)
BibTeX
@misc{dc622329f132fa,
title = {Fengshui: Demystifying Chiplet Ecosystem and Bespoke Neural Network Accelerator Codesign — arXiv:2609.10970},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.10970},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_622329f132fa; sha256 8893e535cc7ae282b2243e4a6bf4a23903fbc6a7fb094fb13b1d77282d37e2be},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.10970. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Modern ML workloads, with stringent latency and energy constraints, are increasingly hard to run efficiently on homogeneous commodity hardware. We argue that operator-level disaggregation--tailoring microarchitecture, batching, and memory hierarchy to each operator--is essential to overcome these limitations, though the resulting highly bespoke accelerators incur prohibitive Non-Recurring Engineering (NRE) costs. Chiplet-based integration amortizes NRE across applications, but choosing which chiplets to build and how to compose them into accelerators is circularly dependent--a chiplet pool's value depends on the constructed accelerators, while accelerator quality is constrained by available chiplets. This paper introduces Fengshui, a chiplet ecosystem and accelerator co-design framework that jointly optimizes chiplet pool composition and bespoke application-specific integrated circuit (BASIC) design. Fengshui constructs BASICs through operator-level disaggregation, co-exploring chiplet and memory heterogeneity, tensor fusion, and pipeline/tensor/expert parallelism with place-and-route validation for physical implementability. With just 8 strategically selected chiplets, encompassing network switches, processing-in-memory units, and accelerators with diverse microarchitectures, Fengshui-generated BASICs achieve 48.5%, 88.1%, 93.0%, and 97.8% reductions in energy, energy-cost product (EC), energy-delay product (EDP), and energy-delay-cost product (EDPC) over homogeneous accelerators, while scoring within 4.1% of unconstrained heterogeneous designs across diverse neural networks. For datacenter MoE and dense LLM serving, Fengshui reduces prefill energy and EC by up to 16.8% and 28.7%, respectively; for edge autonomous vehicle perception, it achieves 12.0% energy and 23.6% EC reductions under real-time latency constraints. Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-09-11 in cs.AR). Categorie arXiv: Hardware Architecture (cs.AR). Pagina del lavoro: https://arxiv.org/abs/2609.10970 — 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 2026-10-16T05:26:59.334Z dalla pagina https://arxiv.org/catchup/cs.AR/2026-09-11?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).