EUVIMEDEUVIMED entdecken
Crossref

SpikeLet: Empowering Multichiplet Neuromorphic System With Hierarchical Interconnect and Spatial–Temporal Parallelism

Jiajun Lu, Xiaoyang Zeng, Chixiao Chen, Haozhe Zhu

IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems · 2026-11 · 45 (11) · 5272-5277

DOI: 10.1109/tcad.2026.3663301 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 2026-11. Verfasst von Jiajun Lu und 3 weiteren Autor:innen. Die Arbeit nennt 19 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Fundamental and Interdisciplinary Disciplines Breakthrough Plan of the Ministry of Education (MoE) of China, Shanghai 2024 “Science and Technology Innovation Action Plan” Fundamental Research Program in Integrated Circuit.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

As spiking neural networks (SNNs) scale, their neuron counts exceed the capacity of a single chip, necessitating multichip integration. Multichiplet systems provide dense, high-bandwidth interconnects but suffer from inefficient interchip spike transmission, limited parallelism, and the lack of transmission-aware mapping. We present SpikeLet, the first multichiplet neuromorphic system tailored for large-scale SNNs. SpikeLet employs a hierarchical transmission interconnect with an extend-AER protocol to aggregate spikes and decouple intrachiplet and interchiplet communication, an event-driven spatial–temporal-parallel dataflow that enables cross-layer and cross-timestep parallelism, and a communication-aware hierarchical mapping based on cooperative co-evolution to balance computation and communication across chiplets and cores. On a $4\times 4$ -chiplet prototype, SpikeLet reduces transmission latency by 74.2%, improves throughput by $36.6\times $ , and boosts energy efficiency by $39.2\times $ over a 2-D multichip baseline, demonstrating a compact and scalable platform for next-generation neuromorphic computing.

Förderung

  • National Natural Science Foundation of China
  • Fundamental and Interdisciplinary Disciplines Breakthrough Plan of the Ministry of Education (MoE) of China
  • Shanghai 2024 “Science and Technology Innovation Action Plan” Fundamental Research Program in Integrated Circuit

Zitieren

APA

Lu, J., Zeng, X., Chen, C., & Zhu, H. (2026). SpikeLet: Empowering Multichiplet Neuromorphic System With Hierarchical Interconnect and Spatial–Temporal Parallelism. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 45(11), 5272-5277. https://doi.org/10.1109/tcad.2026.3663301

Vancouver

Lu J, Zeng X, Chen C, Zhu H. SpikeLet: Empowering Multichiplet Neuromorphic System With Hierarchical Interconnect and Spatial–Temporal Parallelism. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems. 2026;45(11):5272-5277. doi:10.1109/tcad.2026.3663301

BibTeX
@article{Lu2026,
  author = {Lu, Jiajun and Zeng, Xiaoyang and Chen, Chixiao and Zhu, Haozhe},
  title = {SpikeLet: Empowering Multichiplet Neuromorphic System With Hierarchical Interconnect and Spatial–Temporal Parallelism},
  journal = {IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems},
  year = {2026},
  volume = {45},
  number = {11},
  pages = {5272--5277},
  doi = {10.1109/tcad.2026.3663301}
}
Kostenlos starten →