TeraNoC: A Multi-Channel 32-bit Fine-Grained, Hybrid Mesh-Crossbar NoC for Efficient Scale-up of 1000+ Core Shared-L1-Memory Clusters
Yichao Zhang, Zexin Fu, Tim Fischer, Yinrong Li, Marco Bertuletti, Luca Benini · 2025
Read TeraNoC: A Multi-Channel 32-bit Fine-Grained, Hybrid Mesh-Crossbar NoC for Efficient Scale-up of 1000+ Core Shared-L1-Memory Clusters by Yichao Zhang, Zexin Fu, Tim Fischer, Yinrong Li, Marco Bertuletti, Luca Benini (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-02446.txt.
Description
Yichao Zhang and co-authors present TeraNoC, a chip wiring scheme that lets a thousand processor cores share one fast memory without choking, in a 2025 paper. Two classic layouts fail at this scale: grids keep wires short but force long trips across the chip, while crossbars answer fast but grow in cost with every added port.
TeraNoC mixes the two, running several 32-bit channels of grid wiring studded with crossbars. A remapping router spreads traffic evenly across the channels. The team built a cluster of 1,024 small cores sharing a memory split into 4,096 banks, in a modern 12-nanometer process. On heavy AI-style kernels the cores stayed busy at high rates, and the wiring took a small slice of total power. The design is open source.
Catalog record
| CREATOR | Yichao Zhang, Zexin Fu, Tim Fischer, Yinrong Li, Marco Bertuletti, Luca Benini |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 42 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-02446