Welcome to Lightning, the next-generation photonic computing system.
Lightning is a reconfigurable photonic-electronic smartNIC that serves real-time deep neural network inference requests at 100 Gbps. Lightning uses a novel datapath that feeds traffic from the NIC into the photonic domain without creating digital packet processing and data movement bottlenecks. Lightning achieves this by employing a reconfigurable count-action abstraction, which keeps track of the computation Directed Acyclic Graph (DAG) of each inference packet. Our count-action abstraction decouples the compute control plane from the data plane by counting the number of operations in each task of the DAG and triggers the execution of the next tasks as soon as the previous task is finished, without interrupting the flow of the data.
Due to the count-action abstraction that seamlessly connect the photonic and electronic components, our Lightning prototypes performs photonic computing at the record-setting 4.055 GHz!!
To know more about this project:
- Read our full technical paper at the ACM SIGCOMM 2023: https://dl-acm-org.ezproxy.canberra.edu.au/doi/10.1145/3603269.3604821
- Join our Slack channel: #project-lightning
- Sign up for an open-source Lightning Dev Kit here: https://forms.gle/hMXgTdb8XoE5gYV69
- Try out our open-source code (including datapath RTL, photonic hardware control, kit fabrication, simulation, emulation) : https://github.com/open-photonics/lightning-lts
Technical publications
[1] Lightning: A Reconfigurable Photonic-Electronic SmartNIC for Fast and Energy-Efficient Inference
Zhizhen Zhong, Mingran Yang, Jay Lang, Christian Williams, Liam Kronman, Alexander Sludds, Homa Esfahanizadeh, Dirk Englund, Manya Ghobadi
Paper | Artifact | Demo | Slides | Video
MIT News article: System combines light and electrons to unlock faster, greener computing
[2] On-Fiber Photonic Computing
Mingran Yang*, Zhizhen Zhong*, Manya Ghobadi (co-first author)
[3] Delocalized Photonic Deep Learning on the Internet's Edge
Alexander Sludds, Saumil Bandyopadhyay, Zaijun Chen, Zhizhen Zhong, Jared Cochrane, Liane Bernstein, Darius Bunandar, P. Ben Dixon, Scott Hamilton, Matthew Streshinsky, Ari Novack, Tom Baehr-Jones, Michael Hochberg, Manya Ghobadi, Ryan Hamerly, Dirk Englund
[4] NetCast: Low-Power Edge Computing with WDM-defined Optical Neural Networks
Ryan Hamerly, Alexander Sludds, Saumil Bandyopadhyay, Zaijun Chen, Zhizhen Zhong, Liane Bernstein, Dirk Englund
[5] IOI: In-Network Optical Inference
Zhizhen Zhong, Weiyang Wang, Manya Ghobadi, Alexander Sludds, Ryan Hamerly, Liane Bernstein, Dirk Englund