Accelerate Distributed Deep Learning With a Fast Reconfigurable Optical Network (W2B.23)
Presenter: Wenzhe Li, Institute of Computing Technology, CAS
We propose a fast-reconfigurable and scalable optical network architecture, which employs a flow-based transmit scheduling scheme to accelerate data parallelism in distributed deep learning. Experimental results demonstrate that the 4-node prototype achieves training times comparable to those of ideal electrical switching.
Authors:Wenzhe Li, Institute of Computing Technology, CAS / Guojun Yuan, Institute of Computing Technology, CAS / Zhan Wang, Institute of Computing Technology, CAS / Guangming Tan, Institute of Computing Technology, CAS / Peiheng Zhang, Institute of Intelligent Computing Technology, Suzhou, CAS / George Rouskas, North Carolina State University