Efficient all-to-all collective communication schedules for direct-connect topologies

P Basu, L Zhao, J Fantl, S Pal… - Proceedings of the 33rd …, 2024 - dl.acm.org
The all-to-all collective communications primitive is widely used in machine learning (ML)
and high performance computing (HPC) workloads, and optimizing its performance is of …

Fast and scalable all-optical network architecture for distributed deep learning

W Li, G Yuan, Z Wang, G Tan, P Zhang… - Journal of Optical …, 2024 - opg.optica.org
With the ever-increasing size of<? TeX 2pc 0pt?> training models and datasets, network
communication has emerged as a major bottleneck in distributed deep learning training. To …

Efficient neural network accelerators with optical computing and communication

C ** a wide
variety of real-life applications, from decision-making tasks, such as image classification and …