SC20 Proceedings

The International Conference for High Performance Computing, Networking, Storage, and Analysis

Towards a Scalable and Distributed Infrastructure for Deep Learning Applications

Workshop:The 5th Deep Learning on Supercomputers Workshop

Authors: Bita Hasheminezhad, Shahrzad Shirzad, Nanmiao Wu, and Patrick Diehl (Louisiana State University, Center for Computation and Technology); Hannes Schulz (Microsoft Research, Montreal); and Hartmut Kaiser (Louisiana State University, Center for Computation and Technology)

Abstract: Although recent scaling up approaches to train deep neural networks have proven to be effective, the computational intensity of large and complex models, as well as the availability of large-scale datasets require deep learning frameworks to utilize scaling out techniques. Parallelization approaches and distribution requirements are not considered in the primary designs of most available distributed deep learning frameworks and most of them still are not able to perform effective and efficient fine-grained inter-node communication. We present Phylanx, which has the potential to alleviate these shortcomings. Phylanx presents a productivity-oriented frontend where user Python code is translated to a futurized execution tree that can be executed efficiently on multiple nodes using the C++ standard library for parallelism and concurrency (HPX), leveraging fine-grained threading and an active messaging task-based runtime system.

Back to The 5th Deep Learning on Supercomputers Workshop Archive Listing

Back to Full Workshop Archive Listing