Sign up for our newsletter and get the latest HPC news and analysis.
Send me information from insideHPC:


Scalable and Distributed DNN Training on Modern HPC Systems

DK Panda, Ohio State University

In this video from the Swiss HPC Conference, DK Panda from Ohio State University presents: Scalable and Distributed DNN Training on Modern HPC Systems.

The current wave of advances in Deep Learning (DL) has led to many exciting challenges and opportunities for Computer Science and Artificial Intelligence researchers alike. Modern DL frameworks like Caffe2, TensorFlow, Cognitive Toolkit (CNTK), PyTorch, and several others have emerged that offer ease of use and flexibility to describe, train, and deploy various types of Deep Neural Networks (DNNs). In this talk, we will provide an overview of interesting trends in DNN design and how cutting-edge hardware architectures are playing a key role in moving the field forward. We will also present an overview of different DNN architectures and DL frameworks. Most DL frameworks started with a single-node/single-GPU design. However, approaches to parallelize the process of DNN training are also being actively explored. The DL community has moved along different distributed training designs that exploit communication runtimes like gRPC, MPI, and NCCL. In this context, we will highlight new challenges and opportunities for communication runtimes to efficiently support distributed DNN training. We also highlight some of our co-design efforts to utilize CUDA-Aware MPI for large-scale DNN training on modern GPU clusters.

Dr. Dhabaleswar K. (DK) Panda is a Professor and Distinguished Scholar of Computer Science at the Ohio State University. He obtained his Ph.D. in computer engineering from the University of Southern California. His research interests include parallel computer architecture, high performance networking, InfiniBand, network-based computing, exascale computing, programming models, GPUs and accelerators, high performance file systems and storage, virtualization and cloud computing and BigData (Hadoop (HDFS, MapReduce and HBase) and Memcached). He has published over 400 papers in major journals and international conferences related to these research areas.

See more talks from the Swiss HPC Conference

Check out our insideHPC Events Calendar

Leave a Comment

*

Resource Links: