3-7 December 2017
Velmoré Hotel Estate
Africa/Johannesburg timezone

Accelerating Big Data Processing and Associated Deep Learning on Modern HPC Clusters and Clouds

4 Dec 2017, 14:30
30m
De Voir (Velmoré Hotel Estate)

De Voir

Velmoré Hotel Estate

96 Main Road (M26) Hennops River Erasmia
Talk HPC Techniques and Computer Science Cognitive Computing and Machine Learning

Speaker

Dr Xiaoyi Lu (The Ohio State University)

Description

The convergence of HPC, Big Data, and Deep Learning is becoming the next game-changing business opportunity. Apache Hadoop, Spark, gRPC/TensorFlow, and Memcached are becoming standard building blocks in handling Big Data oriented processing and mining. Modern HPC bare-metal systems and Cloud Computing platforms have been fueled by the advances in multi-/many-core architectures, RDMA-enabled networking, NVRAMs, and NVMe-SSDs during the last decade. This talk will provide an in-depth overview of the architecture of Hadoop, Spark, gRPC/TensorFlow, Memcached, and the associated Deep Learning frameworks. We will examine the challenges in re-designing networking and I/O components of these middleware with modern interconnects, protocols (such as InfiniBand, RoCE) and storage architectures. This talk will provide case studies of the new designs for several Hadoop/Spark/gRPC/TensorFlow/Memcached components and their associated benefits. Through these, we will also examine the interplay between high-performance interconnects, storage (HDD, NVM, and SSD), and multi-core platforms (e.g., Xeon x86, OpenPOWER) to achieve the best solutions for these components and applications on modern HPC clusters and clouds. We also present in-depth case-studies with modern Deep Learning tools (e.g., Caffe, TensorFlow, BigDL) running over RDMA-enabled Hadoop, Spark, and gRPC.

HPC content

The convergence of HPC, Big Data, and Deep Learning is becoming the next game-changing business opportunity. Modern HPC bare-metal systems and Cloud Computing platforms have been fueled by the advances in multi-/many-core architectures, RDMA-enabled networking, NVRAMs, and NVMe-SSDs during the last decade. This talk will discuss how to re-design networking and I/O components of the popular Big Data processing middleware and associated Deep Learning frameworks with modern high-performance interconnects, protocols (such as InfiniBand, RoCE) and storage architectures over HPC clusters and clouds.

Primary author

Dr Xiaoyi Lu (The Ohio State University)

Presentation Materials

There are no materials yet.