Reference Guide

4 Dell EMC Ready Solutions for AI Deep Learning with NVIDIA | v1.0
Executive summary
Deep Learning techniques has enabled great success in many fields such as computer vision, natural language
processing (NLP), gaming and autonomous driving by enabling a model to learn from existing data and then to
make corresponding predictions. The success is due to a combination of improved algorithms, access to large
datasets and increased computational power. To be effective at enterprise scale, the computational intensity of
Deep Learning neural network training requires highly powerful and efficient parallel architectures. The choice
and design of the system components, carefully selected and tuned for Deep Learning use-cases, can make
the difference in the business outcomes of applying Deep Learning techniques. In addition to several options
for processors, accelerators and storage technologies, there are multiple Deep Learning software frameworks
and libraries that must be considered. These software components are under active development, updated
frequently and cumbersome to manage. It is complicated to simply build and run Deep Learning applications
successfully, leaving little time for focus on the actual business problem.
To resolve this complexity challenge, Dell EMC has developed an architecture for Deep Learning that provides
a complete, supported solution. This solution includes carefully selected technologies across all aspects of
Deep Learning, processing capabilities, memory, storage and network technologies as well as the software
ecosystem. This document presents the architecture of this Deep Learning solution including details on the
design choice for each component. The performance aspects of this complete solution have also been
characterized and are also described here.
AUDIENCE
This document is intended for organizations interested in accelerating Deep Learning with advanced computing
and data management solutions. Solution architects, system administrators and others interested readers within
those organizations constitute the target audience.