Computational Storage Solves Big Problems with Big Data

Author(s)/Presenter(s):
Library Content Type:
Publish Date: 
Thursday, August 8, 2019
Associated Event Name: 
Focus Areas:
Abstract: 

Compute requirements are skyrocketing with ever-increasing amounts of data, huge data-centric applications such as artificial intelligence, high-performance computing, and virtual reality, and competitive needs to perform deeper analysis, produce results faster, and achieve higher levels of both coverage and accuracy. Computational storage, bringing compute to the data, is a powerful solution that eliminates roadblocks such as large data moves and limited network and processor capacity. It also provides a readily achievable growth path to handling tomorrow’s even larger requirements. Initial test runs show an over 40% increase in performance, power savings, and net TCO for typical AI/ML applications and a huge improvement for a widely used genomic sequencing tool that is a key to creating treatments for both cancer and genetic problems. Compute power to the edges is a new architecture that also offers tremendous potential for web-based services, cloud computing, deep pattern inspection, content delivery, and scientific computation

Keywords: