Exploiting HPC for Big Data, Analytics and AI
SessionData Analytics
Presenter
Event Type
Exhibitor Forum
TimeTuesday, November 14th3:30pm -
4pm
Location501-502
DescriptionIn this presentation, we will cover the use of HPC
technologies for Big Data problems. We will share the
design philosophy behind the architecture of Cray
analytic systems and discuss results from benchmarks on
three kinds of workloads: graph analytics, matrix
factorization and deep learning training. These results
will demonstrate how the combination of the network
interconnect and application of HPC best practices
enables the ability to process 1000x bigger graph
datasets up to 100x faster than competing tools on
commodity hardware, provides a 2-26x speed-up on matrix
factorization workloads compared to cloud-friendly
Apache Spark and promises over 90% scaling efficiency on
deep learning workloads (i.e. potentially reducing
training times from days to hours). We will end by
presenting success stories of organizations that have
leveraged HPC thinking both in the enterprise and
scientific computing sectors.
Presenter
