Event Details

Nowadays, GPUs have become much more powerful while being energy efficient. With the rise in popularity of 3D applications, growth in Big Data analytics, High Performance Computing, and streaming video, GPUs are being pressed into service to gain performance. Traditionally in virtualization usages,  a physical GPU is allocated to a VM using PCI device pass-through, an approach that limits resource sharing. However, graphics virtualization support today enables GPU sharing on par with CPU sharing with tenant isolation, enabling cloud workloads to benefit from GPU processing while allowing cloud providers better returns on their infrastructure investments and thus reducing total cost of ownership.

This presentation will introduce how to accelerate GPU intensive workloads in OpenStack with Intel GPU virtualization technology. We shall demo its use in OpenStack. Come learn how to make your cloud richer and even more performant to meet the demands of the workloads of today and tomorrow.