Supercharging OpenStack clouds for Deep Learning using NVIDIA GPUs with NVLink

1,097 views · Published 8 November 2017 · 46:09 · Indexed 12 November 2026

Channel: OpenInfra Foundation · 2017 · Science & Technology

Watch on YouTube

Increasingly, data scientists are demanding cloud-based resources which contain one or more NVIDIA GPUs in order to accelerate training and inference of their Deep Learning models. OpenStack in combination with qemu/KVM provide the perfect toolset to providing these scientists with high-performance VMs while maintaining the benefits of a fully-virtualised, multi-tenant host. In this talk, you will learn how to pass through one and multiple NVIDIA GPUs (with NVLink) to a Linux VM, and some of the

More from this channel