GPU Configuration on the Fly Using Dynamic Resource Allocation - A Tale of Two Drivers

Learn how NVIDIA's GeForceNow cloud gaming service leverages Kubernetes and Dynamic Resource Allocation (DRA) in this technical conference talk. Discover the intricate details of dynamically switching GPU drivers to enable both full GPU passthrough and GPU virtualization while maintaining optimal datacenter utilization. Explore the migration journey from traditional Kubernetes device plugin API to DRA, understanding the challenges faced and solutions implemented. Gain practical insights into optimizing GPU-accelerated workloads in cloud environments, with specific focus on managing Kubevirt VMs and GPU configurations. Master the best practices for implementing similar migrations in your own infrastructure while maintaining high performance and seamless user experience.