Explore the challenges and solutions of GPU sharing in Kubernetes clusters through this informative conference talk. Learn about the MetaGPU open-source project, which enables dynamic configuration of GPU sharing with zero downtime. Discover how to maximize GPU utilization for AI/ML workloads, overcome limitations in older GPU generations, and implement fractional GPU allocations. Gain insights into production-grade fractional GPU operations on Kubernetes clusters, including tips, challenges, and lessons learned from developing and implementing this technology.
Fractional GPU Allocations with MetaGPU Device Plugin for Kubernetes
CNCF [Cloud Native Computing Foundation] via YouTube
Overview
Syllabus
Sharing is Caring – Fractional GPU Allocations With MetaGPU Device Plugin - Dmitry Kartsev, cnvrg.io
Taught by
CNCF [Cloud Native Computing Foundation]
Related Courses
-
GPU Sharing and Container Device Interface in Kubernetes Device Plugins
-
Increasing GPU Utilization on Kubernetes Clusters for AI/ML Workloads
-
Scaling AI Workloads with Kubernetes - Sharing GPU Resources Across Multiple Containers
-
Are You Really Out of GPUs? How to Better Understand Your GPU Utilization
-
Precision Matters: Scheduling GPU Workloads on Kubernetes
-
Improving GPU Utilization and Accelerating Model Training with Kubernetes Scheduling Framework and NRI
