800VDC MGX Accelerated Computing Rack and Energy Storage for Improved GPU Power Utilization
Open Compute Project via YouTube
Overview
Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn about advanced power management solutions for large-scale GPU computing infrastructure in this technical presentation by NVIDIA power architects. Explore how synchronous AC input current swings in large synchronized GPU clusters create significant challenges for datacenter infrastructure and discover practical mitigation strategies through energy storage systems. Examine various energy storage approaches including electrolytic capacitors, supercapacitors, and batteries to smooth power consumption patterns and maximize utilization of available datacenter power capacity. Understand GPU-specific energy storage requirements and gain insights into proven techniques for improving rack input current profiles that datacenter operators and hardware designers should implement. Master practical power optimization strategies essential for deploying high-performance computing clusters while maintaining stable electrical infrastructure in modern datacenters.
Syllabus
800Vdc MGX Acclerated Computing Rack & Energy Storage for Improved GPU Power Utilization
Taught by
Open Compute Project