Berkeley Lab's Breakthroughs in Exascale Supercomputing and AI Energy Efficiency
SAIConference via YouTube
Overview
Explore the evolution of supercomputing and energy-efficient AI in this 27-minute conference talk by John Shalf, Department Head for Computer Science at Lawrence Berkeley National Laboratory. Delve into Berkeley Lab's rich history, from its founding as the first Department of Energy National Laboratory to its current $2 billion annual budget and 15 associated Nobel prizes. Discover how supercomputers process astronomical data, advance predictive biology, and analyze particle accelerator information. Learn about the transition from petascale to exascale computing, achieving 10^18 operations per second, and the challenges of energy efficiency in high-performance computing. Examine market dynamics driving specialized systems, the importance of chiplets in lowering costs and improving performance, and the potential of an open chiplet marketplace. Gain insights into future directions for supercomputing, including the need for more powerful systems to tackle climate modeling and other scientific imperatives. Understand the shift from general-purpose processors to specialized hardware and the critical role of targeted specializations in advancing scientific progress.
Syllabus
Berkeley Lab's Breakthroughs in Exascale Supercomputing and AI Energy Efficiency | John Shalf
Taught by
SAIConference