Mastering AI Chip Complexity
Explore challenges and solutions in AI chip development
Download eBook
SNUG Keynote Videos
SNUG Silicon Valley 2023 Keynote
The Standardization Imperative for Chiplets
François Piednöel, Distinguished mSoC Chief Architect & Development North America at Mercedes-Benz Research
Chiplet-based design holds great promise to catalyze substantial advances for designed-in intelligence across any markets and applications. The predictability and economy delivered by this approach as compared to a monolithic SoC are compelling. To deliver on the great promise of this approach requires conquering some important obstacles. Standardized interfaces and lower cost, with more robust assembly and test are just some of those obstacles. In this keynote address, François Piednoël explores the challenges that need to be overcome to make chiplet-based design mainstream. He also discusses the impact this work can have for automotive-based design, with examples. Achievements in this domain promise to save lives on the road.
SNUG Silicon Valley 2019 Keynote
Applications of Unsupervised Learning
Bryan Catanzaro, VP of Applied Deep Learning Research, NVIDIA
In this talk, Bryan will discuss several unsupervised learning applications, including video translation, large-scale language modeling and WaveGlow, a flow-based generative model for the vocoder stage of speech synthesis. Because unsupervised learning allows us to try tackling problems where labels would be prohibitively expensive to create, it opens the scope of problems to which we can apply machine learning.
SNUG Silicon Valley 2017 Keynote
Learning and Multiagent Reasoning for Autonomous Robots
Dr. Peter Stone, Professor of Computer Science at the University of Texas, Austin
For robots to operate robustly in dynamic, uncertain environments, we are still in need of multidisciplinary research advances in many areas such as computer vision, tactile sensing, compliant motion, manipulation, locomotion, high-level decision-making, and many others. This talk will focus on two essential capabilities for robust autonomous intelligent robots, namely online learning from experience, and the ability to interact with other robots and with people. Examples of theoretically grounded research in these areas will be highlighted, as well as concrete applications in domains including robot soccer and autonomous driving.
SNUG Silicon Valley 2016 Keynote
What Else Besides FinFET?
Dr. Chenming Hu, Professor, University of California, Berkeley
The "thin body" concept that FinFET introduced will allow gate length reduction to continue with low leakage current. The new frontiers are cost reduction and voltage / power reduction toward 0.1V. While the solutions remain elusive, progress has been made.