GPU Computing: From Computer Games to Deep Learning and Self-Driving Cars

This talk will discuss the evolution of GPU computing over the last ten years from computer gaming to high performance and scientific computing to most recently self-driving cars and speech recognition in the data center.  Along the way, there have been many challenges and innovations in hardware architecture, memory technology, power efficiency, resiliency and programming models.  Most recently, deep learning have emerged as a dominant workload that is driving many of the new applications.  This new workload shares many aspects of prior applications while at the same time introducing new and unique computational demands that are driving next-generation design.