Car Compute


The processing power in cars is moving even faster than Moore's law. That's because, in addition to gains from faster processors, there is also additional need and thus increased per-car spend on computing power. Current high-end cars carry an arsenal of diverse sensors, display screens, human interfaces, ADAS features, Internet, and Infotainment apps, as well as the usual battery of car management functions. Something needs to crunch all that data to make sense of it. Our meeting will look at the needs and uses of car compute power. But car compute is not the whole story. Much like Google Voice recognition or Siri, the mobile unit is better when it leverages the power of the cloud. In turn, the data gathered from the car can be aggregated and uploaded in order to augment the knowledge stored in the cloud. Thus, the optimal structure is a partnership between car compute and cloud compute – however, the car must be able to function independently when connectivity is (inevitably) disrupted. So, this duality offers us ample room for discussion and debate for our meeting: · How much compute is needed, and for what uses? · What is the impact of autonomy on the processor requirements? · Do cars need an AI level of capability? · What about redundancy, like aviation? · Where is that compute power best located in the car, and outside? · What are the privacy implications of this dual brain? What about the security risks?


  • Date: 9/11/2015 08:30 AM
  • Location: Nvidia, Santa Clara (Map)

Sorry, registration has ended.