Virtual worlds, 80 cores, and 20,000 golden pigs

Why show 20,000 golden pigs to a select group of 85 press and analysts? Because it was a cool way to show both a future application capability (massive collision detection) and a new parallel programming environment called Ct, i.e. C with throughput extensions.

Here is some footage from PodTech highlighting three of the demos we had at the event. You’ll see the 80-core chip first, which is an example of architecture research to scale to large numbers of cores on future chips. Then, you’ll see Ct, which is an example of what we’re doing to make chips like this programmable. Finally, an application that you might want to have coded for such a tera-scale chip — and example of ongoing work in virtual worlds.

These were just a few of the projects we showed from our Tera-scale effort, which in turn is only a fraction of the total effort (more than 100 projects). But, if you still haven’t had enough, here’s a much less produced (and only somewhat audible) clip of yours truly attempting to speak to nearly all the demos in one take.

Lastly, I wanted to thank everyone who commented on my previous blog about what to do with so many cores. There were some interesting ideas ranging from fractal compression to protein modeling. I forwarded the whole set of ideas to the director of our Application Research Lab and some of his principal engineers. Suggestions like these are very welcome – thanks.