"We believe that to really get the most use out of these many-core systems, there's going to be quite a significant modification to the way people program today," Bautista says. The cores can handle many different instructions at once, he says, and software engineers will have to learn new programming techniques to take full advantage of the added computational capacity. Tilera's Agarwal says that his company has addressed that concern by providing a software environment that helps customers gradually upgrade, debug, and optimize their applications to work on the 64-core system--even applications designed to run on a single core. The company's technology is being presented this week at the Hot Chips symposium at Stanford, in Palo Alto. Nathan Brookwood, founder of Insight64, an analysis firm, says that many people at the conference are excited about Tilera's work, mainly because it could have immediate applications, such as expanding the capacity of videoconferencing systems and analyzing network traffic in routers. "I think they have a potential winner here," says Brookwood. Intel's Bautista says the marketplace may be ready for a chip with more computing power, but it would need to be low power and easily programmed. He says that Intel will keep an eye on Tilera, as it does on many startups that are first to market with new technologies, to see how customers respond and which aspects of the technology could be improved. "We use companies like this to help us test the waters," he says. |
Bringing Light to Computers
03/26/2008


Comments
eightwings on 08/27/2007 at 10:43 PM
2
Multicore CPUs are not easily programmed and that's the problem that neither Intel nor Tilera has adequately addressed. And it's a problem that must be solved before the industry can see a massive switch to chips like Tilera's.
In my opinion, the reason that current multicore processors are hard to program is that they were created for the wrong software model. The algorithmic model has been with us since the days of Charles Babbage and Lady Ada. It is not well suited for parallel processing because it is inherently sequential. It is time that we switch to a non-algorithmic model, one which is implicitly parallel. Sequential ordering should be explicit, not implicit as it is now. Project COSA is a step in the right direction.
http://www.rebelscience.org/Cosas/COSA.htm
The multicore CPU manufacturers have only themselves to blame. You don't design a new CPU architecture and expect the software model to change to accomodate it. It should be the other way around. The software model must come first.