Intel's Peer-to-Peer Strategy
One of the keynote presenters at IDF was Louis Burns, VP, GM, Desktop Platforms Group, Intel. Judging by the title, a very important guy at the company. Great presenter. Pretty much gave the message out loud and clear, "Bye, bye Pentium III, except on mobiles. Hello Pentium 4. That's 2001 done."
Obviously, he was also the guy to get the most questions about AMD, and why would companies buy Pentium 4 when it, obviously, didn't do much for things like Word. His answer was something to do with the emergence of P2P in the enterprise. This struck us as an interesting tack to take for a company that wants high-performance chips to be gobble up in volumes when everyone starts to make their year-end purchases. So, we took a look at some of the sessions on P2P.
One example of a P2P application is available for all to try at www.entropia.com . Sign up and have your PC's idle time used to solve the world's problems. Distributed computing, or something like that. Well, we are not giving up any part of our CPU cycles to anyone, and we are certainly not going to be buying more Intel processors because, it might help solve world hunger. It won't.
Intel did try and push multimedia P2P application development. Basically, these apps would have us streaming video to each other, and sharing localized images, and rich media. We'd videoconference. We'd hold virtual reality conferences. We'd be training on archived digital media, audio and video. We'd be collaborating and interacting with documents, audio and video.
One way that we could do this is to use our trusty CPU power to encode MPEG-2 to MPEG-4 for better transport. Hence, there was a proliferation of MPEG-4 video running on all kinds of devices at IDF. Encoding uses CPU cycles. MPEG-4 is supposed to be device agnostic so, in the connected world you'd want to use MPEG-4. Right?
There's the always reliable networked games. Yet, none of this added up to demand for performance. We found it hard to understand how P2P, other than the fact that it is a marketing buzzword, could be an effective driver of high-performance PCs.
Intel makes the argument that companies are building P2P infrastructures, and they should be planning ahead by buying a high performance PC now so that it can meet the demands of future applications. Of course, this is not too different to what Intel has been saying in the past. In the past, it was always frustrating to think that what you bought this week was out of date next week. The trouble recently has been that Intel's competitors have done a better job of selling a safe performance bet than Intel has.
