Intel: Death to all wires

Intel chief technology officer Pat Gelsinger says every computer-related wire, except the power cord, should be done away with

With the race for ever more powerful processors less of an issue these days, Intel is searching for other areas in which to differentiate itself and build a niche, such as wireless broadband, security, and mobile platforms.

ZDNet UK caught up with Intel CTO Pat Gelsinger at the company's travelling Developer Forum in Barcelona, Spain, to talk about the great hope for wireless broadband, WiMax, and his claims that the industry is suffering from "irrational exuberance".

Who will actually deliver WiMax technology to users?
Intel won't do this directly. We will build silicon, which we expect to go into access points and base stations, and eventually PCs. Companies like Samsung, Siemens, Alcatel will do base stations, and others like Cisco will do access points.

When can we expect to see WiMax products on the market?
This year people will start doing wireless broadband deployments that are "pre-WiMax", that is, they'll follow the draft standard, with 2005 being when real deployments will occur.

Will WiMax be deployed by organisations for their own wireless networks?
It depends on the situation. WiMax is good for campus environments, where everything is quite spread out, but not high-density buildings. Wi-Fi is better for that.

Is WiMax a replacement for GPRS or 3G for mobile users? Will it be able to cope with people on trains, for instance?
There are improvements to protocol coming. 802.16e is under development for mobile users. We expect that WiMax will have the benefits of higher bandwidth [than GPRS/3G] and it's designed for a data-centric environment. WiMax is easier to deploy as well. If I'm a service provider, I may not have a 3G licence, so I'll provide a WiMax service.

How has wireless technology affected the market share between desktop and mobile PCs?
Previously the shift in market share from desktop to mobile has been around 1 percent. The Centrino experience shifted that growth rate from desktop to mobile. Last year, we saw a shift of 7 points for the first time. Now that's really brought excitement to the mobile market. Intel likes the shift to mobility, but we're going to continue to innovate in the desktop.


You've mentioned lots of innovations that are coming to Intel's processors, such at 64-bit extensions to Xeon, but haven't spoken about NX, [No eXecute, a memory protection technology] which seems to be far more important. Why is this?
There's an optimistic panacea view of NX. The view that this patches all worms and holes is somewhat optimistic. NX will solve a certain class of attack. There have been lots of attacks which this wouldn't affect. Part of our hesitation to deploy this technology more widely is that it does create problems. People do make use of this behaviour [modifying executable code at run time] in real programs. There's a whole validation programme to go through, plus we're waiting on Microsoft to deliver their parts of the toolkit. If major enterprise applications are broken that's a problem.

Could you do something similar to how you deployed HT technology, with the capability present but not enabled?
It may be present and not enabled today. We built the basic capabilities into the architecture a long time ago.

Intel is pushing ultrawideband (UWB) wireless as a technology, but there's been a split in the industry over standards. Can you tell us why this has happened?
It's been an ugly political battle at the IEEE. The good thing about the IEEE is everyone gets a vote. The bad thing is that everyone gets a vote. If we can't go through the IEEE directly, we'll develop a standard independently and then go back to them. We're part of Multiband ODFM Alliance (MBOA), which is committed to standardisation through the IEEE.

Is UWB just another wireless networking technology, or will it be more than that?
Every in-room wire, except the power cord, should go away. Wi-Fi will be very stressed by an uncompressed high-definition video stream. I'd take down all of my Wi-Fi by doing that. UWB can handle that easily. You'd have to get very creative to exhaust 500Mbps within a room.

Are you saying that UWB can replace all cables?
Ultimately that will be our goal. Get rid of the wires. The only exception will be the connection to the display panel itself, which needs around 1Gbps for an uncompressed signal.

In your keynote speech, you talked of the industry having an "irrational exuberance". Have we got past that stage, and learned the lessons of it?
Is nanotech today in a period of irrational exuberance? Is biotech in a period of irrational exuberance? Is UWB? It's a very natural tendency for us to get exuberant about things. Very often it's the only way we can get the press to write about it. Until you've fired someone, you can't make good decisions about hiring someone. Bad times make you mature. People are going to be very thoughtful about their investment in development. We're not going to stop taking risks, but we're going to be deeper in thought about our risks.