Intel lab looks far beyond next-generation with ‘blue sky’- walkers projects
The head of global chip giant Intel’s Cambridge research hub has lifted the lid on some of the Star Wars-style projects underway at the lab.Lab director, Frank Binns has unveiled some of the work aimed at driving a ‘Brave New World’ in computing, and ultimately, making technological advances so compelling that businesses and ordinary consumers alike cannot bear to be without them.
Binns told Business Weekly: “If we can accelerate peoples’ buying decisions by a year or more, drive a wave of ‘computer refresh,’ then that has a significant effect on sales of Intel-powered products.”
The facility has now achieved critical mass, growing the number of full-time research posts from three when it was set up in March 2003 to around 18. The overall research effort is effectively doubled through internships and post-doctoral posts.
Focusing on networking technology, ‘virtualisation’ research and architecture/systems research, Intel Cambridge operates the same open and collaborative model used by its neighbour and research partner, Microsoft Research. The aim is not so much to generate proprietary, product driven intellectual property, but to advance the state of the art in computing more generally.
Only 20 per cent of its work is ‘product-centric,’ with 60 per cent potentially applicable within a five year window and the other 40 per cent “looking way beyond that.” The lab collaborates extensively with academic and research organisations across Europe and is seeking to reverse the current ratio between medium and long-term work in the coming years.
With such wide technological horizons it is not surprising that much of the work underway at the lab is many steps removed from the mere next-generation.
Among the science fiction-like advances Binns’ operation is working on today are very high-speed, low power computers driven by lasers; ‘many-processor’ computing, where a task is split equally between potentially dozens of processors, with the attendant increases in speed and productivity; and intelligent software that actively hunts out ways to deliver an email for example, if it doesn’t have direct access to the internet.
Last year Intel CEO, Paul Otellini announced the company’s diversification beyond being solely a chip player, reinventing itself as a platform company. Intel is now far more interested in the wider world outside the processor and the development of the “whole solution.”
Working closely with the University of Cambridge, Intel’s lab is aiming to replace the printed circuit board and its copper wires with something altogether higher tech – light.
“At the moment, an Intel processor delivers its ‘buzz’ to the rest of the system through the PCB. In the future, that’s not going to be on copper but optical fibres, with lasers interconnecting the various devices on the motherboard. Not only will this significantly increase performance, but also be much more power efficient – critical for battery-powered devices.
“The high-end enterprise server market is crying out for this kind of technology now and I think it will find its way into the first products in that market in four to five years and filter down from there. The internal workings of a PC will eventually look like something out of Star Wars.”
The current boom in networking will only continue and accelerate as the years unfold, Binns believes, and much of what the Cambridge lab does in the space is aimed at making it more pervasive, less infrastructure-intensive and better managed.
Two major strands of its networking research are aimed at helping wireless networks to peacefully co-exist: “I recently moved into a new apartment complex and when I started up my laptop and tried to connect to my wireless network, I found that there were 36 in total,” Binns said.
“That’s clearly a good thing, but not so good when you think about it in terms of 36 capable networks all competing and degrading each others’ performance. As the number of wireless networks grows it will be vital to introduce management capabilities.”
Another research group is working to push computers, especially mobile devices, to be more inventive in their efforts to deliver information to another remote computer. Effectively injecting some ingenuity in getting your email from point ‘A’ to point ‘B.’
Binns said: “Security is an incredibly important aspect of network management, but it does provide a barrier to effective communication.
“A commercial intranet must be secure, so if I don’t have permission to connect I am blocked from communicating with you.
“But there are other ways of communicating, for example infrared or Bluetooth, but current applications and software don’t cater for it. We are working on protocols and hardware that will enable ‘opportunistic’ networking – so that if you’re driving past Starbucks for example, your laptop will detect the presence of a Wi-Fi hotspot, take the email that’s sitting in your outbox and send it.”
Moore’s Law seems to have hit something of a developmental block in recent times and what appeared at one stage to be an irrepressible advance in processing power has slowed markedly. Intel’s answer to this is to develop systems that use more than one processor.
The first step along this path was taken as long ago as 2000, with Pentium 4’s ‘hyper-threading’ capability, which allowed a single processor to act – and compute – like two. There are now, of course, the Pentium M and Core Duo processors, which harvest the obvious benefits of using two processors; the technology is known as multi-processing.
“We see the move to multi-processing continuing and accelerating within the horizons we are looking at. We loosely describe the work as many-processor computing, meaning more than four processors working in harness, in fact potentially a lot more.
“The difficult part of all of this is writing software capable of taking advantage of 32 processors at once, for example. It is difficult enough working with a single ‘compute thread’ never mind dozens.
“Our work centres around making the software development environment much easier to take advantage of all of these processors.”
Proving that collaborative, open-source research can have very real impact on the wider commercial world is the company’s virtualisation research, a project called Xen, partnered with Microsoft Research and Cambridge University’s computer lab.
In basic terms, the Xen technology allows any operating system to be run on the same computing platform simultaneously. While the benefits of this to the average user are not immediately obvious, Binns says that the benefits to large corporations with mission-critical computing cannot be underestimated.
Xen provides a buffer or soft partitioning between the computing platform and the operating system sitting on top.
It could mean the end of downtime while a network is being maintained and even lessen the amount of damage that a virus can do.
The work has already generated a spin-out company from the University of Cambridge’s computer lab in the form of XenSource, now based in Palo Alto, California.
The company has quietly gone about the business of raising $23m in venture capital from Accel Partners, Kleiner Perkins Caufield and Byers, New Enterprise Associates and Sevin Rosen Funds.