Inside Intel
By Niall McKay in Silicon Valley, Mon Jul 30 00:00:00 GMT 2001

Intel is set to launch an attack on the mobile wireless device market with the release of its new XScale microprocessor expected this fall.


The new product, which the company is marketing as the "Internet on a Chip" will feature its StrongARM processor core, flash memory technology, digital signal processor technology together with a base-band communications chip-set.

To date, the chip giant has been satisfied to supply its StrongARM processor to vendors such as Compaq and HP and leave the communications processor market to the likes of Motorola, Texas Instruments and ST Microelectronics. But now with the arrival of third generation (or 3G) network services, consumers will be able to access multimedia content such as audio and video over wireless networks and the lines between what was traditionally the voice market and the data market have become blurred.

Put this together with declining PC sales and a slowdown in the technology sector and you can see why the chip giant is eagerly eying up new markets. In 2000, the company's sales totaled $33.7 billion, three-quarters of which were generated by PC equipment, but sales dropped 76 percent in the second quarter of this year. Furthermore, Intel has slashed its prices by up to 51 percent for its Pentium III and Pentium 4 processors. Now the company is feeling the pain of the high tech slowdown and has told workers at its Silicon Valley facility to take unpaid vacation during the summer months and has started to announce layoffs at its European facility in Dublin, Ireland.

Conversely, the world wide wireless data devices market is still expected to grow in the next few years. Micrologic Research, a consultancy in Phoenix, Arizona, predicts that 1.6 million third generation handsets will ship this year, mostly in Korea and Japan. Europe is expected to adopt 3G by the end of 2001.

Anticipating change


What does all this mean? Well, the industry is hoping for an explosion in personal digital assistant and personal communicator type devices.

Pent up demand for handheld computers that download audio and video, phones that provide complete Internet access and connected devices of all flavors are the last bastion of hope for the beleaguered technology industry.

"(For Intel) this market could be worth tens of billions of dollars," Ron Smith, senior vice-president and general manager of Intel's wireless communications and computing group, told Reuters in an interview at the launch of the XScale at the Intel developers conference in Amsterdam, in May.

Despite its cutbacks the chip-giant is expected to spend in excess of $1.4 billion (almost a third of its annual budget) on research and development of communications products this year. Put this together with the $5 billion that it has already spent on acquisitions in this area and you get a feeling for where Intel wants to be in the next couple of years.

StrongARM tactics


To achieve this the company has redesigned its StrongARM microprocessor. Intel acquired StrongARM when it bought Digital Equipment Corp's Microprocessor division some years ago, but for those of you wondering why the letters ARM keep cropping up in the mobile processor marketplace (Motorola's Dragonball processor is also based on the ARM as is TI's OMAP) you need only cast your memory back to one of the great losers of the technology industry, the UK's Acorn Computer Group.

Acorn was the creator of the famous BBC Micro. When it came time to replace the BBC Micro its engineers sat down and designed a 32-bit RISC processor. They came up with the chip that was to become the ARM processor. They spun it out into a separate company called Acorn RISC Mechanics.

The Italian firm Olivetti and Apple Computer invested in the company (Apple needed the processor for its Newton personal digital assistant) and called it Advanced RISC Mechanics. ARM has done very well but Acorn, which also came up with the idea for the Network computer, didn't really get the benefit for its good work. Now, there is about as much similarity between the original ARM and the XScale processor as there is between a 386 and a Pentium processor.

With XScale, Intel is trying to do to the mobile wireless market what it did for the PC market, namely, get industry standardized on one platform. Intel. The first step was to design a system on a chip that could provide the new class of personal communications devices with the brawn needed to operate a small computer, connect to the internet and process multimedia data such as digital audio and video.

So it took the StrongARM processor and enhanced it so that it can run at between 50 MHz and 1 GHz depending on the application, added flash memory, integrated the Digital Signal Processing technology it acquired from Analogue Devices. The chip will be sold either with or with out the base-band wireless communications capabilities.

Design strategies


"Intel's big challenge now is to get the power consumption under control," says Jupiter Communications wireless analyst Seamus McAteer. "If it can do that then it has a good shot at gaining significant foothold in the wireless data market."

Roy Want is manager of the embedded systems group at Intel's research facility in Santa Clara, California believes that the company has solved the power consumption problem. The former Xerox PARC computer scientist has been hard at works rebuilding the software module called the 'scheduler,' which doles out the processor time to the applications running on the device.

"We are using Dynamic Voltage Management which allows us to reduce the supply voltage depending on what application we are running," says Want. "For example, if you are running an MP3 Player then the device an tailor its power needs to run that application."

Furthermore, Intel has introduced it Personal Internet Client Architecture (also called PCA).

"PCA is its separates the communications and computing engines on the chip," says Intel spokesperson Mark Miller. "Currently, every time you change an application on a phone it impacts the communications capability so you have to get approval form the various telecommunications governing bodies."

As well as this, low-level communications chip programming is somewhat of a black art so few companies have the expertise or the will to do so.

PCA will, according to Intel, enable hardware and software building blocks to be interchanged. That way, adding more functionality to a device such as the home networking technology Bluetooth will be as relatively cheep and easy.

"PCA will help reduce the manufacturing cycles for the cell phone manufactures such as Nokia from 18 to 6 months," says Miller.

That's certainly good for service providers but it is really good for mobile phone manufactures. Jim Tully, analyst at Gartner Research thinks not. "Do you think that the OEMs (original equipment manufacturers) will be happy to see Intel taking a leading role in the communications business?" he asks. "If you look at what Intel and Microsoft have done to the PC business you will see that they dominate the innovation rather than the OEMs such as Compaq, HP and Dell."

That is not to suggest that Intel and Microsoft are working together to unseat the current handset vendors. They are working separately to unseat everybody. Both companies have the same goal, to separate the applications development from the hardware platform but each is pursuing different strategies.

Competition in the markets


With Intel's PCA applications developers will still write applications that run on the chip. Microsoft is, of course, an operating system vendor and is developing its .Net initiative. With it developers will, according to Microsoft, be able to write applications in its Visual Studio software development tool and deploy them on any (Microsoft) platform.

Both initiatives may be good for the telecommunications operators allowing them to rapidly develop new services and offer them to customers. However, Intel and Microsoft are not the only game in town, indeed far from it. Neither are they necessarily the best placed to take the market.

The Symbian consortium will give Microsoft significant competition in the software space and Intel, late to the party, has to compete against Texas Instruments' competing architecture to PCA - OMAP (Open Multimedia Applications Platform) which is also and effort to create a set of hardware and software building blocks for the next generation of cell phones and hand-held devices. TI's chip also combines a RISC processor core, flash memory and a high-end DSP.

Rumor has it that Tom Engibous, chairman, president and chief executive officer of TI has a pocket calculator sitting on his desk as a reminder of what can happen to giant that becomes complacent in this very competitive market.

Then of course there's Motorola's Dragonball technology, which is also a dominant player in the communications market.

Certainly, Intel has quite a battle on its hands. Handspring, Sony, Nokia and Ericsson all have already signed on to support TI's OMAP architecture, while Intel has signed up Compaq, HP and some say Palm will announce an Intel-based product later this year.

Will these new power blocs be good for the cellular market? You decide.

Niall McKay is a contributing editor for the Red Herring magazine. He can be reached at www.niall.org.