Power struggle in the processors market

Intel has a raft of new processors on the way, but there are serious challenges to the company’s market share on all fronts

Intel has a raft of new processors on the way, but there are serious challenges to the company’s market share on all fronts

Not so long ago, choosing what type of processor you wanted for your new system or upgrade was easy. In fact, the choices were fairly limited and most discerning shoppers opted for an Intel chip. Through its technological advantage, Intel created a dominant position in the market with its Pentium processors and had most of the major PC manufacturers tied into exclusive contracts. But Intel's crown is starting to slip, and the company could soon be faced with serious competition in all sectors of the PC marketplace. But the question remains: will buyers want to risk moving away from Intel for the chance of improved performance at a lower price?

With the emergence of the value PC market, Intel seemed to take its eye off the ball, letting competitors gain a significant foothold. AMD emerged with its K-series as a good value alternative to Intel's more expensive Pentium processors. Intel tried to counter with its low-end Celeron range of chips, but serious problems with the first generation of these processors allowed AMD to grab a significant chunk of the consumer PC market. The problems with the Celeron have since been sorted out but the damage was already done, and in the US AMD accounted for more high street sales of PCs than those with Intel inside. AMD's introduction of the higher performance Athlon chip last year further boosted its consumer presence, but so far it has had great difficulty breaking Intel's stranglehold in the business desktop and server marketplace. This could soon change as AMD produces a product roadmap to challenge Intel's core business. But it is not only AMD that Intel has to worry about. Other challengers, such as Transmeta, are entering the arena, a situation that could threaten Intel's mobile processor market share.

Of course, Intel is not resting on its laurels, and most of the developments that have been reported by Intel's competitors are in response to work already underway by the chip behemoth. Plans are afoot to regain significant ground in the value PC market segment. Code-named Timna, Intel's forthcoming low-end chip is designed as a natural replacement for the Celeron and will integrate the CPU, memory and graphics controller onto one chip. The integration of this so-called "system on a chip" should mean much lower manufacturing costs and therefore a lower price to the consumer. Reports have suggested that systems using the Timna chip could cost as little as $600, with the chip itself coming in at only $69.

Original plans were to release Timna later this year, but technical problems have held this back. A flaw causing system-freeze in the Memory Translator Hub, a component of Timna, is believed to be at the root of the delay. A similar problem caused the recall of one million Pentium III motherboards in May. The release of Timna is now expected in the first quarter of 2001. A successor to Timna, called the Timna+, is also on the product roadmap, but details of this are still sketchy.

The second half of this year will see the introduction of a new chip however. Currently named Willamette, the chip is an update to Intel's current 32-bit offering. The chips will be introduced at speeds of well over 1GHz and will feature a number of performance-enhancing features. Firstly, a new hyper-pipelined design will enable instructions inside the chip to be queued and executed at a much faster rate, thus increasing the overall clock speed. A set of 144 new instructions will also be introduced which enhances performance to accelerate video and encryption, and support forthcoming Internet applications. Willamette will also use a 400MHz system bus, which, according to Intel, is the first in the industry. The bus will transfer information from the processor to the rest of the system at speeds three times faster than current PIII processors.

Intel, however, may have trouble convincing motherboard and PC manufacturers to adopt the initial Willamette chip as it appears to have a very limited lifespan. From release in the fourth quarter of 2000, it is expected that a new motherboard chipset, named Tulloch, will be introduced in the second quarter of the following year, initially for Willamette, but it will pave the way for the Northwood processor in the second half of 2001. Northwood is expected to be a 0.13 micron processor, and Willamette is believed to be just a stepping stone on the way to the Northwood release.

More significant is the forthcoming introduction of Intel's 64-bit processing architecture in the form of Itanium, formerly code-named Merced. This practically creates a new high-end category of its own and is aimed at e-business. Obviously, speed will be improved due to its ability to execute more instructions simultaneously and the much larger memory gives greater room to store, deliver and mine data. Availability has also been a key issue with Itanium - especially due to the constant nature of the Internet - and Intel expects the uptime of its 64-bit chip to be exceptional. Expect to see Itanium chips appearing towards the end of this year.

One downside to Itanium is that due to the change of architecture, it will be less effective in running current 32-bit software applications. In fact, the processor is believed to run slower than current Pentium III chips when executing 32-bit software instructions. Intel seems unduly concerned by this, claiming that most 32-bit software can be recompiled as a 64-bit program. Also, Intel believes most customers who adopt Itanium will be looking to 64-bit software rather than porting old applications onto a new system.

AMD has taken this as an opportunity to enter the 64-bit market. Not best known for its development at the high-end, AMD believe that by offering a 64-bit computing solution that is truly backwards compatible with 32-bit software, it will offer an easier migration path to users wishing to install the new chips. Sledgehammer, AMD's alternative to Itanium, has been designed to do exactly that. It will incorporate current x86 32-bit computing at similar speeds to its 64-bit operation.

Nathan Brookwood, principal analyst at Insight 64, said: "Time and again, processor architects have looked at the inelegant x86 architecture and declared it cannot be stretched to accommodate the latest innovations. Time and again, a clever designer has proven them wrong. It seems truly ironic that Intel, the inventor of x86 architecture, has bet its 64-bit future on a radical new design, while the upstart AMD proposes a more conservative and computable path. If AMD's approach pans out, customers will benefit from a wider array of choices for their high-end systems. Perhaps when architects in 2025 begin to debate the move to 128-bit computing, they won't be so quick to reject extensions to the x86-64 features AMD has laid out."

AMD's main problem with Sledgehammer is that it isn't due for release until the end of 2001. By that time Intel will have moved onto its second generation of 64-bit processors, code-named McKinley.

But Sledgehammer isn't the only plan that AMD have in store for the next year or so. The success of the Athlon, partly down to AMD's ability to get the chips out while Intel has struggled to ship its PIII Coppermine processors out of the factory fast enough, has convinced AMD that it can go after Intel's core market as well as the value PC market.

Athlon recently beat the Pentium III to reach the landmark 1GHz processing speed and AMD now wants to utilise the Athlon to break into the mainstream market in a major way. Recently, it announced a new version of the Athlon chip, codenamed Thunderbird. The new chip, available initially in six speeds from 750MHz to 1GHz, features a much larger on-chip L2 cache of 256KB, increasing performance by holding large amounts of relevant data closer to the processor. These chips will also be the first to use 0.18 micron copper technology as opposed to aluminium. With these and several other performance improvements, AMD hopes it can raise a serious challenge to Intel in the mainstream market, and has already got several manufacturers to include Thunderbird in their systems, including Compaq, Fujitsu-Siemens, Gateway, HP and IBM.

Later on in the year we should see further developments to the Athlon processor. Scheduled for release during the second half of this year, Mustang will have a much reduced core size, requiring less power and around 1Mb of L2 cache. Alongside this will be Corvette, which is to be developed specifically for the mobile market.

Of course, AMD is not ignoring its stronghold of the value PC market and has just started shipping its Duron processor, a scaled down version of the Athlon, to compete with Intel's Celeron. This will initially be available in speeds of 700, 650 and 600MHz.

Intel not only faces a challenge from AMD, but new company Transmeta looks set to steal a significant proportion of business in the mobile sector. Transmeta's main selling point is that the processor needs much less power than currently available processors, meaning extended battery life for mobile devices. While the company is still in its early stages, it is expected to make some major announcements with computer manufacturers soon.

While Intel still dominates a large part of the processor market, it looks set to face some major challenges in the near future, and while this may not be good for Intel in the short term, it should be good for the market as a whole. At least, it should certainly benefit the consumer, with prices continuing to drop while performance and choice increase.

Paul Grant

Read more on PC hardware