Performance Soars Amid Intel-AMD Dual-Core Duel

Underneath their dueling dual-core processor announcements, demonstrations, and, most recently, actual releases to the market, Intel, which unveiled its Pentium Extreme Edition chips last month, and Advanced Micro Devices, which unleashed its Opteron chips, represent very different approaches to dual-core. Both, however, achieve a significant increase in computing power for their customers, who are getting what is generally described by industry analysts as the biggest computing boost in more than two decades.

For some time, both Intel and AMD have been building dual-core capabilities into their products, and some of these have been announced, revealed, and released within days of each other during the last couple of months. While the two chip makers fought a similar public relations and technical battle over 64-bit computing, with AMD at the fore until recently, the ability to take advantage of dual-core technology from both companies already exists, according to Gartner research vice president Steve Kleynhans.

“I think dual-core is more important than 64-bit out of the chute because the infrastructure [for dual-core] is already in place,” Kleynhans told TechNewsWorld. “In dual-core, everything’s ready to roll — the operating systems, the applications. There are a lot of applications where multi-threating could help. Multi-threading has been added already. Users already have multi-threading platforms, whether they’re aware of it or not. There’s really a usage model in place to make use of the technology for most users immediately.”

One New Technology, Two New Visions

While dual-core technologies from both Intel and AMD amount to the use of an additional processor core to allow for more, multiple-running programs, the two chip players take different routes to achieve this.

With its Pentium Processor Extreme Edition 840 running at 3.2 GHz and its 955X Express chipset, Intel enables multi-core performance with its Hyper-Threading technology that allows each extension core to present itself as two logical processors. The company — which is producing dual-core technology in 15 projects across its line of processors — said its Pentium Extreme Edition 840 could process four software threads at the same time by using resources more efficiently.

For its part, AMD touts that its multi-core processing technology — available in its popular Opteron chips — was built from the ground up to create a non-disruptive, upgradeable platform that can speed time to market and lower design and support costs.

Both companies tout hardware manufacturer support, with Intel lining up Dell, Alienware and Velocity Micro, while AMD’s dual-core customer list includes Hewlett-Packard, IBM and Sun.

Cautious Optimism

Kleynhans said it would appear on the surface that AMD has a dual-core advantage because of the way memory is accessed with its technology. However, the analyst added it is still too soon to tell which platform will provide better in terms of real-world performance.

“In the real world, who knows,” he said. “Either platform is going to provide a significant step up in performance in a wide variety of applications.”

Kleynhans reported most IT customers have expressed “cautious optimism” on the new dual-core chips, which do represent higher price and additional complexity. He added while there is demand for the technology based on a need to run more applications simultaneously — particularly security programs — there is also a limited number of customers who will pay to be first adopters.

“Most are going to wait until dual-core comes for free, which will likely be the case in 2006,” he said.

‘Lego’ Approach

Mercury Research president Dean McCarron told TechNewsWorld the companies’ approaches to dual-core and multi-core computing, while different, both represent a change from the pursuit of performance through increased processor clock speed, or megahertz.

“This is significant, and it’s a pretty fundamental change,” he said. “[Multiple processor cores] is now how we’re going about getting additional performance. It’s one of the biggest changes in the past 20 years.”

McCarron said AMD designed its multi-core technology to scale up for additional performance, describing it as a “Lego approach” to dual-core and multi-core processing.

Intel, on the other hand, has a legacy of putting the multi-core processing capability largely in its front side bus design, which efficiently arbitrates and delegates processing tasks, according to McCarron. He said the two companies had different reasons for their different designs, adding that AMD’s cost per component is higher than Intel’s, which has the manufacturing might to get more advantage from economies of scale.

McCarron added software licensing concerns for dual-core hardware are beginning to vanish as vendors — with the exception of Oracle so far — indicate they will base software license pricing on a per central processing unit (CPU), rather than per core, basis.

A Joust and a Jump

Gartner research vice president Martin Reynolds said the differences between the dual-core approaches from AMD and Intel amount to the difference between the basic Opteron and Pentium 4 core designs.

“AMD has fast buses and direct memory connect,” Reynolds told TechNewsWorld. “Intel has a slower bus, but they make up for it with larger cache on chip.”

Reynolds alluded to both companies’ previous work toward multi-core technology, indicating that their dueling marketing strategies and dual-core rollouts highlight the lack of difficulty in producing the new chips.

“What’s interesting is it feels as though they’ve both pulled the ship dates forward and demo’d within days of each other,” he said. “This reflects it’s relatively easy to do a dual-core design. It’s not like they’re starting over and completely re-engineering.”

Reynolds said a straight performance comparison between the two dual-core designs might give the edge to Intel. However, he added that in the all-important two-way and four-way server categories, AMD’s design scales better to give it an advantage.

“Intel is addressing it, but AMD does have an edge in that regard,” he said.

Neck and Neck

Reynolds said despite some trending up for AMD in the last couple of quarters, the two chip makers “pretty much joust around the same market share,” which he estimated at about 20 percent in units and 10 percent in revenue for AMD, the rest belonging mostly to Intel in both categories.

“It’s amazing how little performance impacts the market,” he said.

The analyst said both companies are including the multi-core functionality throughout their chip lines and users will not have to wait to take advantage of it.

“It will be in their systems relatively quickly,” he said.

Reynolds said the multi-core capabilities would prove most useful in workstation, graphics and video editing, CAD simulation and game development, where dual-core represents one of the biggest performance increases ever.

“Those employees in those areas are highly paid and highly leveraged,” he said. “For those employees, it’s worth the price.”