Intel aims to fend off AMD servers

Intel plans to use its biannual Developer Forum in San Francisco this week to get into the nitty gritty of a new microprocessor architecture that the company says will help it close the gap with Advanced Micro Devices, which has been eating into the chip giant’s market share as customers look for high-performing, low power-consuming systems.

That pressure got turned up a notch last week with the news that Google, which runs hundreds of thousands of servers internally, plans to switch from Intel-based systems to AMD-based servers. In a research report issued last week, Morgan Stanley analyst Mark Edelstone said Google has started buying AMD Opteron-based servers for almost all new purchases.

A Google spokeswoman would not confirm the switch, saying only that “Google continually evaluates the price and performance of its computing infrastructure to guide purchasing decisions.”

“We typically do not disclose details about our various supplier relationships,” she says.

Corporate buyers in growing numbers are turning to AMD, which has been able to provide higher performance at a lower power threshold than similar Intel-based systems. Intel executives hope their new generation of microprocessors will help stem that tide.

Not that Intel is in danger of losing its command of the x86 market. But AMD, whose market share has languished in single digits, is fast expanding its customer base. In the fourth quarter of last year, AMD held nearly 13% of the overall x86 market, compared with just 5% in the fourth quarter of 2004, according to Gartner.

“AMD has had a great run with Opteron,” says Charles King, principal analyst with Pund-IT. “I’m not predicting that Xeon will finally surpass it any time soon.”

But King notes that the real proof will come as Intel and AMD roll out their next-generation processors.

The spring Intel Developer Forum will include a number of sessions providing deeper technical insight into where Intel is going with its new processor architecture.

Called the Next Generation Micro-Architecture, the new design is based on the Pentium M processor, a low-power chip for mobile personal computing. The new chip platform, which is a shift from the NetBurst architecture introduced with the Pentium 4 in 2001, will result in smaller, more energy efficient systems, operating at around 80 watts, Intel says. Today’s Intel-based systems operate at as high as 130 watts.

AMD is using this week to launch three faster Opteron processors and discuss plans for its next-generation chips. On Monday, AMD plans to unveil its x85 series dual-core Opteron chips that run at 2.6GHz, compared with today’s 2.4GHz processors.

HP says it plans to refresh its Opteron-based ProLiant systems this month, and customers can expect IBM and Sun to do the same.

AMD also is holding an event in San Francisco this week, inviting the press for briefings on its processor road map.

Intel says systems based on next-generation micro-architecture chips – code-named Woodcrest for servers, Conroe for desktops and Merom for mobile computing – will begin shipping in the second half of this year.

Intel also is expected to provide more detail about Dempsey, its dual-core Xeon processor for two-way systems that is expected to begin shipping in volume in May. In addition, attendees can expect to hear about Montecito, the first dual-core Itanium processor that was to have begun shipping late last year. Montecito is scheduled to instead begin shipping in the second half of this year.

While Dell, HP and IBM are expected to introduce systems with the new Xeon processors, HP remains the only Tier 1 systems vendor to sell Itanium. It will begin shipping Integrity servers based on Montecito midyear, HP CEO Mark Hurd said during a Webcast last week in which he, Intel President and CEO Paul Otellini and Oracle CEO Larry Ellison gathered to sing the praises of Itanium.