Intel's new dual-core Xeon targets four-chip servers

Intel  has now brought dual-core technology into all of its Xeon processors for low-end servers with the launch of four new Xeon 7000 series chips, the company announced Monday.

Intel was joined by perennial partners Dell and Microsoft on a conference call Tuesday detailing the results of Intel's plans to accelerate the introduction of its first dual-core Xeon processors. The chip company launched its first dual-core Xeon processor for two-chip servers three weeks ago, and now offers dual-core processors across all of its x86 server products.

Dual-core processors are the chip industry's answer to the problems caused by excessive heat given off by fast single-core processors built with today's generation of leaky transistors. A dual-core chip contains two separate CPUs on a single piece of silicon, which allows chip designers to improve performance despite lowering the clock speed of those chips.

To continue reading, register here and become an Insider. You'll get free access to premium content from CIO, Computerworld, CSO, InfoWorld, and Network World. See more Insider content or sign in.

Intel  has now brought dual-core technology into all of its Xeon processors for low-end servers with the launch of four new Xeon 7000 series chips, the company announced Monday.

Intel was joined by perennial partners Dell and Microsoft on a conference call Tuesday detailing the results of Intel's plans to accelerate the introduction of its first dual-core Xeon processors. The chip company launched its first dual-core Xeon processor for two-chip servers three weeks ago, and now offers dual-core processors across all of its x86 server products.

Dual-core processors are the chip industry's answer to the problems caused by excessive heat given off by fast single-core processors built with today's generation of leaky transistors. A dual-core chip contains two separate CPUs on a single piece of silicon, which allows chip designers to improve performance despite lowering the clock speed of those chips.

Intel's dual-core Xeon chips were originally expected to launch in the first quarter of next year. However, Intel moved up the launch date partly because the chips were ready earlier than expected after the company rapidly developed the dual-core desktop processor that serves as a blueprint for the Xeon 7000 chips, and partly in response to competitive pressures from AMD's dual-core Opteron processors, which have been available since April.

Intel has included hardware support for virtualization technology into the Xeon 7000 processors, but that capability won't be available until next year as part of a BIOS upgrade, said Kirk Skaugen general manager of Intel's server platforms groups. The company has been actively promoting its VT technology, which will improve the performance of virtualization software on these servers.

Dell's new PowerEdge 6800 and 6850 servers will feature the new chips, said Neil Hand, vice president of worldwide enterprise systems marketing at Dell. Those systems can be ordered today, and will start shipping in two weeks. Prices start at around $6,400, and additional details are available at Dell's Web site.

IBM's new xSeries 460 servers will also use the Xeon 7000 series processors. IBM's servers use a unique chipset developed by the company called the X3 chipset, while Dell's use Intel's chipset technology. The X3 chipset allows IBM to reduce the latency in moving data from the processor to memory, a drawback of Intel's current processor and chipset designs.

Intel is still reliant on a front-side bus and external memory controller to coordinate the movement of data from the processor to memory, a key link in system performance. In this setup, two or more processors must share a single connection to the rest of the chipset, and the memory controller sits outside the processor.

AMD, on the other hand, has integrated the memory controller directly onto its dual-core Opteron processors and allows each processor to connect directly to the chipset. This means the memory controller can run at the same speed as the processor, and each processor does not have to compete for bandwidth over a shared connection.

The IDG News Service is a Network World affiliate.