Intel is aiming the processors at the high-end market and has announced 20 stability improvements to make sure the chips are trusted to run mission-critical systems such as transaction processing and virtualisation.

“We are talking about bringing mission-critical to the masses and mission-critical to the market,” said Kirk Skaugen, vice president of Intel’s architecture group and general manager of its datacentre group.

The new chips use Intel’s Machine Check Architecture currently used in the Itanium and RISC chips to allow virtualised environments to deal with errors without shutting down the entire system.

The Xeons are not designed to compete with Itanium chips, Skaugen said in a question and answer session, but will offer high-performance computing at the value end of the market.

Skaugen also stressed that these features will be supported by software vendors, including Microsoft, Red Hat, Novell, SAP and SUSE, as well as by 12 original equipment manufacturers.

“We have talked to the operating system firms,” he said. “We learned our lesson of waiting for software with the move to 64-bit, which you may recall.”

Skaugen added that the new chips are ideal for companies looking to replace outdated hardware, claiming that IT managers could shut down 20 single-core four-chip servers and replace them with one Xeon 7500 server without affecting performance.

The chips are around three times as fast as Intel’s Xeon 7400 series, according to Skaugen, a claim which was backed up by IBM.

“On systems like virtual machine benchmarks we have seen x3.8 performance increases,” Alex Yost, vice president of IBM’s System X and BladeCenter division, told V3.co.uk. “On SAP the boost is about x3 and on TPC x2.8. ”

Yost explained that the new chips could allow companies to pack more computing power into their systems, backed up by large amounts of memory.

Using IBM’s xFlash technology, for example, a company could replace 100 racks with a single high-end system, something that would prove a compelling purchase for companies running out of datacentre space.