Intel’s Nehalem EX to gain error correction technology

Intel's eight-core Nehalem EX server processor will include a technology derived from its high-end Itanium chips that helps to reduce data corruption and ensure reliable server performance, the company said.

The processor will include an error correction feature called MCA Recovery, which will detect and fix errors that could otherwise cause systems to crash, said Boyd Davis, general manager of Intel's server platforms marketing group, at a press conference.

The chip will be able to detect system errors originating in the CPU or system memory and work with the operating system to correct them. That could help make systems more fault tolerant and provide greater uptime, he said.

Error correction is a feature Intel includes in its high-end Itanium chips, designed for servers that require very high levels of availability. Itanium uses a different instruction set to x86 server chips and competes more with RISC-based processors like Sun's Sparc and IBM's Power.

Though Intel is trying to push the Nehalem EX chips into higher-end environments, Davis brushed off concerns that the processors might cannibalize sales of its more expensive Itanium chips.

Itanium sales are driven largely by software and operating system choices, he said. For example, customers running Unix-based systems would be more likely to adopt Itanium over Xeon. In addition, the error correction in Itanium is more advanced than that in the Nehalem EX and can resolve a larger number of errors, Davis said.

“Our expectation is Itanium will be healthy for us based on software” and services, Davis said.

Still, Itanium has been less successful than Intel originally hoped, with only a few vendors such as Hewlett-Packard selling Itanium-based servers.

Last week Intel pushed back the release of the next version of Itanium, code-named Tukwila, to early next year. It said it wanted to speed the performance of highly threaded workloads.

Including the error correction feature in Nehalem EX marks a shift in Intel's strategy, said Nathan Brookwood, principal analyst at Insight 64. Intel tried to keep Xeon confined to the lower end of the market and wanted customers to buy Itanium for mainframe-type applications. By equipping Xeon chips with higher-end features, Intel is pushing Xeon up into that market, Brookwood said.

However, customers running legacy applications or who need very high levels of availability, such as stock markets and banks, may continue to buy Itanium systems, he said.

Adding error correction to the Xeon line will be useful as workloads get more complex and use more memory, Davis said. It may also be helpful as workloads are spread across virtual machines in data centers. Data corruption in one virtual machine can spread to other VMs and cause a server to crash, Davis said. Nehalem EX will be able to isolate an error and restart individual virtual machines without crashing an entire system, he said.

The Nehalem EX chips will be the first x86 server chips to include such error correction features, Davis said. Intel is sampling the chips to server makers, and systems powered by them will begin shipping early next year, Davis said. Intel won't say yet what clockspeeds the chips will run at.

IBM displayed an eight-socket server at Tuesday's press event with 64 cores running 128 software threads simultaneously.

Intel is targeting the EX processors at high-end systems running data-intensive applications such as databases. The chips are built on the Nehalem microarchitecture, which improves system speed by cutting data bottlenecks that plagued Intel's earlier chips. Intel's existing Xeon 5500 quad-core chips, which launched in March, are also based on Nehalem.

Along with error correction, servers based on Nehalem EX will include separate buffered memory chips that can temporarily store data alongside the main memory for faster execution.