RLX rolls out new Transmeta-based blade

RLX Technologies Inc. has released a new model of its Transmeta Corp.-based blade server that runs on one of the fastest Crusoe processors to date, the company announced Monday.

RLX has been a rare animal in the server world, using Crusoe processors typically found in laptops to power large racks of blade servers. The ServerBlade 1000t from RLX will use 1GHz Crusoe chips and replace existing systems based on a 667MHz version of the processor, said Kim Elsey, product manager at RLX.

RLX also sells blade servers based on Intel Corp.'s Pentium III processor. It positions the Crusoe systems as an answer for customers concerned about the power consumption and cooling costs of their server racks. The Crusoe processors consume relatively little power when running at full speed and almost no power when idle, which can benefit users who stack hundreds of thin blade servers in a rack, Elsey said.

RLX has priced the 1000t server at US$1,329 with one 1GHz TM5800 Crusoe, 640M bytes of memory and no hard drive. The servers run Microsoft Corp.'s Windows Server and Advanced Server and Red Hat Inc.'s Linux OS.

Wu-chun Feng, team leader of the Research And Development In Advanced Network Technology (RADIANT) group at Los Alamos National Laboratory, New Mexico, has used the RLX systems to create a large cluster of servers running the Linux operating system. He built a 240-processor cluster called Green Destiny earlier this year, and is now using the faster 1000t systems in a 480-processor cluster called The Green Machine.

Feng found that using the slower but low-power Transmeta chips can cut down on the cooling and administrative costs associated with running a large cluster, he said. The processors generate less heat than most other server processors, he said, which helps lessen the amount of failures of other components in a server. In addition, The Green Machine can run in a hot warehouse with no special cooling equipment, while other high performance computers tend to require cooling systems.

"When you buy a vehicle, the decision metric that you use is not necessarily horsepower or top speed," Feng wrote in an e-mail response to questions. "The metric (or set of metrics) that you use could be fuel efficiency, reliability, safety, cost, or off-road worthiness. Similarly, when you buy a compute cluster, you should choose a cluster based on your intended application for the cluster. "

Join the newsletter!

Error: Please check your email address.

More about IntelMicrosoftRed HatRLXRLX TechnologiesTransmeta

Show Comments

Market Place