X

Linux Networx adopts new cluster design

The company unveils two new supercomputer systems made from clusters of Linux machines. Supercomputers at SC05

Stephen Shankland Former Principal Writer
Stephen Shankland worked at CNET from 1998 to 2024 and wrote about processors, digital photography, AI, quantum computing, computer science, materials science, supercomputers, drones, browsers, 3D printing, USB, and new computing technology in general. He has a soft spot in his heart for standards groups and I/O interfaces. His first big scoop was about radioactive cat poop.
Expertise Processors, semiconductors, web browsers, quantum computing, supercomputers, AI, 3D printing, drones, computer science, physics, programming, materials science, USB, UWB, Android, digital photography, science. Credentials
  • Shankland covered the tech industry for more than 25 years and was a science writer for five years before that. He has deep expertise in microprocessors, digital photography, computer hardware and software, internet standards, web technology, and more.
Stephen Shankland
2 min read
Linux Networx, a company that specializes in supercomputers made from clusters of Linux machines, plans on Monday to unveil two new systems: the LS-1 and the LS-X.

The LS-1, to be announced at the SC05 supercomputing show in Seattle, is geared for less sophisticated users, said Ben Passarelli, vice president of product marketing. That means the company's market could expand from labs and high-end users to more ordinary commercial companies.

"The LS-1 is positioned at those who have never owned a cluster before and who were a little timid because they heard the stories that you had to attach a sysadmin to every rack," Passarelli said. "That's not good in commercial areas."

Supercomputers at SC05

The higher-end LS-X has a faster internal network using InfiniPath technology from start-up PathScale. The network connects directly to HyperTransport links on Advanced Micro Devices' Opteron chip, the only processor available on the LS-X.

In a market where heavy hitters--such as Sun Microsystems, IBM, Hewlett-Packard and even Dell--have launched corresponding products, Linux Networx has tried to make a name for itself. And it has had some success, including a soon-to-be announced sale of a 5,000-processor cluster to an Energy Department laboratory.

At the same time, however, a handful of companies are pushing more traditional supercomputer designs that often have a single operating system spanning many processors. Those companies include Cray, IBM, Silicon Graphics, NEC and Sun.

At the outset, the LS-1 will have just computing nodes, but its expandable design means the company will add different hardware options later in 2006, Passarelli said. Those extensions will include modules for storage; graphics; and hardware accelerators that speed up some tasks, he said.

Both systems come with hardware and software components tested to work with each other, Passarelli said. Both will use Advanced Micro Devices' Opteron processor at the outset, but in the first quarter of 2006 the LS-1 will be available with Intel Xeon chips after the new dual-core "Dempsey" models arrive.

The LS-1 has a starting price of $40,000 for a 16-processor machine, the company said; a 64-processor machine costs $130,000 and up. Prices weren't immediately available for the LS-X. Both will be shipping in limited quantities this quarter and will become generally available in the first quarter of 2006.

Each server in a typical LS-1 is a dual-processor node, though four-processor nodes will be available, with 40 nodes per rack, Passarelli said. The LS-X uses four-processor nodes only and as many as 24 per rack. At maximum, the LS-1 accommodates as many as 128 nodes, while the LS-X handles as many as 6,144 nodes through the use of a more elaborate network.

At Seattle's supercomputing show, Linux Networx will show a visualization-oriented system for showing complex graphic images. The system uses visualization software from Computational Engineering International and graphics chips from Nvidia.