to Showcase New Energy Efficient HPC Technology at SC 2009

Share Article

System Fabric Works to demo the first 16 node plugcluster featuring lower power and lower cost per node computing

At the SC 09 conference and exhibition in Portland, Oregon, will showcase in its booth (No.137) new computing solutions developed by System Fabric Works (SFW), designed specifically for the high performance computing market (HPC).

The mission of the Plug Computing initiative is to make high-performance, always on, always connected, energy efficient, low power and environmentally friendly computing readily available for developers and end users, and to facilitate a free and open exchange of ideas and software.

SFW will demonstrate the first 16 node plug cluster built for HPC market based on the Marvell SheevaPlug platform. The SheevaPlug platform uses a Marvell Kirkwood processor based on an embedded 1.2GHz Sheeva™ CPU equipped with 512 Mbytes of FLASH and 512 Mbytes of DRAM. Connection to the SC09 SCinet network is via Gigabit Ethernet. Disk storage for the file system is directly attached storage using a USB 2.0 port. A standard Fedora Linux 2.6 kernel distribution is running with MPI for the parallel applications and NFS for the file system. The plugs take power from a commercial power strip and draw less than one tenth of the power of a typical PC being used as a server.

Visit during SC09 (Nov. 16-19, 2009) at the Oregon Convention Center, Booth No. 137
777 NE ML King Blvd.
Portland, OR 97232

Schedules of the presentation about can be found at:

Information about can be found at:

Information about System Fabric Works can be found at:

More information:
The uses of Kirkwood processors and the plug computing ecosystem are rapidly expanding to include a full value chain of product partners, application developers and educational institutions.


Share article on socal media or email:

View article via:

Pdf Print

Contact Author

Bill Boas

Daniel Yoo
Visit website