Networking Solution Delivers High ROI For Commercial HPCC Applications, Featuring Seamless Deployment and 3X Improvement In Latency and Throughput

Sunnyvale, CA and Westborough, MA, February 14, 2005 – Chelsio Communications, Inc., the established leader in 10-Gigabit Ethernet server adapters and protocol acceleration technology, and Scali Inc., the leader in scalable, high performance cluster software solutions, today announced the Chelsio T210 MPI Connect 10 Gigabit Ethernet cluster networking solution. The solution provides advanced performance levels for commercial High Performance Cluster Computing (HPCC) applications such as Eclipse, Fluent, Magma, LS-Dyna, TurboWorx, Abaqus, and others. In addition, the T210 MPI Connect offers up to 3X throughput and latency improvement over existing standard cluster interconnect configurations.

Based on Chelsio’s recently announced T210 Protocol Engine and Scali’s MPI Connect, the leading, fully integrated commercial MPI solution, the T210 MPI Connect will be demonstrated this week in the Chelsio booth (#1607) at LinuxWorld in Boston.

Robert Desautels, president of Harvard Research Group, said “Scale-out clusters exhibit unparalleled performance gains for users of high-performance computing. The Chelsio/Scali partnership clearly shows the strategic value of standards-based, 10 Gigabit Ethernet cluster fabrics for enterprise-class, commercial HPC environments.”

Kianoosh Naghshineh, president and CEO of Chelsio, said “Clustered solutions have exhibited a high rate of growth in the technical computing market, with Gigabit Ethernet clusters remaining very attractive due to their high return on investment (ROI). The T210 MPI Connect enables the vast installed base of Ethernet-based commercial HPCC clusters to seamlessly gain new levels of performance and scalability through harnessing the power of 10Gb Ethernet.”

New levels of performance for commercial HPC applications were achieved when the T210 MPI Connect was deployed on clusters built from Intel Itanium and AMD Opteron based servers. Actual application results showed that throughput vaulted from 222 MB/s to 830MB/s – a 270% increase – and latency dropped from 26.2 micro-seconds to 10.5 micro-seconds -- a 250% reduction when compared with GigE solutions.

Bjørn Skare, president and CEO of Scali said, “With clusters built from Linux-based commodity computers increasingly taking on the challenges of commercial high performance computing, the importance of the networking, messaging solutions and management has become critical. Scali and Chelsio both firmly believe that fully supported, industry-standard networking hardware and software technology is a critical requirement for rapid adoption of Linux clustering in the commercial HPC market.”

For applications running MPI, Scali MPI Connect is designed to reduce the complexity of running a mix of platforms within a Linux cluster environment. Chelsio and Scali have worked together to certify the Scali MPI Connect software on Chelsio’s T210 Protocol Engine.

To aid in cluster deployment and maintenance, Scali Manage dramatically eases system management for customers that build Linux cluster solutions from a range of hardware and software combinations. Scali Manage is configured to seamlessly install Scali MPI based applications in cluster environments. Scali Manage and the T210 MPI Connect solution are available immediately from Chelsio Communications.

About Chelsio Communications Chelsio Communications is leading the convergence of networking, storage and clustering interconnects with its robust, high-performance and proven protocol acceleration technology. Featuring a highly scalable and programmable architecture, Chelsio is shipping 10-Gigabit Ethernet adapter cards with protocol offload, delivering the low latency and superior throughput required for high-performance computing applications. For more information, visit the company online at www.chelsio.com.