"The Cray XD1 stands out from the competition because it combines
a very high-speed interconnect, high-scalability and high-availability features, and unique possibilities for additional application acceleration," said Rich Partridge, enterprise systems analyst at D.H. Brown Associates. "It's no surprise that major ISVs are rapidly aligning themselves with Cray to exploit this innovative system for CAE applications."
The Cray XD1 supercomputer is specifically designed to enable CAE
users and their employers to bring higher-quality products to market
faster and at lower overall cost than Linux clusters and other
alternatives in the Cray XD1 system's price range. The new product
makes true supercomputer performance available for the first time at
mainstream pricing and in a familiar environment -- the Cray XD1
supercomputer runs all x86 Linux applications using the MPI
"Cray will focus initially on the automotive, aerospace and marine
segments within the CAE market," said Himanshu Misra, CAE business
manager at Cray. "We have a long and proud history in these markets,
more than any other HPC vendor, having sold the first supercomputer
into the automotive industry in 1979. And Cray pioneered the use of
HPC with CAE customers, such as Ford, Boeing, the Army High Performance Computing Center and others."
Cray is currently working with more than a dozen ISVs on CAE codes
for the XD1, including Abaqus, Inc., ACUSIM Software, Inc., CD Adapco, ESI Group, Fluent, Livermore Software Technology Corporation, Mecalog SARL, and Software CRADLE, plus a co-development effort with Oceanic specifically for the Marine industry. Misra added that additional ISV agreements are expected in the coming months.
About the Cray XD1 Supercomputer
With U.S. list prices starting under $100,000, the Cray XD1 supercomputer combines AMD's Opteron(TM) processors and HyperTransport(TM) technology with an advanced Cray interconnect to
provide industry leading performance and value. Designed for superior
sustained application performance, the Cray XD1 system uses the direct connected processor architecture to directly link processors to each other and memory, eliminating interconnect bottlenecks and providing 30 times greater bandwidth and 30 times lower latency than typical cluster systems available today.
HPC users can also take advantage of the Cray XD1 system's
sophisticated management, self-monitoring and self-healing features to simplify system administration and ensure high availability. Other
Cray XD1 innovations include applications acceleration capability
based on Field Programmable Gate Arrays (FPGAs) linked directly to
processors over the Cray XD1 interconnect.