Appro International, a systems maker based in Milpitas, Calif., is announcing Nov. 13 that it has delivered two supercomputer Linux clusters to the Lawrence Livermore National Laboratory.
In building the clusters for the Lawrence Livermore, based in Livermore, Calif., Appro said the machines have the potential to deliver 100 teraflops of computer processing power on a Linux-based platform.
One of the clusters that has already been delivered to the lab and tested—called “Rhea”—offers 22 teraflops of processing power.
Although impressive, these clusters pale in comparison to IBMs Blue Gene/L system that is installed at the lab and performs at 280.6 teraflops or 280.6 trillion calculations per second.
Even so, when the four clusters are combined, the machines will rank in the top three at the lab, which is run by the U.S. Department of Department of Energy, according to Appro.
In building these four interconnected clusters, Appro used Advanced Micro Devices Opteron 8000 series processor.
When all four clusters are finished, the machines will deliver a total of 2,592 four socket dual-core Opteron-powered nodes. This will give each node eight processing cores.
Each node carries 16GB of DDR2 (double data rate) of RAM, according to Appro. In the first cluster delivered to the lab, for example, that machine contained 576 nodes.
Addison Snell, an analyst at IDC, believes that Appros contribution to the Lawrence Livermore lab will help establish its reputation in the field of supercomputing.
“This is a significant win for Appro and establishes their products as viable contenders in the HPC [high-performance computing] space,” Snell wrote in an e-mail to eWEEK.
“Mark Seager [assistant department head for Advanced Technologies] at LLNL is a leader in high-performance cluster computing, and Appro should benefit from the association.”
The company received the commission from the lab to build the clusters in 2005 and developed each machine as a scalable unit, said Adrian Wu, a senior field application engineer for Appro.
The first two of these machines, Rhea and “Zeus,” have been delivered to the lab. Appro is in the process of delivering the third cluster, called “Atlas,” and the fourth and final cluster, “Minos” will arrive in the first quarter of 2007.
The clusters are also the only Linux-based machines operating at the lab. Wu said that lab officials were looking for machines that would be cost-effective, standardize all the Linux functions on a single platform and give the scientists at the lab a capacity computer.
“This is a real work horse for them,” Wu said. “Its a work horse that will run 24 hours a day, seven days a week without any tweaking, and it doesnt fail but will keep running.”
By using AMD processors, Wu said Appro will give the scientists at the lab a chance to upgrade the clusters without changing most of the hardware. This also is expected to help with reducing costs associated with the project.
“The U.S. Department of Energy and NNSA are committed to the acquisition of cost-effective systems that can be rapidly deployed to benefit a broad range of scientific research in the national interest,” said Seager, who leads the labs new platform development program, in a statement.