September 21st, 2017
About Us
Our Sponsors

Sirti Bestows Award of Honor
October 21, 2010. Billie Moreland, PhD, of Billie Moreland and Associates and Steve Simmons, PhD, E...

Triangle %u201CGraduates%u201D at Sweet Sixteen
The Idea (1994)
In June 1994, while on a layover in the San Francisco airport, Dr. Steve Simmons cam...

  More News


06/30/2008 - Steam Plant Hosts Multi-Teraflop ASL Computer Cluster

Washington State University’s Applied Science Laboratory’s (ASL) has installed a new high-performance computing (HPC) cluster for applied research in the physical sciences and engineering. The machine has been installed in the Steam Plant and is connected to the Riverpoint Campus via VPnet. At present, the cluster is being optimized for performance and benchmarked. There are jobs waiting in the queue for this teraflop-level computing resource – as soon as all the code is adapted to the new platform, according to Santanu Chaudhuri, ASL research scientist. ASL is the Spokane-based applied research component of WSU’s Institute of Shock Physics. Funding for the HPC cluster comes from the State of Washington and the Office of Naval Research. Additionally, Chaudhuri thanks Avista for hosting the cluster in the Steam Plant building and Sirti for administering the network connectivity.

ASL performs contract research for clients in Spokane and the region. This is scientific work -- including materials research, modeling and simulations – with problems that require a lot of computing power. Ten years ago, this kind of work could only be done at large national laboratories, e.g. PNNL, because only they could support the infrastructure and expense for very specialized computers, according to Chaudhuri. “Today we have commodity computing,” says Chaudhuri, “the capabilities of modern HPC clusters have increased and are now affordable for research universities.” Chaudhuri and the other ASL researchers plan to keep the cluster busy by using 90% of the computing power, twenty-four hours a day, seven days a week.

The ASL HPC cluster is unlike many of today’s supercomputers – it is flexible and can work on a wide variety of scientific problems. The HPC cluster was produced and customized by IBM based on their 1350 series Linux clusters and will initially be used for quantum chemical calculations.

What a machine it is! The HPC consists of three 42U racks mounted with densely stacked compute “nodes”. There are sixty-four of these compute nodes. Each pizza box shaped node has two dual-core, 3.0 GHz Intel Xeon 64-bit EM64t chips, which translates into four acting processors for each node. This means 256 working processors (4 x 64). This distributed memory, highly parallel HPC cluster has 16 gigabytes of RAM for each node – a titanic amount of memory. Overall, there is 1 terabyte of RAM in the computer nodes. The cluster receives and processes data at 20 Gbps using Cisco InfinibandTM connectivity infrastructure and OpenIB protocols. The operating system is SUSE Enterprise Linux 10. Disc storage of 1.6 terabytes is available for users, 4.6 terabytes of total scratch space and more hot-swappable expansion slots are available for future storage needs. The file system runs ultrafast GPFS parallel file system to handle the highly demanding parallel-I/O requirements.

Storage is not the biggest need with this cluster. Scientific computing needs speed and number crunching power. The calculations will produce large data sets that will be shared and distributed between the nodes – parallel computing. The management node is very smart and distributes the computing requests to whichever node has RAM and other system resources available. After the data sets have been distilled to the final values, these values are stored so that the resulting final files are not large. As smart as this HPC machine is, it still requires Chaudhuri and his fellow ASL researchers to design the problem for the computer. They have to gauge how much computing power to throw at a problem, and use mathematical analysis to determine how (or whether) the problem can be “parallelized.” The scientists will adopt new algorithms as they are improved and become available, and will plug those into the computer’s modular programs for even faster utilization of the total computing power.

A computer of this size requires a huge amount of power. This one requires up to 48 kW power, (currents of 270 amps at 208 volts) when operating at the peak performance levels. It gives off a huge amount of heat, so cooling is critical. An 18-ton air conditioning unit is required to keep it sufficiently cooled. The system is designed with heat exchanger doors that can be linked in the future to chilled-water systems to lower the load on the HVAC system. The system monitors temperature and throttles up or down, by controlling the speed of the cooling fans in relation to the load on the nodes.

This is the first step in building ASL’s scientific computing facility. ASL will continue to do contract research in energy, national security, and advanced materials, and is working to recruit more scientists to Spokane. They also hope to attract more businesses to Spokane.

For more information about the ASL , visit

Billie Moreland
» News Archive

Sirti Bestows Award of Honor
Triangle %u201CGraduates%u201D at Sweet Sixteen
Triangle Concludes Newsletter Operations
ISAGA 2010 Conference Slated at Riverpoint
Terabyte Tidbits
Education Robotics Expands in Region
Terabyte Tidbits
Triangle Welcomes Caelus Consulting
Area High Schools Join Robotics Competition
Terabyte Tidbits
Triangle Welcomes Apple Guy
Fine Solutions Offers Flexible ERP
Lewis and Clark Robotics Faces FIRST Showdown
Terabyte Tidbits
Interlink Debuts Surveillance System
Spokane Entrepreneurial Center
Terabyte Tidbits
EWU Computer Science Celebrates 25 Years
Terabyte Tidbits
Triangle Welcomes Spokane Web Communications