November 24, 2006
Researchers in a number of national and international universities are using the shared-memory processing power of SGI high performance compute technology to explore and solve a wide range of complex mathematical problems in pure and applied math. The University of Waterloo (UW), in Ontario, Canada, recently upgraded the processing power of two SGI Altix systems with SGI InfiniteStorage arrays for a variety of fields, including ocean wave modeling and computer sciences.
The Mathematics Faculty at UW currently uses two Altix systems. The University has added processors, memory and storage as their needs have grown since the initial Altix installation in 2004. UW currently has one Altix 3700 system with 64 Itanium 2 processors and a total of 4.6 TB in two InfiniteStorage TP9100 arrays, and one Altix 350 system, recently upgraded to 16 processors from the original dual Itanium 2 processor configuration.
The Fluids group in Applied Math uses the Altix 350 system for ocean wave modeling, and writes their own home-grown code, which has been enhanced by SGI's tuned numerical libraries. Professor of Applied Mathematics Kevin G. Lamb, currently department chair, specifically requested the Altix 350 for high-resolution computational fluid dynamics (CFD) simulations, running code he originally wrote over 15 years ago for his study of internal gravity waves in the ocean.
Professor Lamb purchased an SGI S330 storage array with 5.6 TB for the Altix 350 system; the Canadian Foundation for Climate and Atmospheric Science was a significant source of funding. Other purchases and upgrades were made possible by Canada Foundation for Innovation, the Ontario Innovation Trust and Canada's National Science and Engineering Research Council (NSERC) "Research Tools and Instruments" grants.
"I bought the Altix to look at the interaction of stratified tidal flow with bottom topography in the ocean. This process generates internal gravity waves of many different length and time scales, and I needed high resolution to simulate as large a range of scales as possible," said Lamb.
"Over the years I've altered my code considerably and used it on many different computers. When I was in the market for a new computer, I ran my model on a lot of different machines. I tested my code, as well as a number of other things, on an Altix and it was the fastest and most cost-effective of all the machines I was looking at. One thing I found is that making use of SGI's scientific computation library sped up the code significantly. I've been quite happy with the results."
One of the Mathematics groups at UW that first used the SGI Altix 3700 system, when it was originally delivered, is the School of Computer Science. Current fields of research include design and implementation of algorithms for exact solution of linear algebra problems, and exploring pattern information for program visualization tools to aid in the development and debugging of parallel code.
"We chose Altix 3700 because Computer Science specifically wanted Itanium processing for its integer performance compared to other processors. Floating point performance was impressive as well," said Robyn Landers, a software specialist in the Math Faculty Computing Facility, University of Waterloo. "They wanted open-source Linux and shared memory processing. At the time, SGI was the only solution with more than 16 processors in a single system image. We were anticipating future expansion, so the modular design of SGI's architecture was very attractive."
With the most recent upgrade in September 2006, the Altix 3700 system has been expanded to a 64-processor machine with 192 GB memory, running Novell SUSE Linux Enterprise Server, Version 9. Along with the School of Computer Science, several other groups from the Mathematics Faculty are now using Altix 3700 technology. These include: Pure Math for number theory and random matrix theory; Combinatorics and Optimization for data mining, optimization of algorithms in both theoretical and physical realms (manufacturing, computer applications); and Applied Math, where mesh-free methods for modeling 3D events (with applications in medicine, climatology and geophysics) are being explored.
"The ready to use, single system performance and flexibility provided by SGI Altix shared memory systems allows mathematics researchers and scientists to explore the largest experimental data sets without doing them piecemeal," said Michael Brown, sciences market segment manager for SGI. "Ever-expanding data sets and simulations in almost all areas of science require reliable and economic solutions that allow researchers to develop new approaches to solving problems and SGI Altix systems have proven time and again that they meet those high performance demands. SGI InfiniteStorage solutions easily scale to back up and share the vast amounts of valuable research that will lead to the biomedical, environmental, and computational innovations of tomorrow."
10/30/2013 | Cray, DDN, Mellanox, NetApp, ScaleMP, Supermicro, Xyratex | Creating data is easy… the challenge is getting it to the right place to make use of it. This paper discusses fresh solutions that can directly increase I/O efficiency, and the applications of these solutions to current, and new technology infrastructures.
10/01/2013 | IBM | A new trend is developing in the HPC space that is also affecting enterprise computing productivity with the arrival of “ultra-dense” hyper-scale servers.
Ken Claffey, SVP and General Manager at Xyratex, presents ClusterStor at the Vendor Showdown at ISC13 in Leipzig, Germany.
Join HPCwire Editor Nicole Hemsoth and Dr. David Bader from Georgia Tech as they take center stage on opening night at Atlanta's first Big Data Kick Off Week, filmed in front of a live audience. Nicole and David look at the evolution of HPC, today's big data challenges, discuss real world solutions, and reveal their predictions. Exactly what does the future holds for HPC?