| |
| We wanted the industry's highest performance interconnect for Red Squall. We found that our applications scale better with Quadrics' QsNetII interconnect than any other HPC interconnect on the market at this time. | Doug Doerfler, Principal Member of Technical staff in the Computer and Software Systems Department at Sandia National Laboratories |
| - | |
| The type of Science and Engineering calculations required by Livermore's national security mission require a cluster of this size and a very high bandwidth and low-latency interconnect that provides demonstrable and scalable performance. | Dr. Mark Seager, Assistant Department Head for TeraScale Systems, Lawrence Livermore National Laboratory. |
| - | |
| Quadrics QsNetII offers the highest MPI performance available on systems based on standard processing nodes. | Fabrizio Petrini, Computer & Computational Sciences Division, Los Alamos National Laboratory. |
| - | |
| Quadrics is a major vendor of high-speed interconnect technology, a crucial element in building scalable and well-balanced computational cluster. Instead of researchers deploying small clusters in their offices, it's far more productive to build larger machines. Larger machines, when properly run, significantly lower the cost of ownership. The GEaRS group serves over 250 researchers in several academic departments which include engineering (aerospace, chemical, electrical, mechanical), mathematics, statistics, physical sciences, and biological and life sciences. For such a diverse and interdisciplinary set of users, it is particularly important to design compute engines which enable parallel computations for codes with varying but demanding communication requirements. When it comes to measuring and tuning the performance of real applications as opposed to generic benchmarks, highest performing interconnects are a must for large computational systems. | Vijay Agarwala, director of graduate education and research services, Penn State University. |
| - | |
| It is important to choose the right interconnect to meet the needs of the applications. Some applications are embarrassingly parallel and don't need to solve a single large matrix calculation . Other applications need to use all the processors in concert with each other and scale to thousands of processors. For these problems, the premium interconnects are clearly the only option. We've run our codes on Quadrics interconnect and they scale phenomenally | Scott Studham, MSCF Operations group leader, Pacific Northwest National Laboratory. |
| - | |
| Our simulation program requires outstanding performance from the network interconnecting the nodes of such a large machine, especially in terms of latency and barrier synchronization. Quadrics, who we chose for our previous computer TERA-1, is still today the only one able to meet such a challenge. | Jean Gonnord, Numerical Simulation & IT program director at CEA/DAM |
| - | |
| An essential aspect for choosing Quadrics QsNet as the high-speed interconnect for this system was reliability, performance and robustness for an industrial production environment. Quadrics technology combines best performance in latency and bandwidth, a robust MPI production software environment and scheduling software for managing a large user base. | Jürgen Winkelmann, Head of System Services at ETH Zurich |