Andrey Semin
Keynote speakers
- Kenji Doya
- Alon Halevy
- Astrid Prinz
- Andrew Schwartz
- Shankar Subramaniam
- Arthur Toga
Workshop speakers
- Bart ter Haar Romeny
- Uri Eden
- Klaus Linkenkaer-Hansen
- Tim Clark
- Alan Ruttenberg
- Jeffrey Grethe
- Arnd Roth
- Wulfram Gerstner
- Peter Hunter
- Markus Diesmann
- Andrey Semin
- Pietro Liò
- Albert Cardona
- Giorgio Ascoli
- Rolf Kötter
Workshop 4, High performance computing and grid infrastructure for neuroinformatics applications
Andrey Semin
Title: Advancing High Performance Computing solutions for neuroinformatics applications
Intel, Munich, Germany
Abstract: Neuroinformatics applications for many years are among the most demanding for computation performance, and scientists foresee continued demands for large computational capacity for wide range of medical and life sciences applications. However there are number of challenges in software and hardware development that needed to be solved in order to provide scientists with scalable supercomputers capable to deliver high sustained performance for large-scale simulations. We first review the history of supercomputer hardware developments, and then outline trends highlighting limiting factors which designers foresee on the way of building large scale supercomputers, usually involved power and cooling requirements of the supercomputers, their size and cost. Though hardware difficulties can be overcome with continued technology development and we'll likely see continued increase in the peak performance of large-scale systems, but it is important that these supercomputers continue deliver high sustained performance for challenging scientific applications. And this is where the software starts paying the bigger role. By software we understand not just the application itself, but the whole stack of software, starting with programming paradigm and language, spanning to middleware and system software. Many of the previously developed software paradigms and programming languages have not been designed to take the full advantage of hardware of the future. Thus many companies have started research and development of new programming languages and paradigms better suited for the supercomputers we'll likely see in several years, so we devote significant part of discussion to the software aspects of supercomputing, and impacts that these software paradigm shifts have on software development practices.
Bio sketch: Andrey Semin leads Intel's HPC Technology group in EMEA (Europe, Middle-East and Africa) region, responsible for assistance of enabling, design, and deployment of Intel-based supercomputers for scientific and industrial applications. His team is engaged with leading European HPC users and vendors helping deliver new and innovative HPC solutions to grand challenge problems at affordable cost of commodity components. Prior to moving to Germany in 2004, he was for 4 years with Intel in Russia in various roles in HPC solutions engineering and development. Andrey Semin received a Master of Science in Physics from Moscow State University (Russia) in 2000, and completed Ph.D work in 2002 at the Department of Physics specializing in possibility theory and its applications for physical experiment analysis. His current research interests are in the area of applications of possibility theory to image analysis, object detection and classification in real-time.