Parallel and Distributed Computing in Scientific Applications
- Victor Bazterra, University of Illinois at Chicago
Friday, May 10, 10:00 AM - Special Seminar
Nuclear Conference Room
The increase of the computing power at low cost in the last decades made possible the widespread adoption of parallel and distributed computing for large scientific problems. Loosely coupled clusters of commodity computers are now a well established technology, enabling the creation of Grid infrastructure to share computational resources between geographically separated institutions. Another recent trend is that processing power is growing over the time due to architecture changes toward a massive increase of the number of cores per processor rather than an increase of the processor speed. In this talk, I will show some of my contributions in enabling scientific applications to use this type of resources, from structure prediction of materials to high energy physics. I will illustrate different user case scenarios and the solutions I helped to develop. I will also discuss the current efforts in multithreading and multiprocessing computing for data analysis and simulation in high energy physics.