SC Conference - Activity Details

Efficient, Real-Time Parallel Computing

Gail Walters  (CPU Technology Inc.)
Exhibitor Forum Session
Software Applications and Virtual HPC
Tuesday,  03:30PM - 04:00PM
Room 12A/12B
Parallel and distributed computing allows large and complex compute tasks to be broken up into smaller, more manageable jobs, and run on separate cores and servers. This is particularly important when you have a large or complex simulation class computing problem. Although the smaller jobs run more quickly in parallel, the challenge of this approach lies in the delays associated with communicating the results of each job, synchronizing the execution of the many jobs and then propagating the next round of tasks for each job. CPU Tech has eliminated this communication and synchronization bottle neck for the large and complex simulation class of computing problems and is achieving real-time simulation performance. This session will discuss the results we have seen accelerating simulations of real-world problems at real-time speeds. We will also discuss how we have achieved this while reducing the size, power and cooling requirements by 90+% compared to traditional computing resources.
   IEEE Computer Society  /  ACM     2 0   Y E A R S   -   U N L E A S H I N G   T H E   P O W E R   O F   H P C