GPU Computing: Designing flexible systems for rapid evolution
SCIENTIFIC COMPUTING invites you to an educational webcast GPU Computing Designing flexible systems for rapid evolution   Using graphics processing units to run applications has become one of the hottest trends in high performance computing. The past few years have seen major changes, including increasing migration of single instruction, multiple data (SIMD) computations to highly-parallelized GPU environments available in current hardware configurations, and the ensuing significant performance increases many HPC algorithms have achieved through the use of hybrid GPU computing. Given this opportunity to dramatically improve performance, it is imperative to build cost-effective systems that utilize GPUs effectively. It is equally vital that these systems not be mere point solutions that satisfy only the needs of today‚Äôs GPU applications when these applications are evolving so rapidly. In this discussion, our expert panelists will look at approaches to GPU system configuration design and will examine the importance of maintaining flexibility to adapt to the inevitable evolution of applications and tool sets, including the ability to mix and match components to meet application, workload and user requirements. Benchmark test results also will be presented to illustrate how a flexible approach to GPU computing works and how it affects performance and system design.   Presented by: In Association with:  Wednesday, October 26, 2011 2:00 p.m. ET; 11:00 a.m. PT Duration: 1 Hour   Moderator Suzanne Tracy Editor-in-ChiefSCIENTIFIC COMPUTING Speakers Tommy Minyard, Ph.D. Director of Advanced Computing Systems Texas Advanced Computing CenterThe University of Texas Austin      Rob Farber Chief Scientist and Co-founderBlackDog Endeavors, LLC    Mark R. Fernandez, Ph.D. HPC Computer Scientist Advanced Systems GroupDell Inc.