SC Conference - Activity Details

GPU Computing: A New Approach to an Existing Problem

Andy Keane  (NVIDIA)
Exhibitor Forum Session
GPU Software Tools
Wednesday,  11:30AM - 12:00PM
Room 19A/19B
The inherently parallel architecture of today's GPUs can offer HPC professionals a boost in speed anywhere from 8X to over 400X, depending on the application. However, if developing software for the parallel processor is complicated, developers will not use it no matter the potential performance. New programming models, such as the CUDA programming environment have also entered the industry seeking to address this challenge - allowing developers to use the familiar, industry standard language of C. CUDA allows developers to distribute work across the "many-cores" architecture of GPUs without the need to learn a new and complex programming language. As a new, alternative way of approaching the challenges associated with data-intensive computation on a processor technology already present in consumer PCs, workstations and servers, the potential of GPU Computing is huge. This presentation will discuss these exciting new technologies in direct relation to real challenges facing organizations and will examine deployments of this technology in fields such as oil and gas, finance and medical imaging.
   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