GPU 2014 - SPECIAL SESSION ON GPU COMPUTING

For the next decade, Moore's Law is still going to bring higher transistor densities allowing Billions of transistors to be integrated on a single chip. However, it becomes obvious that exploiting significant amounts of instruction-level parallelism with deeper pipelines and more aggressive wide-issue superscalar techniques, and using most of the transistor budget for large on-chip caches has come to a dead end. Especially, scaling performance with higher clock frequencies is getting more and more difficult because of heat dissipation problems and too high energy consumption. The latter is not only a technical problem for mobile systems, but is even going to become a severe problem for computing centers because high energy consumption leads to significant cost factors in the budget. For the moment, improving performance can only be achieved by exploiting parallelism on all system levels. Multicore architectures like Graphics Processing Unit (GPU) offer a better performance/Watt ratio than single core architectures with similar performance. Combining multicore and coprocessor technology promises extreme computing power for highly CPU-time-consuming applications like in image processing. The Special Session on GPU Computing and Hybrid Computing aims at providing a forum for scientific researchers and engineers on hot topics related to GPU computing and hybrid computing with special emphasis on applications, performance analysis, programming models and mechanisms for mapping codes.
Topics of interest include, but are not limited to:
GPU computing, multi GPU processing, hybrid computing;
Programming models, programming frameworks, CUDA, OpenCL, communication libraries;
Mechanisms for mapping codes;
Task allocation;
Fault tolerance;
Performance analysis;
Applications: image processing, signal processing, linear algebra, numerical simulation, optimisation;
Domains: computer science, electronic, embedded systems, telecommunication, medical imaging, finance.