Developing PIC Codes for the Next Generation Supercomputer using GPUs

Viktor Decyk
University of California, Los Angeles (UCLA)

The current generation of supercomputer (petaflops scale) cannot be scaled up to exaflops (1000 petaflops), To achieve exascale computing, new classes of highly parallel, low power devices are being designed. The most successful so far has been the Graphical Processing Unit (GPU). I will describe how our research in developing algorithms for Particle-in-Cell codes for GPUs should help us exploit the next generation supercomputer when it arrives.

Presentation (PDF File)

Back to Workshop I: Computational Challenges in Hot Dense Plasmas