Come for Table discussions, Member Self-Intro, What's New, Application Showcase, and Advanced Application Development Techniques! Exchange ideas, meet experts, share code... all HPC & GPU, all practical, all cutting-edge.
6:15-7:00pm What’s new: Report from SIGGRAPH (Steve Worley, Worley Labs)
7:00-7:10pm Member self-intros: 30 seconds for each member
OpenCL on FPGA (Delayed until future meetup)
7:10-8:10pm When Galaxies Collide: Treecodes with Dynamic Parallelism (Stephen Jones, NVIDIA)
Refreshments courtesy of NVIDIA.
Bio for Stephen Jones, NVIDIA
Stephen Jones is an engineer in NVIDIA's high-performance computing group, working on the CUDA language and programming model for the GPU. His background is in computational fluid dynamics and plasma physics, and he has worked on CUDA's parallel mathematical libraries as well as hardware architecture, system infrastructure, and language design. His current work is focused on the design of the CUDA language to better support HPC applications, and the future GPU architectures required to enable it.
Abstract for When Galaxies Collide: Treecodes with Dynamic Parallelism
Astrophysical N-body simulations are some of the most compute-intensive scientific models that are used today. The direct all-to-all gravity calculation scales exponentially as problems grow, which has led researchers to try alternative approximation approaches to handle large scientifically interesting datasets which are common today.
We will look at one of these approaches - the Barnes-Hut N-Body treecode - and the benefits and challenges involved with implementing this on GPU hardware. At the recent GTC 2012 conference, a real-time implementation at astrophysically interesting scale was demonstrated on the latest GK110 GPU hardware. This talk will use the demo to illustrate how the new dynamic parallelism support in the hardware was used to benefit the simulation.
Carnegie Mellon Silicon Valley;
NASA Research Park Bldg 23;
Mountain View, CA 94043;
Directions to Carnegie Mellon Silicon Valley;
Google Map showing parking, check point, and building entrance;
NOTE: You will need a government issued ID (e.g. Driver's License) to enter NASA Research Park