Wednesday, March 12

MS8
Revolutionary Technologies for Acceleration of Emerging Petascale Applications - GPUs

10:00 AM - 12:00 PM
Room: Atlanta B

The HPC community is in the midst of a significant paradigm shift, as power limitations are driving hardware architects to increasingly rely multi-core technologies. As a result, computational scientists are examining alternative architectures that address the limitations of modern cache-based designs. In this minisymposium, we examine the potential of several radically different micro-architectural approaches: the heterogeneous STI Cell, the many-core NVIDIA and ATI GPGPUs, and the double-precision—enabled Xilinx FPGA. The invited speakers will present an overview of the hardware, programming environment, and application performance of these architectures, while highlighting their tremendous potential for high-end scientific computations.

Organizer: Jeffrey S. Vetter
Oak Ridge National Laboratory
Leonid Oliker
Lawrence Berkeley National Laboratory
Rupak Biswas
NASA Ames Research Center

10:00-10:25 NVIDIA CUDA Software and GPU Parallel Computing Architecture

Michael Garland, NVIDIA

10:30-10:55 Implementation of the Navier-Stokes Stanford University Solver (NSSUS) on a GPU

Patrick LeGresley, Eric F. Darve, and Erich Elsen, Stanford University

11:00-11:25 Accelerating Molecular Modeling Applications with Graphics Processors

John Stone, James C. Phillips, Peter Freddolino, David J. Hardy, and Leonardo Trabuco, University of Illinois at Urbana-Champaign; Klaus Schulten, University of Illinois at Urbana-Champaign

11:30-11:55 Performance and Productivity of Graphics Processing Units for a Quantum Monte Carlo Application

Jeremy Meredith, Oak Ridge National Laboratory