The largest high-performance computers that will be available in the near future will have important differences from those machines we are used to programming. The primary change is that gains in parallelism will be obtained through the addition of more processor cores, and more hardware thread contexts will be available to hide memory latency. As a result, programmers will have to expose extreme levels of parallelism (billions of threads) to efficiently utilize these machines. Another major concern for application programmers and system designers is the reduction in mean time between failures expected for machines with so many components. This talk will elaborate on these issues, discuss the state-of-the-art, and present results of early prototypes of new parallelism models for quantum chemistry that are designed to accommodate the challenges presented by extreme-scale computers.

Challenges in the utilization of extreme-scale high performance computers by quantum chemistry applications

242nd National Meeting (2011)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.