Lead Chief Investigator
|
Project Title
| Towards the Development of Self Optimizing Electronic Structure Software for High Performance Computing Platforms |
|---|
Brief Description for General Publications
Building fast, reliable, and adaptive software is a constant challenge for computational science, especially given recent developments in computer architecture and a move towards peta-scale computing. Our research seeks to address these issues in the context of computational chemistry applications. To maximize execution speed we are developing simple performance models that can be used to predict performance as a function of the underlying machine architecture. For reliability, we are seeking to use interval arithmetic to bound the effect of both truncation and rounding errors on the results obtained from complex calculations. For adaptivity we are investigating new programming models and methods that enable a running application to dynamically change its behaviour based on current runtime conditions. |