Stormy weather drives massively parallel algorithms

Prof. Lou Rossi has joined forces with a large interdisciplinary team at UD and the National Center for Atmospheric Research (NCAR) to develop new algorithms to perform multiscale computations of cloud physics. The team hopes to achieve computational speeds in the Peta-Flop (10^15 floating point operations per second) regime on some of the world's fastest computers. The team leader is Prof. Lian-Ping Wang from Mechanical Engineering. In addition to Prof. Rossi, team members include Wojciech W. Grabowski, senior scientist in the Mesoscale and Microscale Meteorology Division at NCAR, Prof. Guang R. Gao, from Electrical and Computer Engineering; Prof. Chandra Kambhamettu from Computer and Information Sciences; Prof. Xiaoming Li, from Electrical and Computer Engineering and Andrzej A. Wyszogrodzki, scientist in the Research Application Laboratory at NCAR. This project will be funded by a $1,000,000 NSF grant and a separate $300,000 grant to NCAR.

The specific aim of this proposal is to capture the essential physics of droplet coalescence coupled to the large scale fluid dynamics of a turbulent cloud. This type of investigation requires resolving flow features from over lengths scales traversing many orders of magnitude. Performing this type of science requires advances in algorithms and computing hardware. Historic studies covering the last few decades have shown that improvements in scientific computation can be attributed in equal parts to improvements in algorithms and hardware. Prof. Rossi and his graduate student Claudio Torres will bring their expertise on fast algorithms, parallel computing and computational fluids to boost the performance and efficiency of the existing cloud physics model. The team members will work together to take advantage of new hardware paradigms including the use of graphical processing units (GPUs) to boost performance. The team will work together to answer critical scientific questions about cloud evolution and dynamics.

For more information, see the UDaily article.

Article created: September 2, 2009