Scott Haney is a technical staff member at the Advanced Computing Laboratory at Los Alamos National Laboratory. He works on the POOMA framework, which can be downloaded at http://www.acl.lanl.gov/pooma. His interests include high-performance scientific computing and active library design using C++. He received BS degrees in physics and nuclear engineering from the University of California, Santa Barbara and a Ph.D in fusion reactor technology from the Massachusetts Institute of Technology.

POOMA (Parallel Object-Oriented Methods and Applications) is a C++ framework for constructing computational science applications on platforms ranging from laptops to parallel supercomputers. POOMA 2.2, our latest release, provides a sophisticated, generalized Array class along with Field and Particles abstractions. Multi-threaded, out-of-order execution based on dataflow analysis is supported using the SMARTS runtime system. These facilities allow scientists to write their codes using a high-level data parallel language, leaving issues like parallelism and cache management (mostly) to the framework. I will present an overview of POOMA 2.2 along with our plans for future releases. I will also discuss aspects of the framework's design, which allows developers to easily plug in their own convenient data representation along with context-sensitive optimizations of the evaluation process to achieve high performance.