MC2 Logo Multicore Computational Center

Vision

To revolutionize the productivity of computational applications by taking advantage of multi-core parallel processing software and hardware technologies in collaboration with industry, government and universities. Closing Quote

Mission

To evolve to a leading national center in providing service oriented computational solutions employing multi-core technologies for the optimization of problems in the fields of environmental and geophysical sciences, chemical, aerospace, defense, bio-medical informatics, financial, and event driven simulations and visualizations. Closing Quote

Recent Updates:


MC2 Brown Bag Lunch Series

Click EVENTS tab for MC2 brown Bag Lunch Series.



Director's Blog, June 20th, 2009

As we conclude the fourth quarter of the second year of the MC2, it gives me great pleasure to share with you and other visitors to this web blog the exciting research and accomplishments of our staff.

First , I want to extend congratulations to our two graduate students, Nancy Walia and Navid Golpayegani, who were awarded their Master's degree in May both having completed highly innovative and challenging research thesis. Nancy's work on "Parallel External Suffix Array Construction" showed the scalability of the DC3 algorithm executed on the IBM bluegrit system and hence can be used to construct and search for strings in huge (TB) suffix arrays. This met the objectives of her sponsored research by the Laboratory of Telecommunication Sciences. With her mentors, John Dorband, Yaacov Yesha, and Michael Ferguson, a paper for publication is being prepared for journal submission. Navid showed in his thesis titled "Gridding Earth Science Data with Hadoop" that by implementing the Hadoop framework on our bluegrit JS20 series that the MapReduce parallel programming paradigm showed significant speed improvements over our previous approach of separately distributing the gridding of a days collection of satellite radiance to a given blade. His work also showed that the code produced with Hadoop is independent of the gridding resolution, allowing the generation of gridded data sets at arbitrary, user specified, resolutions without incurring heavy compute delays. A paper joint with his mentor, Milt Halem, titled "Cloud Computing for Satellite Data Processing on High End Compute Clusters" was submitted to the IEEE International Conference on Cloud Computing that can be viewed on our MC2 web site. Read more

CSS Template by Rambling Soul | Valid XHTML 1.0 | CSS 2.0