Faculty of Engineering and Natural Sciences
Combinatorial Algorithms for Parallel Scientific Computing
Sparse matrix-vector multiplication (SpMxV) operation abounds in scientific computing. Therefore, efficient parallelization of SpMxV operations is of prime importance to many applications. Previous works on parallelizing SpMxV operations consider maintaining the load balance among processors and minimizing the total message volume. In this talk, we will discuss that communication cost depends on some other factors in addition to the total message volume. We will clearly define those factors and propose combinatorial techniques to minimize them.
Bora Uçar received the B.S. (1997), M.S. (1999), and PhD (2005) degrees in Computer Engineering from Bilkent University, Ankara, Turkey. After receiving the PhD degree he joined to the Department of Mathematics and Computer Science, Emory University, Atlanta, Georgia as a post-doctoral research associate. Much of his work has been in combinatorial scientific computing--the design, analysis and application of discrete algorithms for applications in scientific computing.
July 25, 2006, 10:40, FENS G035