The Colloquium Series of the Department of Computer Science, University of Wyoming presents Dr. Paolo D'Alberto Post Doctoral Fellow, Carnegie Mellon University "Compiler Analysis and Optimizations: A Vertical Approach for Recursive Algorithms" Tuesday, March 20, 2007 EN 1062 4:00 - 5:00 p.m. Abstract: The development of divide and conquer (D&C) algorithms for matrix computations has led to the widespread use of high-performance scientific applications and libraries. In turn, D&C algorithms can be implemented using loop nests or recursion. Recursion is extremely appealing because it is an intuitive means for the deployment of top-down techniques, which exploit data locality and parallelism naturally. However, recursion has been considered impractical for high-performance codes, mostly because of the inherent overhead of the division process into small subproblems and the difficulty of analyzing the resulting computation leaves, where the bulk of the work is done and implemented as loop nests. In this talk, Dr. D'Alberto will present a vertical approach for the analysis and optimization of D&C recursive algorithms, including: * techniques for the analysis of the division process of recursive algorithms, these techniques are embodied into a lite C compiler JuliusC; * polyhedra-based approach for the analysis and optimizations of cache interference for parametrized loop nests (STAMINA). This work has been developed in collaboration with Prof. Nicolau (UCI). Biography: Dr. D'Alberto is currently a Post Doctoral Fellow, Electrical and Computer Engineering, Carnegie Mellon University. He is affiliated with the SPIRAL project, responsible for the design and implementation of techniques for the SW-HW partitioning of DFT for low-power and high-performance custom systems. He received his Ph.D. in Computer Science from the University of California, Irvine in 2005; a Dottorato di Ricerca in Computer Science from the University of Bologna in 2000; and the B.C. in Computer Engineering from the University of Padua in 1995.