Maria Garzaran

Ph.D. University of Zaragoza, 2002

Research Statement

My research focuses on automatic program optimization. Computers are becoming very complex, as the addition of new architectural features makes the interaction between hardware and software very difficult to understand. The result is that program optimization is hard and costly. The approach that I am pursuing is to use empirical search to try to automate the process of program optimization. The idea is to search among the different versions of a program to find the one that perform best. These different versions can be obtained by applying a program transformation with different parameter values, such as loop unrolling and different degrees of unroll. Some other times, however, these versions come from different algorithms that solve the same problem, such as quicksort and radix sort when sorting a sequence of numbers.