Dr. Thomas H. Cormen
October 27, 2000
Suppose you need to perform a Fast Fourier Transform (FFT) on so many
points that the data do not fit in the memory of your largest computer.
Instead, the data reside on a parallel disk system (since one disk is not
large or fast enough). In such an "out-of-core" setting, the usual FFT
algorithms run slowly due to their inefficient pattern of disk accesses.
Significant performance gains can be realized by permuting the data in
particular ways throughout the computation. Dr. Cormen will discuss how
to perform out-of-core FFTs with parallel disks on both uniprocessors and
multiprocessors, and in one dimension and multiple dimensions.
[Joint work with David Nicol, Jake Wegmann, Lauren Baptist, and Michael Ringenburg.]
Dr. Thomas H. Cormen is an Associate Professor in the Dartmouth College
Department of Computer Science, where he has worked since 1992. He received
the B.S.E. degree in Electrical Engineering and Computer Science from Princeton
University in 1978 and the S.M. and Ph.D. degrees in Electrical Engineering
and Computer Science from the Massachusetts Institute of Technology in 1986
and 1993, respectively. Professor Cormen is coauthor of the textbook,
Introduction to Algorithms, which he wrote with Charles E.
Leiserson and Ronald L. Rivest. His primary research interests are in
out-of-core computing: developing I/O-efficient algorithms for
out-of-core problems and applying the technology of parallel I/O
algorithms to design virtual memory systems for data-parallel
computing. He is also interested in debugging technology.
Everything you always wanted to know about out-of-core FFTs, but were afraid to ask
Associate Professor
Dartmouth College
Center for Environmental Sciences & Technology Management (CESTM) Auditorium
University at Albany
4 p.m. to 6 p.m.
Sponsored by the University at Albany's Division for Research
