Tutorial 3: Affordable High-Performance Computing via ATM


Patrick W. Dowd, State University of New York at Buffalo
Saragur M. Srinidhi, NASA Lewis Research Center

ATM and cluster-based computing provide a possible solution to the need for supercomputing power without the economic implications. If a fraction of the idle CPU cycles of workstations could be harnessed together, it would be a very valuable resource. Cluster-computing is becoming increasingly important as evidenced by the number of large corporations who have recently replaced supercomputers with clusters. This tutorial begins with a description of ATM and proceeds to an introduction of distributed computing, including the software environment, such as message-passing libraries and other required-system level software. Specific examples are examined, and the advantages and disadvantages for both local cluster-based platforms and geographically distributed systems are quantified.


Patrick W. Dowd received a B.S.E.E. from the State University of New York at Buffalo, and the M.S. and Ph.D degrees in electrical engineering from Syracuse University. He was with the IBM Corporation as a staff engineer with System Design at the IBM Glendale Processor Development Laboratory. His main effort was in the area of processor communication subsystem design for future systems. Dowd joined the Department of Electrical and Computer Engineering at SUNY-Buffalo as an assistant professor in 1989. He is on the editorial board of the International Journal on Computer Simulation. He is a member of IEEE, ACM, and SPIE, with research interests in optical interconnects, distributed and parallel computer architectures, and computer communication.

Saragur M. Srinidhi received a B.E. in electronics and communications engineering from Bangalore University in India and M.S. and Ph.D degrees in electrical engineering from Cleveland State University. He is a former systems engineer in the Networks Group at HCL India, New Delhi, and was a telecommunication analyst with BP America. He joined Sterling Software's Scientific Systems Division in 1992 where he is a senior consulting engineer with the High-Performance Networks group at NASA Lewis Research Center in Cleveland, Ohio. He is a principle investigator for NASA's ACTS ATM experiments and is involved in prototyping wide-area ATM networking in support of dispersed computing. His research interests are in performance modeling, congestion control, and transport level issues in ATM. Srindhi is a member of the Graduate Faculty at Cleveland State University and a member of the IEEE and ACM.