Al Geist's Home Page

Al Geist Oak Ridge National Laboratory Corporate Fellow,
Chief Technology Officer of the Oak Ridge Leadership Computing Facility, and
and Chief Technologist of the Computer Science and Mathematics Division of Oak Ridge National Laboratory

What does Al do?

Exascale Computing Challenges: extreme parallelism, resilience, memory hierarchy, and power

Principal Investigator of the Extreme-scale Algorithms and Software Institute. It's EASI!
A DOE Joint Math/CS Institute develop architecture-aware algorithms for scalable performance and
resilience on heterogeneous architectures.

Development of Oblivous Fault Tolerant algorithms that can survive multiple simultaneous errors.

Extreme-scale system simulation — collaborating with Christian Engelmann on a system simulator
to study the behavior of algorithms at extreme scale. In June 2011 set the world record
of simulating a 100 million task MPI application.

Co-Principal Investigator on Holistic Fault Awareness project developng an API for all layers
of software stack to both tell as well as be notified of faults in the system.

Led a national consortium of labs, universities, and industry to develop Scalable Systems Software
for extreme-scale supercomputer Centers.

Was Principal Investigator on the IAA Algorithms project to develop architecture-aware algorithms for multi-core processors.

Heterogeneous distributed computing

Technical manager PVM development team. To date PVM has had more than 14 million downloads.

Co-Principal Investigator on Harness the next generation virtual machine environment beyond PVM
Here is my 2000 EuroPVM-MPI conference keynote talk (powerpoint slides)
Subcommittee Chairman in the group that designed MPI Standard.

MPI-1 chairman of the collective communication chapter
MPI-2 co-chairman of the process creation and management chapter
MPI-3 investigating how to incorporate fault tolerance into MPI.

Collaborative Technologies and Web Programming

PI on the Electronic Notebook Project - in use by over 10 thousand users world wide for research, education, and medicine.
The software is freely available. Check it out.

JavaLand - Al teaches workshops and maintains a site to help Java programmers

Genomes to Life

Was co-principal investigator on one of three national Genomes to Life Centers.
The goal is to develop advanced computational algorithms and tools needed to identify all the molecular machines and regulatory pathways in a single-cell organism.
Also lead the Computational Environments and Infrastructure subproject team.

Sample Publications

Books
PVM: Parallel Virtual Machine. A Users' Guide and Tutorial for Networked Parallel Computing.: MIT Press, 1994. (with Beguelin, Dongarra, Jiang, Manchek, Sunderam).
MPI: A Message-Passing Interface Standard.: (with The MPI Forum).
Papers
Development of Naturally Fault Tolerant Algortihms for Computing on 100,000 Processors: Journal of Parallel and Distributed Computing (submited 2002 with Engelmann).
Parallel Out-of-core Algorithms for Genome-Scale Enumeration of Metabolic Systemic Pathways: Proceedings of High Performance Computational Biology Conference (2001 with Samatova, Ostrouchov, and Melechko).
M3C - A Java Tool for Managing Multiple PC Clusters: Journal of Parallel and Distributed Computing (1999 with Schwidder).
Cluster Computing: Wave of the Future?: Springer Verlag Lecture Notes in Computer Science, 1994.
Reduction of a General Matrix to Tridiagonal Form.: SIAM J. Matrix Anal. Appl. Vol. 12 No. 2, p. 362-373.
Parallel Superconductor Code on iPSC/860.: The Journal of Supercomputing, 6, 153-162. (with Ginatempo, Shelton, Stocks). Gordon Bell Prize Winning Entry.
Modeling of Nonequilibrium Melting and Solidification in Laser-Irradiated Materials.: Phys. Rev. B 34(4), p.2606. (with Wood).
References to all 182 of Al's papers.