Parallel Supercomputing
Ever since arriving at Los Alamos I have been intrigued by the
power of parallel supercomputers in solving difficult problems in
nonlinear science. My main interest is in extracting physics out of
massive simulations while, at the same time, avoiding becoming a slave
to them. This is not easy to do, but it is possible. I have used
parallel computers to solve PDEs (including stochastic and conditioned
PDEs) and to solve very large N-body problems.
My supercomputing experience began in the early nineties with the
Thinking Machines CM-2000 and then on to the Thinking Machines CM-5
(the most usable parallel platform ever?). Both of these machines were
at LANL's Advanced Computing Laboratory. Next to come were the Cray
T3E at the National Energy Research Scientific Computing Center
(NERSC) and the SGI Origin 2000 cluster at the ACL. At present I am
utilizing the IBM SP3 at NERSC, the QB and QSC machines at LANL (Alpha
clusters), local Beowulf clusters, and, soon, the IBM SP4 at Oak
Ridge.
| Salman Habib / LANL / habib@lanl.gov / revised December 2002 |
 |
The National Energy Research Scientific
Computing Center at LBNL.