This project is a new Darpa funded collaboration led by Joel Saltz at the University of Maryland and also involving Apostolos Gerasoulis at Rutgers. The project is an application oriented approach to performance prediction .
We are developing a suite of simplified applications and a corresponding simulator Petasim which is designed for qualitative performance predictions as is needed in the rapid prototyping phase of machine design. Syracuse has designed this simulator which will be developed as a Web Resource using Javabean technology
In November 97, we released first prototype of Petasim with a Java frontend and C++ backend.
Syracuse will use its work on Black Hole Grand Challenge as a particular motivating application for
Petasim while Maryland will use its I/O intensive applications from data fusion (Land Cover Dynamics) and medical areas.
The name Petasim comes from the series of Petaflops
studies where the rapid turnaround of qualitative models of novel machine architectures
was clearly needed. For further information contact Dr.
Geoffrey Fox at NPAC, Syracuse University.