Research Interests and Activities
Performance Analysis and Visualization of Parallel Applications
My key interests are in the visualization and analysis of performance
data and information from parallel applications. The work is centered around
Projections, a framework with a trace-generation component
built-into the Charm++ Runtime System and as well as a
Java-based visualization and analysis tool.
The Parallel Programming Laboratory has a fairly detailed Project
Page describing our work in this field.
My thesis research attempts to study the additional capabilities to
performance visualization and analysis derived from predictive
performance information generated through What-If techniques. I will
update the following URL as I develop the ideas and research on the topic.
Capabilities in Predictive Parallel Performance Analysis (no link yet).
Projections Analysis of Performance Issues in Classical Molecular Simulations using NAMD
In conjunction with my main research focus, I work closely with in
collaboration with members of the Theoretical and Computational Biophysics
Group to successfully port and maintain the stability and
performance of NAMD on various large parallel machines.
My typical research activites in this area involve the study of the
scaling of the ApoA1 simulation benchmark under NAMD run on different
machines like the Apple G5
Cluster (turing) at the Computational Science and Engineering
(CSE) Department, UIUC; the SGI
Altix machine at the National Center for Supercomputing
Applications (NCSA), UIUC; the