Caitlin Ross

Institution/Organization: Rensselaer Polytechnic Institute.

Department: Computer Science.

2nd Institution/Organization: Argonne National Laboratory.

Academic Status: Graduate Student.

What conference theme areas are you interested in:

Computational science and machine learning;
Statistical modeling, methods, and computation;
High performance software: packages and design;
Algorithms at extreme scales;
Data science, analytics, and visualization.


My PhD research is in performance analysis and visualization of parallel event-driven simulation, in order to support the codesign of next generation supercomputer systems. The synchronization method used in our simulations allows for simulation events to be processed speculatively, so sometimes the system must be rolled back to a prior state in order to correct any events processed out of order. My work has been developing tools that will help us understand performance issues related to this synchronization protocol because traditional parallel performance analysis tools cannot detect the periods of unproductive work.