Computer software exhibits complex interactions with underlying hardware, making understanding performance difficult. Using a memory reference trace, a list of memory transactions performed by software at runtime, enables deep exploration into the structure of software execution. Unfortunately, this data is extremely large and cryptic behavior makes direct analysis difficult.
In my talk I will overview 4 approaches which combine visual and computational analysis approaches to better communicate the data. Our first approach uses abstract visual encodings, animating event streams so that the user can directly observe runtime behavior. Next, a topologically-based approach is developed that finds and visualizes cyclical patterns in the normally linear reference trace as spiral structures expanding out into the time dimension. Third, an ensemble-based method visualizes side-by-side several reference traces, or a single trace simulated through different cache configurations, bundled into a 'cache simulation ensemble'. Finally, I will overview recent work visualizing massively parallel systems programmed using CUDA. Several case studies will illustrate the approaches.
Paul Rosen received his PhD in 2010 from Purdue University where his dissertation was about Camera Model Design, a problem solving paradigm that advocates designing dynamic, application specific camera models for problems in computer graphics, visualization, and computer vision. Dr. Rosen joined the Scientific Computing and Imaging Institute at the University of Utah in 2010 and currently holds the position of Research Assistant Professor.
Dr. Rosen’s research interests include a wide variety of areas in computer graphics, scientific visualization, and information visualization with a focus on approaches which combine computational analysis and interactive visual exploration to address problems with large heterogeneous datasets.
Date: Friday, October 12, 2012
Time: 10:30am to 11:30am
Location: ECS South 2.410
Refreshments will be served at 10:15am