Richard Vuduc

Richard Vuduc

Assistant Professor

Faculty Bio: http://www.cse.gatech.edu/people/richard-vuduc

High Performance Computing Area

Performance and Autotuning

Research Areas

Software analysis, tuning, and debugging for multicore and manycore systems; architecture; compilers; statistical machine learning; and computational science

Honors and Distinctions

Best paper at Parallel and Distributed Testing and Debugging at International Symposium on Software Testing and Analysis
Best paper at International Conference on Parallel Processing
Best student paper at Supercomputing - Finalist
Best student paper at Performance Optimization of High-Level Languages and Libraries (POHLL) Workshop at International Conference on Supercomputing (ICS)
Best student presentation at POHLL/ICS
Best presentation at Feedback-directed Dynamic Optimization workshop, at MICRO
Outstanding Graduate Student Instructor Award at University of California-Berkeley
Cornell Tradition Fellowship - Cornell University

Research Summary

In the emerging era of "parallelism everywhere," a central question facing high performance computing researchers and practitioners is how to produce the fastest and most energy efficient software possible for massively parallel hardware with the least programming effort. Vuduc's lab is examining this question through an integrated research program in three areas: (1) exploring new programming models with the potential to change the way we think about and write parallel software; (2) discovering techniques for building algorithms and software that can self-adapt automatically to the underlying hardware ("auto-tuning"); and (3) identifying new approaches for finding and preventing bugs in massively concurrent programs. Collectively, the ultimate goal of these activities is to simplify the development, analysis, tuning, and debugging of parallel software.