Recent Highlights We have 3 papers accepted at this year's SC supercomputing conference! The first compares 1106 implementations of 6 graph algorithms to determine which parallelization and implementation styles work well and under what circumstances. The second describes a GPU approach for computing minimum spanning trees (MSTs) that is 4.5 times faster than the fastest prior code. The third presents a new parallel algorithm for GPUs to detect strongly connected components (SCCs) in mesh graphs 8 times faster than the fastest prior implementations.
Short Biography
Martin Burtscher is a Professor in the Department of Computer Science at Texas State University.
He received the BS/MS degree from ETH Zurich and the PhD degree from the University of Colorado at Boulder.
Martin's current research focuses on the parallelization of graph algorithms and complex programs for GPUs
as well as on the synthesis of high-speed lossy and lossless data-compression algorithms.
He has co-authored over 120 peer-reviewed scientific publications, which have
been cited over 6500 times.
Martin is a distinguished member of the ACM and a senior member of the IEEE.
Current Research Areas
GPU computing, parallel algorithm design, data compression, graph analytics, algorithm synthesis, performance optimization, energy efficiency
I am always looking for creative and motivated students at all levels who are interested in working on these and related topics with me. Please contact me if you are interested.
Efficient Computing Laboratory
Martin directs the Efficient Computing Laboratory (ECL). Its research goals are developing general
strategies for parallelizing complex and irregular programs, creating techniques to automatically synthesize
high-performance data-compression algorithms, and designing optimizations to improve performance and
energy efficiency. Animations and more information are available here.
Faster computations and better algorithms can help save lives, solve health problems, increase safety, improve
the environment, and keep us connected. At this point, parallelization is the primary way to make more powerful
and energy-efficient computers possible. As we are reaching the limits of human ability, automatic synthesis is
the most promising avenue for creating new and improved algorithms.
Projects CIVIC (verification of irregular parallel programs) [NSF/CCF - SHF] LC (high-speed lossy data compression framework) [DOE/ASCR - DRS] PECOS (simulation of inductively coupled plasma torch) [DOE/NNSA - PSAAP III]
Teaching Material Lecture slides:
Teaching modules (introduction to parallel programming for undergraduates) Programming projects:
Peachy assignments (computing a movie of zooming into a fractal)
Selected ECL-member highlights
NSF Fellowships ($138,000): Jared, Kristi, Molly
Outstanding thesis awards: Sepideh
Outstanding research awards: Jared, Molly, Sepideh
Best paper awards: Yiqian, Noushin
PhD positions at Texas State, U. of Oregon, UT Austin, U. of Utah
Industry positions at NVIDIA, Intel, Samsung, Thermon, USAA, Charles Schwab, Uber, etc.
