UTCS Colloquia - Albert Cohen, Snr. Research Scientist, INRIA, "A Streaming Data Flow Approach to Correct and Efficient Parallel Programming"

Contact Name: 
Keshav Pingali
GDC 4.304
Jun 6, 2013 11:00am - 12:00pm

Talk Audience: UTCS Faculty and Graduate Students

Host:  Keshav Pingali

Talk Abstract: This talk will engage into a short exploration, starting with weak memory models and ending with directions for automatic parallelization, discussing runtime system and compiler construction along the way. We will present recent and ongoing work on the generation of efficient parallel code with in-place, array-based computations, from dynamic data-flow representations of programs.

Speaker Bio: Albert Cohen is a senior research scientist at INRIA and a part-time associate professor at Ecole Polytechnique, Paris, France. He graduated from Ecole Normale Superieure de Lyon, and received his PhD from the University of Versailles in 1999 (awarded two national prizes). He has been a visiting scholar at the University of Illinois in 2000 and 2001, and an invited professor at Philips Research and NXP, Eindhoven in 2006 and 2007.  Albert Cohen works on optimizing compilers for high-performance and embedded systems, automatic parallelization, data-flow and synchronous programming. He coauthored more than 100 peer-reviewed papers, and he is or has been the advisor for 18 PhD theses. Several research projects initiated or led by Albert Cohen resulted in the transfer of advanced compilation techniques to production compilers.