Recent News
UNM to host nuclear engineering students from around the country
March 14, 2025
CCEE faculty encourage prospective students to attend UNM
March 14, 2025
Computer Science Colloquium will discuss strategies for sustainable AI data centers
March 10, 2025
Wei selected as New Mexico Women in Tech awardee
March 7, 2025
News Archives
UNM Computer Science PhD Student George Bezerra Awarded Excellence in Graduate Research Award
June 12, 2012
Computer Science PhD student George Bezerra recently received the 2012 UNM Chapter of Sigma Xi “Excellence in Graduate Research” Award. The UNM Chapter of Sigma Xi, an international, multidisciplinary research society, presents the Excellence in Graduate Research Award to encourage and recognize the research performed by a doctoral student near the end of his or her Ph.D. dissertation.
Bezerra will share his research in a colloquium titled “Communication Locality and Energy Consumption in Chip Multi-Processors” on Friday, June 15 at 10:00 am in the Centennial Enginering Center Auditorium. A reception follows at 11:00 in the Stamm Commons.
Bezerra is graduating from UNM this summer and starting a postdoc at MIT in the Fall. He works with modeling and optimization of energy consumption in modern computer architectures, in particular multicore chips with dozens of cores. He is also interested the similarities between power consumption in chips and metabolism in biological organisms as these systems scale in size. He holds a Bachelors degree in Electrical Engineering and a Masters degree in Computer Engineering both from the University of Campinas, Brazil.
About the Colloquia:
Performance increase of future computer architectures will be driven by the growing parallelism of multi-core chips and constrained by power consumption. To take full advantage of the multi-core design, the communication patterns of parallel applications must be optimized through careful mapping of data to cores, so that communication distances and energy consumption are minimized. We present a new method for data placement in Chip Multi-Processors (CMPs), which reduces on-chip energy consumption by targeting communication locality and load-balancing. Our method is exact and can be solved in polynomial time, improving on earlier heuristic approaches that do not provide guarantees on solution quality. Simulations on a 64-core system showed an average reduction of dynamic energy consumption of 49.8% (and as much as 84.1%), with performance gains of up to 16.9% on parallel scientific benchmarks.