Advanced Comput, Math & Data
Staff Awards & Honors
CSMD Adds Familiar Face to its Scientific Ranks
Since first joining PNNL as a post-graduate research assistant in July 2012, Gokcen Kestor has continued to grow as a valued part of the Computational Sciences & Mathematics Division’s High Performance Computing (HPC) group. In early May 2014, Kestor, who completed her Ph.D. in 2013, was promoted to a full-time research scientist with the HPC group, where she will continue her efforts in performance and power modeling, analyzing the energy cost of data movement in HPC application, and designing and implementing fault-tolerant algorithms for task-based applications. One of her most recent achievements included work on an energy characterization of data movement that was profiled in an article featured in HPCwire—work categorized as “really important, impacting practical aspects of co-design for extreme-scale systems” by CSMD Director Adolfy Hoisie. (Read the related research highlight here.)
In addition to her work in power and performance modeling, Kestor’s research interests include programming models, runtime and compiler support for programming models, fault-tolerance, transactional memory, transactional memory benchmarking, thread level speculation, and data race detection. In 2011, she was the primary author of “RMS-TM: A Comprehensive Benchmark Suite for Transactional Memory Systems,” which earned Best Paper honors from the International Conference on Performance Engineering. She also was the lead author of “Quantifying the Energy Cost of Data Movement in Scientific Applications,” which was nominated for a 2013 IEEE International Symposium on Workload Characterization Best Paper award. Kestor earned both her masters and Ph.D. in computer engineering from the Universitat Politècnica de Catalunya in Spain.