"Data Analytics of Supercomputer Field Data for Reliability" presented by Nathan DeBardeleben, Los Alamos National Laboratory
In this talk, Dr. DeBardeleben will discuss results from data analysis of supercomputer field data in collaboration with AMD. He will talk about neutron-induced errors and how his team tests electronics in the neutron beam at Los Alamos National Laboratory. Results from tens of billions of device hours will be shown related to DRAM, SRAM, and the impacts of error correction codes on the reliability of some of the most powerful supercomputers in the world.
Nathan DeBardeleben is the lead of the Ultrascale Systems Research Center (USRC) at Los Alamos National Laboratory. At USRC, Nathan works in the areas of data analytics and resilience with a focus on studying environmental effects to supercomputer hardware. Nathan came to LANL in 2004 after finishing his PhD in computer engineering from Clemson University. He works extensively with vendors on next generation hardware designs and evaluating the fault tolerance of future technologies.
Friday, September 8 at 2:30pm to 3:30pm
McAdams Hall, 119
821 McMillan Rd., Clemson, SC 29634, USA