Our load balancing work “Fast Dynamic Load Balancing Tools for Extreme Scale Systems” will be presented at SIAM CSE17 on March 1st from 4:30-6:30 pm as part of the PP4 Poster Session.
@KennethEJansen and I spent a week at the Argonne Leadership Computing Facilties (ALCF) Theta Workshop learning about Knights Landing, the Intel and Cray tools, and running performance and scaling studies. Thanks to the ALCF, Intel, and Cray staff we were able to run PHASTA on over 3Ki nodes on meshes balanced with our tools.
Mark Shephard, @gahansen, @mperego, Vijay Mahadevan, and I, presented FASTMath Unstructured Mesh Technologies at this year’s Argonne Training Program on Extreme-Scale Computing (ATPESC). Along with the talk we prepared a hands-on demonstation of an adaptive workflow for simulating Greenland ice-sheet flow.