Presenter

Ke Fan is a fourth-year Ph.D. student in the computer science department at the University of Alabama at Birmingham, advised by Sidharth Kumar. My research orientations are high-performance communication and I/O and scientific data analytics and visualization. Applications running on exascale systems must adapt to increasing degrees of parallelism that are going to put tremendous stress on overall data movement, be it over the interconnect network or to the parallel storage system. It becomes necessary to have systems that have highly scalable data exchange and I/O phases — any degree of load imbalance will lead to sub-optimal performance. My research focuses on developing solutions for managing imbalances in both parallel I/O and a sub-set of communication that have been traditionally overlooked by HPC users. In particular, I have developed 1) solutions for dealing with load-imbalance created by modern compression libraries and 2) optimization of both uniform and non-uniform all-to-all communication among processes. I am also a straight-A student in UAB, and I have already published two conference papers at HiPC and HPDC, a workshop paper at IPDPS, and a poster at SC -- all top-tier conference venues in high-performance computing (HPC). I am deeply committed to my research and will devote myself to HPC.