Presenter

Vivek Kale is a Computer Scientist at the University of Southern California's 's Information Sciences Institute. Vivek works on improving performance of scientific codes run on supercomputers with multi-core nodes. During his PhD, Vivek worked under the supervision of Professor William D. Gropp and was a Lawrence Scholar at Lawrence Livermore National Laboratory, working with Todd Gamblin and Bronis de Supinski. Vivek's dissertation work focused on developing lightweight loop scheduling strategies to improve the scalability of bulk-synchronous MPI+OpenMP applications run on clusters of SMPs. He also was involved in work on MPI shared memory extensions model for MPI-3, also known as the MPI+MPI model. Vivek's current work involves performance optimizations that include techniques of auto-tuning and loop scheduling of ptychography solvers intended to run on supercomputers having nodes with GPUs and/or MICs.