Not so long ago heterogeneity in HPC was coming mostly on the processing side: within a compute node we had several processing units of different architectures (let it be the Cell processor, or a combination of traditioanl CPUs, GPU accelerators, and/or FPGAs). Currently, heterogeneity has expanded with the advent of multiple explicitly-addressable memory subsystems featuring different technologies, but also in the form of clusters of compute nodes featuring completely different hardware configurations (we find a recent example in the MareNostrum 4 supercomputer). Programming models and runtime systems have been charged with the task of easing the programming burden of this heterogeneity. In this talk I will review current, upcoming, and foreseeable heterogeneity in HPC, as well as some of the proposed and available ways of handling it. I will also highlight some of the remaining challenges.
Antonio holds a BS + MS degree in Computer Engineering (2006), and MS and PhD degrees in Advanced Computer Systems (2010, 2013), from Universitat Jaume I, Spain. He is currently a Sr. Researcher at the Barcelona Supercomputing Center (BSC), Computer Sciences Department. At BSC, Antonio works within the Programming Models group where he leads the "Accelerators and Communications for HPC" team. Antonio is also the Manager of the BSC/UPC NVIDIA GPU Center of Excellence and member of the BSC Outreach Working Group. He is a Juan de la Cierva Fellow and prospective Marie Curie Fellow. He is a recipient of the 2017 IEEE TCHPC Award for Excellence for Early Career Researchers in High Performance Computing. His research interests in the area of runtime systems and programming models for high performance computing include resource heterogeneity and communications.
Antonio was formerly with Argonne National Laboratory (USA, 2012-2015), driving the heterogeneous memory and accelerator computing areas of research within the Programming Models and Runtime Systems group, where he was the technical lead of heterogeneous memory and accelerator virtualization projects. He was also part of the core MPICH R&D team. His PhD dissertation, in a joint collaboration between the Universitat Jaume I and the Universitat Politecnica de Valencia (Spain) started the rCUDA remote GPU virtualization project, for which he was awarded the Extraordinary Doctoral Award from the Jaume I University.
