Tarek Ghazawi
Exascale and the Convergence of High-Performance Computing, Big Data, AI and IoT
Abstract
The field of high-performance computing (HPC) or supercomputing refers to the building and using computing systems that are orders of magnitude faster than our common systems.
The top supercomputer, Summit, can perform 148,600 trillion calculations in one second (148.6 PF on LINPAC). The top two supercomputers are now in the USA followed by two Chinese supercomputers.
Many countries are racing to break the record and build an ExaFLOP supercomputer that can perform more than one million trillion (quintillion) calculations per second.
In fact the USA is planning two supercomputers in 2021 one of which, when fully operational (Frontier), will perform at 1.5 EF.
Incidentally, data volumes due social media and the internet of things (IoTs) have been exploding and AI has been a successful technique with advances in deep learning to leverage those large volumes of data.
Those concurrent developments have thus resulted in what is seen as the Convergence of Big Data and HPC as processing massive data amounts become impractical without HPC. In this talk we examine the progress in HPC and potential applications and capabilities of such convergence as the basis for a future smart world.
We also briefly take a bird’s eye peak at the race for creating postMoore’s law processors to address existing challenges beyond the exascale.
