Keynote 3: Tuesday (August 23, 2022) 2:00 pm - 3:00 pm

Abstract:

Today, AI, big data and scientific computing use different software stacks，although they share many common requirements, such as expressing the parallelism, load balance and fault tolerance. It is interesting to witness that these applications domains are converging. For example, AI for science has already made impressive achievements in areas such as protein folding and molecular dynamics. The data processing tasks are usually used to prepare, transform and manage data for both AI and scientific computing applications. In this talk, we propose FABS(Fused AI, Big data and Scientific computing programming) which uses a single software stack for all these domains, which could reduce the cost and complexity of inter-system operations. We also discuss some advances in the FABS paradigm.