Efficient and Scalable Communication and System Software for Exascale Computing

点击翻译按钮获取中文摘要

High-Performance Computing (HPC) is used to tackle computationally-intensive problems in fields as diverse as green energy and biofuels, cancer research and drug discovery, weather forecasting and climate change, seismic processing for oil and gas, genomics and bioinformatics, astrophysics, material science, automotive, defense, data mining and analytics, and financial computing. It is the key to many scientific discoveries and engineering innovations. However, the demand for more computational power is never ending. Far more computational power on much larger-scale machines is required to unravel the scientific mysteries. The Message-Passing Interface (MPI) is the de facto standard for communication in HPC systems, and by far the dominant parallel programming model used by large-scale scientific and engineering applications. Processes in such applications compute on their local data while extensively communicating with each other through the interconnection networks. MPI has proved to be scalable and is smoothly transitioning in the current systems. However, on extreme-scale systems that are characterized by massive parallelism, highly hierarchical architectures and communication channels, smaller memory per core and heterogeneity, there will be immense pressure on the interconnection networks and communication system software to deliver the required performance and scalability. This research seeks to address the challenges for high-performance and scalable communication subsystems on extreme-scale systems. The proposed research is highly original and innovative in the sense that it addresses key questions in high-performance communications and system software in MPI and hybrid MPI+X programming models. Such research will pave the way for adoption by industry. The outcome of this research will be relevant to various sectors in Canada, including Environment Canada, Compute Canada, Canada Genome Sciences Centre, automotive and oil/gas industries, and ultimately the Canadian public at large. It is expected that the findings from this research will have significant impact on the target community, and that it will lead to new directions for future research. In such a key and fast-paced field, the results of this research will keep Canada at the forefront of science and technology. The proposed research is ideal for training HQP in that it has a strong foundation that translates immediately into practical applications and implementations. There is a high demand for graduates in HPC and networking, and the HQP trained will be well positioned to compete for jobs in academia and industry.