The International Conference for High Performance Computing, Networking, Storage, and Analysis
The International Conference for High Performance Computing, Networking, Storage, and Analysis
Authors: Paul Carpenter (Barcelona Supercomputing Center)
Abstract: Radical changes are foreseen in all aspects of high-performance computing systems over the next decade. In this context, in 2017, the European Eurolab-4-HPC project published its long-term vision for academic HPC systems research. This session, which follows a successful BoF at ISC 2019, is part of the process of preparing the 2020 update. In a lively and hopefully controversial discussion, involving expert presenters and the audience, we will identify, discuss and assess the most important long-term challenges and opportunities in HPC systems research. The overall goal is to build a research community that is high quality and relevant.
Long Description: Radical changes are foreseen in all aspects of high-performance computing systems over the next decade. The end of CMOS scaling is bringing new and emerging technologies (quantum, neuromorphic, etc.) and heterogeneous accelerators (GPUs, FPGAs, TPUs, etc.). Meanwhile, HPC simulations are being complemented by machine learning and extreme data and are evolving to model problems more difficult than previously thought possible. In this context, we need a thriving, high-quality and relevant community in academic HPC systems research. This session will bring together academic researchers and interested stakeholders to discuss, assess and contribute to Eurolab4HPC’s vision for academic research in high performance computing. Expert presenters will engage with the audience in a lively and hopefully controversial interactive session.
Eurolab4HPC is a two-year European Horizon 2020 funded project with the bold commitment to make Europe excel in academic research and innovation in HPC technology (https://www.eurolab4hpc.eu/). Eurolab4HPC has the overall goal to strengthen academic research excellence and innovation in HPC in Europe, by structuring the HPC community, disseminating community news, promoting entrepreneurship, and stimulating technology transfer.
Eurolab4HPC published the first version of its Eurolab4HPC Long Term Vision on High-Performance Computing in August 2017 [1]. This document targeted the period 2023–2030, over which it predicted major changes across high-performance computing. It surveyed the landscape of HPC systems research, driven by a push from emerging technologies and a pull from new and evolutionary applications.
This session is intended to be the second public workshop in the process of preparing the 2020 update, and the only one to take place outside Europe. As such, it will be an opportunity to discuss commonalities and differences among the various international efforts towards and beyond exascale. It follows a successful BoF session at ISC 2019, titled “What are the greatest challenges to be addressed by academic HPC systems research?”, which had a full room with 50+ attendees and a lively discussion with wide participation.
This session is targeted at anyone who is interested in identifying the greatest challenges and opportunities to be addressed by 2030 through academic research in high-performance computing. While the primary focus of the Eurolab4HPC project is of course the European academic community, this BoF session will be of interest to any academic researchers and other stakeholders (representatives of funding agencies, scientific and engineering end-users, application developers and HPC suppliers) that are interested in high-quality and relevant academic HPC systems research. We expect an audience at least as large as that of the ISC 2019 BoF, so at least 50 people. Nevertheless, we would prefer not to greatly exceed 50 participants in order to encourage an open expression of ideas and interactivity.
Tentative agenda:
4 mins: Introduction
12 mins: “Speed” presentations from subject-matter experts
4 mins: Discussion
10 mins: Improving the 2017 roadmap
20 mins: Open discussion on missing and controversial aspects.
10 mins: Next steps and call for contributions.
[1] Theo Ungerer and Paul Carpenter. Eurolab-4-HPC Long-Term Vision on High-Performance Computing. August 2017. Available: http://arxiv.org/abs/1807.04521
Back to Birds of a Feather Archive Listing