Sandia National Laboratory, USA
Title: Turbulent Combustion Simulation and In Situ Analytics on Titan and Summit with S3D-Legion
Abstract: Turbulent combustion direct numerical simulation (DNS) is a first principles tool for understanding the combustion of complex fuels over a broad range of aerothermochemical conditions representative of practical engines for transportation and power generation. These simulations require the world’s leading supercomputers to numerically resolve the large dynamic range of flow and chemical scales and to perform in situ analytics side-by-side with the DNS. To facilitate performance portability and composable in situ workflows, the DNS code S3D-Legion, a re-implementation of S3D-MPI in a dynamic task-based asynchronous programming model will be described. Weak and strong scaling performance on Summit and Titan at ORNL and PizDaint at ETH will be presented.
Biography:
Jacqueline H. Chen is a Distinguished Member of Technical Staff at the Combustion Research Facility at Sandia National Laboratories. She has contributed broadly to research in direct numerical simulations of turbulent combustion elucidating turbulence-chemistry interactions in turbulent flames and ignition processes. These interactions govern the overall combustion rate, emissions, the degree of local extinction and ignition timing. She and her collaborators have discovered new physical insights related to turbulent premixed and stratified flame propagation, preferential diffusion, intrinsic flame instabilities, lifted flame stabilization in (non)heated flows, reactive scalar mixing, compression ignition and flashback in boundary layers. These benchmark simulation data have also been used by the modeling community to validate turbulent combustion models.
To achieve scalable performance of DNS on increasingly complex heterogeneous computer architectures, she leads an interdisciplinary team of computer scientists, applied mathematicians and computational scientists to develop combustion direct numerical simulation software with complex chemistry and multi-physics for exascale supercomputers. She is a member of the National Academy of Engineering and a Fellow of the Combustion Institute and the Americal Physical Society. She received the Combustion Institute’s Bernard Lewis Gold Medal Award in 2018 and the Society of Women Engineers Achievement Award in 2018. She is a Director of the Combustion Institute, a former Editor of Flow, Turbulence and Combustion, Co-Editor of Proceedings of the Combustion Institute, Volumes 29 and 30, 2008, 2010 and has served on numerous editorial boards and advisory committees.
Molecular Biomechanics Interdisciplinary Center for Scientific Computing (IWR) Heidelberg University, Germany and Heidelberg Institute for Theoretical Studies, Germany
Title: HPC for biomaterials: why playing soccer hurts
Abstract: Materials - be it a shoe sole or an achilles tendon - respond to mechanical stress involving length scales all the way down to atoms and electrons, rendering computational materials science a prime application area of HPC. I will show how large-scale simulations of collagen have fundamentally changed our understanding how tension in this biomaterial, in our achilles tendon or muscle, can increase pain or muscle sour. I will give an outlook on how exascale can go beyond these achievements. The vision is to watch microcracking in materials all the way to failure, at relevant dimensions and yet at atomistic detail.
Biography: Frauke Gräter is head of the research group “Molecular Biomechanics” at the Heidelberg Institute for Theoretical Studies (HITS) and Professor at Heidelberg University. She investigates how proteins have been designed to specifically respond to mechanical forces in the cellular environment or as a biomaterial, e.g. in the process of blood coagulation, in tendons, or in spider silk fibers. To this end, her group uses and further develops various simulation techniques from the molecular to the mesoscopic scale. Frauke Gräter is member of the scientific steering committee of PRACE, and has been awarded the Ada Lovelace award for HPC in 2017.
Argonne National Laboratory, USA
Title: Exascale Computing, Artificial Intelligence and Cancer
Abstract: In this talk I'll give an update on the US Exascale Computing Initiative (ECI). The ECI is a national plan to develop and field multiple Exascale computing systems starting in 2021. In addition to developing systems the ECI is supporting the development of a broad collection of Exascale applications and software environments to support the mission of the Department of Energy (DOE) and the broad needs of the US research community. Artificial Intelligence (AI) based applications are emerging as important new drivers of advanced computing capabilities across many scientific disciplines. I'll discuss key "AI in Science" opportunities that will be enabled by Exascale systems and walk through some of the architectural, software and methods challenges. Finally, I will present an overview of the joint program of the DOE and the National Cancer Institute to apply Exascale computing to three Cancer research challenge problems (i.e. drug response prediction, RAS biology and understanding patient trajectories).
Biography: Professor Rick Stevens is internationally known for work in high-performance computing, collaboration and visualization technology, and for building computational tools and web infrastructures to support large-scale genome and metagenome analysis for basic science and infectious disease research. He is the principle investigator for the NIH-NIAID funded PATRIC Bioinformatics Resource Center which is developing comparative analysis tools for infectious disease research, and for the Exascale Computing Project (ECP) Exascale Deep Learning and Simulation Enabled Precision Medicine for Cancer project which focuses on building a scalable deep neural network code called the CANcer Distributed Learning Environment (CANDLE) to address three top challenges of the National Cancer Institute. Stevens has been a professor at the University of Chicago since 1999, and Associate Laboratory Director at Argonne National Laboratory since 2004. Over the past twenty years, he and his colleagues have developed the SEED, RAST, MG-RAST and ModelSEED genome analysis and bacterial modeling servers that have been used by tens of thousands of users to annotate and analyze more than 250,000 microbial genomes and metagenomic samples. He teaches and supervises students in the areas of computer systems and computational biology, and he co-leads the DOE national laboratory group that has been developing the national initiative for exascale computing.
National University of Defense Technology, China
Title: Tianhe-3 and the Exascale Road in China
Abstract: After a brief introduction to the Tianhe-1 and Tianhe-2 supercomputers, this talk focuses on the progress of Tianhe-3 exascale system, including the prototype system development, the new chips, and the new interconnection network. Major issues on exascale system development and the Tianhe application ecosystem will be discussed. The perspective of Exascale systems development in China, especially the projects under the national key R&D program are discussed toward the end of the talk.
Biography: Ruibo Wang is an associate professor at National University of Defense Technology. He is the director of Operating System Lab and now holds an appointment as the leading designer of Tianhe System Software. He has been working on operating system and fault tolerant management for Tianhe-1 and Tianhe-2 supercomputers. His current research interests include high performance computer architecture and implementation technologies, transactional memory, and kernel performance measurement. He received a Bachelor of Engineering (2003) and a Ph.D (2011) in Computer Science and Technology from the National University of Defense Technology.
European Processor Initiative, EU
Title: EPI, Europe strikes back on HPC
Abstract: European Processor Initiative (EPI) represents the ambition of Europe in HPC field. This joint project backed by EU H2020 framework contains 23 project members from industrial and scientific segments. It aims to develop critical technologies and generate key components for next HPC systems: a high performance -low energy consumption processor. In this talk we will disclose the project outlook and the vision of the team.
Biography: Lead Architect for the EPI project. He acted as adv. technology officer in the consumer digital division of STMicroelectronics France during 2015-2018. Before he was the head of engineering teams for settopbox chip product line in MediaTek-MStar, and CTO of Sunplus Technology in Taiwan.
Arm
Title: Arm in HPC
Abstract: Arm-based hardware and software products are recent entrants into HPC. The goal is to bring architectural choice back to HPC with Arm-based servers. Already these server-class designs are starting to appear and HPC performance is available. On the software side, Arm is working closely with the growing ecosystem to ensure a mature and robust stack. In this talk, we will discuss the progression of Arm in HPC and macro-trends motivating a shift in architecture for Cloud and Data Center.
Biography: Brent Gorda is the Sr. Director for worldwide HPC business at Arm. After founding Whamcloud to focus on Lustre, he sold the company to Intel where he served as the General Manager for the High Performance Data Division.