The workshop will be held in Edinburgh, Scotland on Tuesday, September 2, 2025 in Room Salisbury.
Note: All presenters and attendees must register at the IEEE Cluster website.
Agenda: (all times are in British Summer Time / UTC+1)
09:30 - 09:45 – Welcome Message & Speed Introduction
09:45 - 10:45 – Keynote: I/O Simulation: From Resources to Complex Workloads
Speaker: Frédéric Suter, Oak Ridge National Laboratory (USA)
10:45 - 11:00 – TBD
11:00 - 11:30 – Refreshment Break
11:30 - 12:15 – Expert Talk I: Data centric workflows: The Maestro Middleware in the Destination Earth Twin Engine, Utz-Uwe Haus, HPE (Switzerland).
12:15 - 13:00 – Expert Talk II: Using objects for data storage in high-performance I/O, Adrian Jackson, EPCC, The University of Edinburgh (UK).
13:00 - 14:00 – Lunch Break
14:00 - 14:30 – Paper Talk I: BBView: A View-Aware Burst-Buffer Mechanism for MPI-IO, Sohei Koyama and Osamu Tatebe. University of Tsukuba (Japan).
14:30 - 15:00 – Paper Talk II: Benchmarking Darshan and Recorder for HPC I/O Profiling and Tracing, Zhaobin Zhu, Leonie Derstroff and Sarah Neuwirth. Johannes Gutenberg University Mainz (Germany).
15:00 - 15:30 – Paper Talk III: Towards an Optimal IO500 Configuration: Literature Meets Empirical Evaluation, Hadi Ahmad, Radita Liem and Jay Lofstead. RWTH Aachen University (Germany) and Sandia National Laboratory (USA).
15:30 - 16:00 – Refreshment Break
16:00 - 16:30 – Paper Talk IV: Optimizing I/O for an Exascale Implicit Kinetic Plasma Simulation using the Rabbit Storage System, Ian Lumsden, Hariharan Devarajan, Izzet Yildirim, Stefano Markidis, Andong Hu, Ivy Peng, Luca Pennati, Dewi Yokelson, Stephanie Brink, Olga Pearce, Tom Scogland, Bronis R. de Supinski, Gian Luca Delzanno, Anthony Kougkas, Xian-He Sun and Michela Taufer. University of Tennessee, Knoxville (USA), Lawrence Livermore National Laboratory (USA), Illinois Institute of Technology (USA), KTH Royal Institute of Technology (Sweden), Advanced Micro Devices, Inc. (USA) and Los Alamos National Laboratory (USA).
16:30 - 17:15 – Expert Panel: The I/O Horizon: Workloads, Storage, and What Comes Next
Panelists: Frédéric Suter, Utz-Uwe Haus, Adrian Jackson
Moderator: Jay Lofstead
17:15 - 17:30 – Closing Remarks
17:30 – End of REX-IO Workshop Day
Keynote Abstract
Simulation is a popular way to conduct performance evaluation studies as it makes it possible to explore hypothetical scenarios, can yield reproducible results, and usually require less time, labor, and/or carbon footprint to obtain results. In this talk, I will present how I/Os can be simulated from individual storage resources to complex hybrid HPC workloads with the SimGrid simulation toolkit and its rich software ecosystem.
Speaker Bio: Frédéric Suter (Oak Ridge National Laboratory, USA)
Fred Suter is a Senior Research Scientist in the Workflow Systems Group of the Computer Science and Mathematics Division at Oak Ridge National Laboratory since February 2022. Before that, he has been a CNRS researcher at the IN2P3 Computing Center in Lyon, France, for thirteen years where he led the research team of the center. His research interests include scheduling, scientific workflows, data science, and cyberinfrastructure and scientific application simulation. Dr. Suter obtained his Ph.D. from the Ecole Normale Supérieure de Lyon, France, in 2002 and his Habilitation to supervise research from the Ecole Normale Supérieure de Lyon in 2014. He is also one of the main developers of the SimGrid toolkit.
The high-resolution, multi-application workflow of the Climate Digital Twin of Destination Earth brings its share of HPC challenges; larger datasets that come with greater resolution, and the conception of a common interface to allow convenient orchestration and data transfers to ease integration of downstream applications. I n this talk we present the contributions to the Digital Twin Engine made so far with the Maestro middleware, and the plan to integrate with the pre-production Climate DT workflow within the next year.
Among the contributions, we have shown the advantage of direct data streaming from IFS-NEMO to a Climate DT downstream application, using the high-speed network via Maestro. We also developed a Maestro-enabled component, the Librarian, to read data from FDB and make it available via RDMA to applications. All this is part of a more generally applicable infrastructure that also allows integration with workflow management side (via a Data Notifier component), and starts to have improved fault tolerance.
(joint work with Christopher Haine and Ali Mohammed, both HPE)
Speaker Bio:
Dr. rer. nat. Utz-Uwe Haus, Head of the HPE HPC/AI EMEA Research Lab (ERL), studied mathematics and computer science at the Technical University of Berlin (TU Berlin). After obtaining a doctorate in mathematics at the University of Magdeburg (Germany) he worked on nonstandard applications of mathematical optimization in chemical engineering, material science, and systems biology. He led a junior research group at the Magdeburg Center for Systems Biology and wassenior researcher at the Department of Mathematics of ETH Zürich. H e co-founded CERL, the CRAY EMEA Research Lab in 2015 , the predecessor of ERL.
His research interests focus on parallel programming and data aware scheduling problems, data analytics in the context of semanticdatabases, and novel compute architectures, as well as sustainable and secure computing.
Speaker Bio: Prof. Adrian Jackson is the Director of Research & Professor of High Performance Computing Technologies at EPCC - The University of Edinburgh. His research spans high performance computing, data and storage, memory technologies, and application optimisation, helping to ensure large scale computing is efficiently and effectively used. Current and past application domains include computational fluid dynamics, plasma physics, nuclear fusion, molecular dynamics, and discrete element modelling.