ExCALIBUR Task Parallelism Workshop (ICCS)
Future supercomputing architectures are expected to exhibit unprecedented hardware parallelism. This parallelism has to be “harvested” on the software side. Traditional, global, loop-based parallelism and notably the bulk-synchronous/BSP paradigm are insufficient, and developers might have to go back to the drawing board and uncover alternatives to equip their code with higher concurrency. Task-based parallelism is one tool to write such new, highly concurrent software. It phrases programs as sequence of steps including their causal dependencies, but leaves the decision what (aka which task), when and where to execute a task to a task runtime.
The crosscutting ExCALIBUR project on Exposing Parallelism – Task Parallelism aims to develop a data-driven taskification toolset and workflow, to investigate task features that are missing within mainstream tasking approaches, and to establish taskification training and an experience exchange landscape.
This Workshop
This workshop aims to bring researchers, research software engineers and vendors together to discuss the current state of the art, issues, experiences and ideas for tasking in research software.
The workshop takes place on June 21, 2022, at Brunel University, London, UK, and is co-located with the ICCS – International Conference on Computational Science 2022. It can be attended either as part of the conference or individually.
Registration: https://www.iccs-meeting.org/iccs2022/co-located-events/ (tab at the bottom of the page)
For any questions, please contact Marion Weinzierl.
Workshop Agenda
(Sessions 6G and 7G on the ICCS agenda.)
| 14.50 | Welcome | Tobias Weinzierl |
| 14.55 | Talk: Overview of the Tasking Parallelism project | Cristian Barrera-Hinojosa |
| 15.15 | Software demo of taskification simulator API | Adam Tuft |
| 15.30 | Questions and discussion | |
| 15.40 | Talk: Tasks in the OpenMP API | Michael Klemm (AMD) |
| 16.00 | Talk: Task based programming in OneAPI: C++, SYCL, TBB, OpenMP | Andrew Mallinson (Intel) |
| 16.20 | Questions and discussion | |
| 16.30 | Break | |
| 17.00 | Talk: Accelerating applications using ISO standard C++ | Gonzalo Brito Gadeschi (NVIDIA) |
| 17.20 | Questions and discussion | |
| 17.25 | Group discussions: issues with tasks from the participants own projects, what would we need/expect from a taskification tool, lessons learned, ideas | |
| 18.10 | Report-back from group discussions | |
| 18.30 | Project Outlook, Training and Workshops | Tobias Weinzierl |
| 18.40 | End |

- I've finally updated my project pages describing our long-standing and fruitful collaboration with Intel at @durham.ac.uk : https://tobiasweinzierl.webspace.durham.ac.uk/intel-academic-centre-of-excellence/
- We will feature our HAI-End and @shareing.bsky.social projects in collaboration with @durham-comp-sci.bsky.social and @arc-durhamuni.bsky.social at the conference as well. Thanks to @cake-dri.bsky.social for making this possible. [contains quote post or other embedded content]
- This is a great contribution highlighting the importance of this event. If you find it interesting, the @shareing.bsky.social team has also a blog on it: https://shareing-dri.github.io/blogpost/dri-retreat-26/ [contains quote post or other embedded content]
- Had a brilliant time down in Abingdon to learn more about the National Federated Compute Services NetworkPlus initiative: nfcs-networkplus.ac.uk Looking forward to read about their roadmap later this year.
- Our @shareing.bsky.social project has a new call open (shareing-dri.github.io/task-map/). We are searching for projects: You propose a project and whatever you want to do around accelerate computing, as long as you meet at least one of the tasks/outcomes that the SHAREing team has identified. https://shareing-dri.github.io/task-map/