- Submission deadline: February 21, 2020
- Participant notification: March 4, 2020
- Workshop: April 21.-23, 2020
On April 21.-23. 2020 LRZ will organize a hands-on coding workshop to scale user applications to machine size (6000+ nodes) on SuperMUC-NG. Programming experts from LRZ, Lenovo, Intel and Allinea will support your code profiling and parallelization efforts.
SuperMUC-NG users whose application already scales up to at least two islands (~1500 nodes, ~70000 cores) are invited to apply for the workshop.
To select the most interesting projects we ask for a short proposal (max two pages) including the following information:
- A short project description including known scaling/performance issues you want to work on
(memory, communication or I/O bottlenecks, etc.).
- A short scientific motivation for achieving higher scalability.
Which application would be your vision for a next generation exaflop compute-system?
- Your goal for the workshop (intended size of scaling, performance, etc.).
- Scaling plot (strong and/or weak) covering at least two islands on SuperMUC-NG.
- Used algorithms and code dependencies (Programming language, compilers, MPI, programming model - MPI and/or OpenMP, I/O pattern, libraries, etc.)
- Contact Data (Name, Affiliation, e-mail, phone, project ID, user IDs).
Please send your proposal via email with subject "LRZ Extreme Scaling Workshop 2020" to Gerald.Mathias@lrz.de until February 21. CPU-hours used during the workshop will not be charged from your budget. Each project is limited to a single application and up to three participants. A mandatory two-page summary is due no later than two weeks after the workshop, including the list of authors, scaling plots, and profiling results (Performance, I/O bandwidth, etc.).