The rapid changes in the multiple supercomputing architectures used to run weather and climate codes and the different programming models used seriously affect the development productivity and the ability to retain a single source code running efficiently everywhere. Domain-specific languages provide a solution to portability of these codes. In this training, we provide insights into DSLs considered in ESiWACE2 (PSyclone, CLAW and GridTools ecosystem) and demonstrate how to apply them to weather and climate models.
Topics:
- Insights into PSyclone, CLAW and GridTools ecosystem
- Demonstration into how to apply them to weather & climate models
Audience:
Participants interested in theoretically and practically learning how to use the DSL languages to implement partial differential equation operators. During a hands-on session, participants are encouraged to implement some of the benchmark models defined by WP2 using DSLs and to build their own toy models, followed by an in-depth evaluation of generated optimised implementation and performance benefits. This training is planned for 25 people.
Schedule:
This training was originally planned as a 5-day face-to-face course in 2020 and was held as an online event instead.
This is the event page of the online DSL training in November 2020. Read more about the training in the Results section.
Our Open Educational Resources (OER) material for the DSL training is available vis the OER Commons page.
Contact person: Carlos Osuna, MeteoSwiss