Preprints
https://doi.org/10.5194/egusphere-2024-2879
https://doi.org/10.5194/egusphere-2024-2879
09 Oct 2024
 | 09 Oct 2024
Status: this preprint is open for discussion.

Porting the Meso-NH atmospheric model on different GPU architectures for the next generation of supercomputers (version MESONH-v55-OpenACC)

Juan Escobar, Philippe Wautelet, Joris Pianezze, Florian Pantillon, Thibaut Dauhut, Christelle Barthe, and Jean-Pierre Chaboureau

Abstract. The advent of heterogeneous supercomputers with multi-core central processing units (CPUs) and graphics processing units (GPUs) requires geoscientific codes to be adapted to these new architectures. Here we describe the porting of the Meso-NH version 5.5 community weather research code to GPUs named MESONH-v55-OpenACC, with guaranteed bit reproducibility thanks to its own MPPDB_CHECK library. This porting includes the use of OpenACC directives, specific memory management, communications optimization, development of a geometric multigrid solver and creation of an in-house preprocessor. Performance on AMD MI250X GPU Adastra platform shows up to 6.0× speedup (4.6x on NVIDIA A100 Leonardo platform), and achieves a gain of a factor 2.3 in energy efficiency compared to AMD Genoa CPU Adastra platform, using the same configuration with 64 nodes. The code is even 17.8 faster by halving the precision and quadrupling the nodes with a gain in energy efficiency of a factor 1.3. First scientific simulations of three representative storms using 128 GPUs nodes of Adastra show successful cascade of scales for horizontal grid spacing down to 100 m and grid size up to 2.1 billion points. For one of these storms, Meso-NH is also successfully coupled to the WAVEWATCH III wave model via the OASIS3-MCT coupler without any extra computational cost. This GPU porting paves the way for Meso-NH to be used on future European exascale machines.

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.
Juan Escobar, Philippe Wautelet, Joris Pianezze, Florian Pantillon, Thibaut Dauhut, Christelle Barthe, and Jean-Pierre Chaboureau

Status: open (until 04 Dec 2024)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse
Juan Escobar, Philippe Wautelet, Joris Pianezze, Florian Pantillon, Thibaut Dauhut, Christelle Barthe, and Jean-Pierre Chaboureau

Data sets

Code and data for Porting the Meso-NH atmospheric model on different GPU architectures for the next generation of supercomputers (version MESONH-v55-OpenACC) Juan Escobar, Phillippe Wautelet, Joris Pianezze, Thibaut Dauhut, Christelle Barthe, Florian Pantillon, and Jean-Pierre Chaboureau https://zenodo.org/doi/10.5281/zenodo.13759713

Model code and software

Meso-NH version MESONH-v55-OpenACC The Meso-NH developers http://mesonh.aero.obs-mip.fr/gitweb/?p=MNH-git_open_source-lfs.git;a=commit;h=498cd0cb968041038ff6c5b0f2a76d5066c55bfd

Juan Escobar, Philippe Wautelet, Joris Pianezze, Florian Pantillon, Thibaut Dauhut, Christelle Barthe, and Jean-Pierre Chaboureau

Viewed

Total article views: 236 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
150 74 12 236 4 5
  • HTML: 150
  • PDF: 74
  • XML: 12
  • Total: 236
  • BibTeX: 4
  • EndNote: 5
Views and downloads (calculated since 09 Oct 2024)
Cumulative views and downloads (calculated since 09 Oct 2024)

Viewed (geographical distribution)

Total article views: 226 (including HTML, PDF, and XML) Thereof 226 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
1
 
 
 
 
Latest update: 08 Nov 2024
Download
Short summary
The Meso-NH weather research code is adapted for GPUs using OpenACC, leading to significant performance and energy efficiency improvements. Called MESONH-v55-OpenACC, it includes enhanced memory management, communication optimizations, and a new solver. On the AMD MI250X Adastra platform, it achieved up to 6× speedup and 2.3× energy efficiency gain compared to CPUs. Storm simulations at 100 m resolution show successful results , positioning the code for future use on exascale supercomputers.