August 25, 2025Open Access

Operational numerical weather prediction with ICON on GPUs (version 2024.10)

View Full Paper

Ask AI

Bookmark

A 5.6x speed-up was achieved in the ICON model after its adaptation for GPUs, enhancing performance significantly.
The transition to GPU required a full-port strategy involving multiple aspects from data assimilation to physical parameterizations.
Performance tuning and mixed-precision optimization were critical for meeting the operational forecasting needs of MeteoSwiss.
These advancements suggest broader implications for transitioning community weather models to high-performance computing platforms.

Discussion

No takes yet. Share an insight, caveat, or question.

Implication

The porting of the ICON model to GPUs improves numerical weather prediction efficiency, suggesting significant operational advancements for MeteoSwiss.

Cite This Study

Lapillonne et al. (2025) studied this question.

synapsesocial.com/papers/68af5d6fad7bf08b1eae1151 https://doi.org/10.5194/egusphere-2025-3585

View Full Paper

Ask AI

Bookmark

Implication

The porting of the ICON model to GPUs improves numerical weather prediction efficiency, suggesting significant operational advancements for MeteoSwiss.

Operational numerical weather prediction with ICON on GPUs (version 2024.10)

Key Points

Discussion

Implication

Cite This Study

Operational numerical weather prediction with ICON on GPUs (version 2024.10)

Key Points

Discussion

Implication

Cite This Study

Also Consider

Also Consider