Can an LLM port a complex Fortran ocean model to C++/Kokkos and make it run on GPUs?
Yes.
We did it with FESOM2: production runs, ~3–4× speedup vs CPU, same code on LUMI, MareNostrum5 and JUPITER.
Fortran as executable specification?
Maybe yes :)
arxiv.org/abs/2606.11356
Large language models (LLMs) can translate and modify source code, and have been shown to do so for codes of different complexity. Whether they can port a complete, production geophysical model to a d...