A GPU‐Based Ocean Dynamical Core for Routine Mesoscale‐Resolving Climate Simulations
Name
J Adv Model Earth Syst - 2025 - Silvestri - A GPU‐Based Ocean Dynamical Core for Routine Mesoscale‐Resolving Climate.pdf
Description
Published version
Size
7.89 MB
Format
Adobe PDF
Checksum (MD5)
b0240f97227b5c682fb0d9d4742d1235
Author(s)
Silvestri, Simone
Wagner, Gregory L
Constantinou, Navid C
Hill, Christopher N
Campin, Jean‐Michel
Souza, Andre N
Bishnu, Siddhartha
Churavy, Valentin
Marshall, John
Ferrari, Raffaele
Date Issued
April 21, 2025
Journal
Journal of Advances in Modeling Earth Systems
Publisher
Wiley
Citation
Silvestri, S., Wagner, G. L., Constantinou, N. C., Hill, C. N., Campin, J.-M., Souza, A. N., et al. (2025). A GPU-based ocean dynamical core for routine mesoscale-resolving climate simulations. Journal of Advances in Modeling Earth Systems, 17, e2024MS004465.
Version
Final published version
Abstract
We describe an ocean hydrostatic dynamical core implemented in Oceananigans optimized forGraphical Processing Unit (GPU) architectures. On 64 A100 GPUs, equivalent to 16 computational nodes incurrent state‐of‐the‐art supercomputers, our dynamical core can simulate a decade of near‐global oceandynamics per wall‐clock day at an 8‐km horizontal resolution; a resolution adequate to resolve the ocean'smesoscale eddy field. Such efficiency, achieved with relatively modest hardware resources, suggests thatclimate simulations on GPUs can incorporate fully eddy‐resolving ocean models. This removes a major sourceof systematic bias in current IPCC coupled model projections, the parameterization of ocean eddies, andrepresents a major advance in climate modeling. We discuss the computational strategies, focusing on GPU‐specific optimization and numerical implementation details that enable such high performance.
MIT Department
Massachusetts Institute of Technology. Department of Earth, Atmospheric, and Planetary Sciences
Terms of Use
Creative Commons Attribution-Noncommercial
Persistent DSpace Link
DOI of Published Version
https://doi.org/10.1029/2024MS004465
