About this project

Infrastructure is the glue that holds our models together. Because it plays a critical role, it is important to ensure it adheres to the CSF standards. This includes ensuring that our couplers, drivers, and I/O libraries are portable across our models, work across different languages and GPU porting methods, follow software standards, and are performant. This work seeks to explore new technologies and expand our existing packages to ensure they meet the challenges posed by our next-generation modeling capabilities, which will need to incorporate AI models/emulators and science written in other languages and run on heterogeneous hardware.

Why this work is important

We want to ensure that the infrastructure we base our science capabilities on can meet the challenges posed by the next generation of science questions that will need to be answered. We want to ensure we can incorporate AI models/emulators into our modeling systems, along with code written to run on heterogeneous hardware and across a variety of compute languages.

How does this fit within the CSF

  1. This project fits under Better Practices
  2. Increases code portability across different architectures
  3. Increase interoperability of code written in different languages
  4. Provides performance portability across different architectures
  5. Allows us to run ultra-high resolution simulations more efficiently

NSF NCAR Labs involved in this project

CISL
CGD
MMM
ACOM
HAO
RAL

External partners

NOAA
DOE

More information

Interoperability

CCPP
https://ral.ucar.edu/solutions/products/common-community-physics-package-ccpp

ESMF
https://earthsystemmodeling.org/

IO Libraries
https://www.cesm.ucar.edu/models/pio
https://docs.e3sm.org/scorpio/html/index.html
https://github.com/MPAS-Dev/SMIOL