Getting Started with GCHP on Various Platforms#
For general GCHP questions, please refer to the GCHP official documentation.
As GCHP versions evolve, dependent libraries and environment settings may change. This documentation primarily covers GCHP v14.0.0 to v14.6.x.
The GCHP operational run script examples provide useful reference examples.
Compute1 (Docker-Based)#
To run GCHP on Compute1 across multiple nodes using MPI, specify the appropriate pre-built Intel compiler Docker image in your LSF script according to your GCHP version requirements.
1dandan/gchp-esmf8.0:latestAn environment with ESMF v8.0.0 to support compiling/running GCHP v13 simulations based on Intel compilers.
1dandan/gchp-esmf8.3:latestAn environment with ESMF v8.3.1 to support GCHP v14 (< 14.7) simulations based on Intel compilers.
yuanjian.z/gchp_env:latestAn environment with ESMF v8.6.1 to support GCHP v14 (>= 14.7) simulations based on Intel compilers.
Important: Starting with GCHP v14.3.0, the introduction of Cloud-J as a submodule requires manually setting stack memory to unlimited in a separate MPI-launched bash script. For details, see the related GCHP pull request.
Compute2 (Module/Spack-Based)#
To request access to Compute2, contact Bill Zhuge at WashU.
On Compute2, load the required environment using either module or Spack instead of pre-built Docker images. Refer to the GCHP operational run script examples for specific commands and configurations.
AWS (Module/Spack-Based)#
To get started with AWS, follow our AWS Quick Start Guide.
Consult AWSGuide by Laura Yang for reference for picking up shared cluster maintained Yuanjian Zhang.
Also consult the GCHP operational run script examples for platform-specific setup instructions.