This topic is to announce new releases of benchcab
. benchcab
is a Python-based software for the scientific evaluation of CABLE on NCI.
To receive email notifications when a new release is announced, please watch this topic.
Release: 2.0.0
This is our first fully public release of this software. Please do not hesitate to give us feedback on this forum or GitHub so benchcab
can become more useful to you.
Get started
Refer to benchcab
’s User guide.
Features:
run several CABLE versions at various flux tower sites with the same compilation* and a standard set of science configurations.
choice of sets of flux tower sites to allow different levels of testing
bitwise comparison of CABLE outputs
CABLE outputs from
benchcab
is compatible with experiments on modelevaluation.org (workspacebenchcab-evaluation
)compatible with CABLE v3 and v2 (limited testing of v2).
allow running tests of CABLE’s branches owned by others
versioned documentation, including how to use modelevaluation.org with
benchcab
outputs.example work directory
allow running a customised set of science configurations on one branch (e.g. test new feature) or all branches.
*: the compilation may differ between the branches if the CABLE source codes are not both based on CABLE v3.
Limitations:
no support for gridded (spatial) simulations
limited standard set of science configurations. Feedback on the choice of configurations is welcomed. Please see this forum topic to leave some feedback.
no testing with any spin-up. But starting the simulations from a given restart file is possible and easy to do.
no testing with CASA
analysis plots will be updated to allow easier comparison between science configurations and include more variables.
limited to use on NCI’s supercomputer
Since
benchcab
is still in active development, future versions might not be backward-compatible