We are currently testing a limited selection of outputs, in particular not including BGC variables.
Based on our discussions at [C]Worthy this week, we should:
establish machine-dependent benchmark netcdf output files in addition to the benchmark log files
Create a catalog of file hashes following something like machine -> test case -> output file -> hash
Compare these hashes with those produced during testing
Once this is accomplished we can establish a pt. ii to this issue concerning more concrete plans for exporting the netcdf files produced during testing in order to establish why they may be different.
We are currently testing a limited selection of outputs, in particular not including BGC variables. Based on our discussions at [C]Worthy this week, we should:
Once this is accomplished we can establish a pt. ii to this issue concerning more concrete plans for exporting the netcdf files produced during testing in order to establish why they may be different.