TRUSTING ORCHIDEE trunk at irene-amd/TGCC

Based on configuration IPSLCM7_work

UNDER CONSTRUCTION....

Started in october 2025 at irene-amd, continuously running

This is a program testing the different revisions on the trunk of ORCHIDEE. Different compile options and run modes are tested offline and coupled to LMDZ. Restart files and diagnostics are compared to a reference run. Intercomparison of some runs are done.
Each line in the grey table corresponds to a specific installation of the model and all related tests. The revision number of ORCHIDEE is set in column 2 and the revsion numbers for the other compontents and tools are listed in the end of the line. The fifth column shows the reference revision which is used for comparing.
When the Status field in the third column is green and says OK it means that all tests finished well and there were no changes compared to the reference run. When the status field says DIFFER it means that all tests finished but some of the results changed. When the Status field becomes red saying FAILED their might be several reasons, read the comments in fourth column to know more. There might be some runs that are not counted for in the Status field. This will be listed below.

See here the discription of the different test cases performed at each execution of the trusting. Default set up is used if nothing else is said.

Offline =>

JobName

Compilation

Experiment Type

PeriodLenght

Total lenght

Other

LMDZOR =>

JobName

Compilation

Experiment Type

Parallelization

PeriodLenght

Total lenght

Other

O1p

prod

FG_CRUJRA_OD_HIST

1Y

2Y

LO1p

prod

clim_pdControl.orch4

mpi_omp

1D

2D

O2p

prod

FG_CRUJRA_OD_HIST

1M

2Y

LO2p

prod

clim_pdControl.orch4

mpi_omp

2D

2D

O3p

prod

FG_CRUJRA_OD_HIST

1Y

2Y

Using different nubmer of cores MPI

LO3p

prod

clim_pdControl.orch4

mpi_omp

1D

2D

Using different number of cores MPI

O4p

prod

FG_CRUJRA_HIST

1Y

2Y

LO4p

prod

clim_pdControl.orch4

mpi

1D

2D

O5d

debug

FG_CRUJRA_OD_HIST

1D

4D

LO5d

debug

clim_pdControl.orch4

mpi_omp

1D

2D

O6d

debug

FG_CRUJRA_OD_HIST

2D

4D

LO6d

debug

clim_pdControl.orch4

mpi_omp

1D

2D

Using different number of cores MPI

O7d

debug

FG_CRUJRA_OD_HIST

2D

4D

Using IOIPSL output. Deactivate routing scheme.

LO7d

debug

clim_pdControl.orch4

mpi_omp

1D

2D

Using interpolation with XIOS

O8d

debug

FG_CRUJRA_OD_HIST

2D

4D

Using interpolation with XIOS

O9d

debug

FG_CRUJRA_OD_HIST

2D

4D

OK_SOIL_CARBON_DISCRETIZATION=y

O10d

debug

FG_CRUJRA_HIST

1D

4D

ICOLMDZOR =>

JobName

Compilation

Experiment Type

Parallelization

PeriodLenght

Total lenght

Other

O11d

debug

FG_CRUJRA_HIST

2D

4D

DLO1p

prod

clim_pdControl.orch4

mpi_omp

1D

2D

O12d

debug

FR_SAFRAN_HIST

2D

4D

DLO2p

prod

clim_pdControl.orch4

mpi_omp

2D

2D

O13d

debug

FG_CRUJRA_SPIN

1Y

2Y

1 pixel from global forcing file

DLO3d

debug

clim_pdControl.orch4

mpi_omp

1D

2D

O14p

prod

FL_IT-MBo_PD_SPIN

1Y

2Y

Exceptions where the results are accepted

1- For cross test O1p and O3p corresponding to the change in number of MPI in offline mode, 5 errors on the restart variables are accepted. These errors corresponds to differences in the routing variables and it has been evaluated as accepteable, read more here: https://forge.ipsl.fr/orchidee/ticket/1046 .
2- For cross test LO1p and LO4p corresponding to the change in MPI and OMP in coupled mode gives different results for all variables. This is due to the optimisation level choosen for LMDZ. The results have been examined and accepted by the LMDZ team. This is the case at irene-amd and irene-skl.