wiki:SourceCode/Performances

ORCHIDEE performance (trunk revision 3623)

At this page the performances for ORCHIDEE in offline mode are detailed.
See here the performances for ORCHIDEE in the coupled model LMDZOR_v6 and IPSLCM6 : https://forge.ipsl.jussieu.fr/igcmg/wiki/PerformancesIPSLCM6

Orchidee scalability for different horizontal resolutions

Following table is based on tests with following set up:

  • Simulation length: 1Y
  • FG2.CWRR configuration:
    • HYDROL_CWRR = y, RIVER_ROUTING = y, STOMATE_OK_STOMATE = y, OK_EXPLICITSNOW = y, OK_FREEZE = n.
    • This is the standard set, read more here ReferenceSimulations
  • Output level: monthly and daily output
  • MPI parallelization
    • 1MPI is used for the server XIOS, the rest are used for Orchidee executable
    • no server is used for 1 MPI case or for the case with IOIPSL
  • using the standard driver dim2_driver.f90

Conclusion

For the trunk ORCHIDEE it is recommended to use XIOS with 1 server. See in this table the recommended number of processes for a given resolution for global simulations with the corresponding CPU time for 10 years of simulation. The computing time is the same at curie and ada.

Forcing file resolution Recommended number of processes Real computing time / 10 years Total computing time / 10 years
2 deg 31 MPI orchidee + 1 MPI xios server 43 m 22 h
1 deg 63 MPI orchidee + 1 MPI xios server 1 h 20m 85 h
0.5 deg 63 MPI orchidee + 1 MPI xios server 4 h 260 h

Note: The total(requested) computing time should be used in the demand for computing resources at TGCC and IDRIS.

The total cpu time is calculated by multiplying the real time with the numbers of MPI used.

Details

XIOS (using restart files)

Reference simulation

1 MPI 4 MPI 8 MPI 16 MPI 32 MPI 64 MPI 128 MPI
twodeg 1h21 29m 13m45 7m06 4m24* 3m11 3m20
onedeg Mem limit 1h40 47m03 23m 12m42 8m13* 7m05
halfdeg Not tested Mem limit 2h50 1h20 42m07 24m02* 20m36

*: Recommended number of processors

XIOS (start without restart files)

XIOS only writes the history outputs files.

1 MPI 4 MPI 8 MPI 16 MPI 32 MPI 64 MPI 128 MPI
twodeg 1h25 32m18  17m1 10m23 7m33 6m12 6m30
onedeg Not possible Mem limit 57m 33m40 23m07 18m32 17m50
halfdeg Not possible Mem limit 3h33 2h03 1h25 1h05 1h02

IOIPSL (using restart files)

1 MPI 4 MPI 8 MPI 16 MPI 32 MPI 64 MPI 128 MPI
twodeg 1h34 26m04 13m39 7m31 4m40 3m15 2m38
onedeg 5h56 1h34 52m57 28m54 17m30 12m 8m53
halfdeg >16h40 (max limit) 6h23 3h50 2h08 1h17 50m34 36m45

IOIPSL (start without restart files)

Only IOIPSL enabled.

1 MPI 4 MPI 8 MPI 16 MPI 32 MPI 64 MPI 128 MPI
twodeg 1h37 29m 16m36 10m20 7m30 5m59 5m21
onedeg 5h39 1h48 1h03 36m44 28m10 21m47 19m33
halfdeg >16h40 (max limit) 7h21 4h32 2h49 1h58 1h32 1h17

Orchidee scalability to the output level - TODO -

In this table the set up FG2.CWRR is used with 63MPI for orchidee_ol and 1MPI for the server XIOS. For the case with IOIPSL, then 64MPI are used for orchidee_ol. The first line describes different levels of output.

yearly
with XIOS
monthly
with XIOS
monthly and daily
with XIOS
daily and 3hour
with XIOS
daily and 3hour
with IOIPSL
twodeg
onedeg
halfdeg
Last modified 7 months ago Last modified on 10/12/16 10:56:08