Changes between Version 105 and Version 106 of DevelopmentActivities/ORCHIDEE-DOFOCO

Timestamp:

2016-11-11T19:15:28+01:00 (7 years ago)

Author:

luyssaert

Comment:

--

DevelopmentActivities/ORCHIDEE-DOFOCO

@@ -12 +12 @@
 === 11.11.2016 ===
 - Checked out revision 3879. The multi-layer set-up worked ran for 20 years for 37 PFTs for a single PFT (4 age classes for forests). Note that only PFT_10 contained veget_max in that set-up. A shorter test (1 months at the daily time step) was passed with all 37 PFT being activated.
-- [NOT solved] Generating daily output took 15 seconds per day or 18 minutes per month. Generating monthly output takes 280 seconds or 80 minutes per month. Annual output takes 63 minutes per year. Part of the timing is due to the size of the *_out_orchidee files which is currently 6GB per year! Setting stomate_printlev=5 already saves some time. 1D: 10 seconds per day and 10 minutes per month. 1Y: XX minutes per year. *_out_orchidee is now YGB per year.
-
-All write statements should be placed within (sub)routine-specific printlev flags (see stomate) such that the *_out_orchidee file can be controlled.
+- [NOT solved] Generating daily output took 15 seconds per day or 18 minutes per month. Generating monthly output takes 280 seconds or 80 minutes per month. Annual output takes 63 minutes per year. This awful timing is partly due to the size of the *_out_orchidee files which is currently 6GB per year! Setting stomate_printlev=5 saved some time for the daily simulations (1D: 10 seconds per day and 10 minutes per month) but did not result in any time savings for the annual simulations. 1Y still took 60 minutes per year. *_out_orchidee is still 6GB per year. All write statements should be placed within (sub)routine-specific printlev flags (see stomate_*.f90) such that the *_out_orchidee file can be controlled. The problem seems to occur in the modules of both stomate and sechiba.
 - [NOT solved] Revision 3879 does only seem to work when NLEVELS_PHOTO = 11. This is a not the intended behaviour.
-- [NOT solved] Revision 3879 should be tested with CWRR because CWRR is required to calculate water stress with sechiba_hydraulic_architecture.f90.
+- [NOT solved] Revision 3879 should be tested with CWRR because CWRR is required to calculate water stress with sechiba_hydraulic_architecture.f90. To be tested.