Changeset 1764


Ignore:
Timestamp:
Aug 7, 2017, 4:38:48 PM (7 years ago)
Author:
jvatant
Message:

Fixed unitialized radius and qext in initracer

-> These arrays were printed and even tested in callsedim without being initialized !
-> Please stop handling uninitialized values !!!

Also updated README for modif from r1725

JVO

Location:
trunk
Files:
5 edited

Legend:

Unmodified
Added
Removed

  • trunk/LMDZ.GENERIC/README

    r1714 r1764  
    13481348== 07/06/2017 == MT
    13491349Resurrection of kcm1d, part I
     1350
     1351== 21/07/2017 == JVO
     1352Optimization of the optci/cv routines
     1353 - The repeated calls to huge matrices gasi/v increased dramatically the execution time because of memory access
     1354 - Added a tmpk variable
     1355 - Save ~ 50% time on the RT, ~30% on the whole code on the tested simulations

  • trunk/LMDZ.GENERIC/libf/phystd/callsedim.F

    r1477 r1764  
    106106 
    107107      do iq=1,nq
    108        if((radius(iq).gt.1.e-9).and.(iq.ne.igcm_co2_ice)) then   
     108       if((radius(iq).gt.1.e-9).and.(iq.ne.igcm_co2_ice)) then ! JVO 08/2017 : be careful radius was tested uninitialized (fixed) ...
     109       
    109110!         (no sedim for gases, and co2_ice sedim is done in condense_co2)     
    110111

  • trunk/LMDZ.GENERIC/libf/phystd/initracer.F

    r1621 r1764  
    6161       alpha_lift(:)=0.
    6262       alpha_devil(:)=0.
     63       
     64       ! Added by JVO 2017 : these arrays are handled later
     65       ! -> initialization is the least we can do, please !!!
     66       radius(:)=0.
     67       qext(:)=0.
    6368
    6469! Initialization: get tracer names from the dynamics and check if we are

  • trunk/LMDZ.TITAN/README

    r1722 r1764  
    13161316Added surfalbedo and surfemis keywords to be used with startphy_file = .false.
    13171317Also made default values in phyetat0_mod unambiguously float
     1318
     1319== 21/07/2017 == JVO
     1320Optimization of the optci/cv routines
     1321 - The repeated calls to huge matrices gasi/v increased dramatically the execution time because of memory access
     1322 - Added a tmpk variable
     1323 - Save ~ 50% time on the RT, ~30% on the whole code on the tested simulations

  • trunk/LMDZ.TITAN/libf/phytitan/initracer.F

    r1672 r1764  
    5454       alpha_lift(:)=0.
    5555       alpha_devil(:)=0.
     56       
     57       ! Added by JVO 2017 : these arrays are handled later
     58       ! -> initialization is the least we can do, please !!!
     59       radius(:)=0.
     60       qext(:)=0.
    5661
    5762! Initialization: get tracer names from the dynamics and check if we are
Note: See TracChangeset for help on using the changeset viewer.