Changes between Version 29 and Version 30 of SourceCode/Performances/r3623


Ignore:
Timestamp:
2016-10-12T10:43:47+02:00 (8 years ago)
Author:
jgipsl
Comment:

--

Legend:

Unmodified
Added
Removed
Modified

  • SourceCode/Performances/r3623

    v29 v30  
    22= ORCHIDEE performance (trunk revision 3623) = 
    33 
    4 == Table 1: Orchidee Scalability == 
     4== Orchidee scalability for different horizontal []resolutions == 
    55Following table is based on tests with following set up:  
    66 * Simulation length: 1Y 
    … …  
    1010 * Output level: monthly and daily output 
    1111 * MPI parallelization 
    12   * while using XIOS, 64 cores simulation = 1 core MPI for the XIOS server + 63 cores MPI for Orchidee executable 
     12  * 1MPI is used for the server XIOS, the rest are used for Orchidee executable 
    1313  * no server is used for 1 MPI case or for the case with IOIPSL 
    1414 
    1515=== Conclusion === 
    1616 
    17 See in this table the recommended number of processes for a given resolution for global simulations with the corresponding CPU time for one year of simulation.  
     17'''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. 
    1818 
    19 ||= Forcing file resolution =||= Num. processes =||= Real computing time per 1 year =||= Total computing time per 1 year =|| 
    20 || 2 deg   || 32p  || 4m24  || 2h20 || 
    21 || 1 deg   || 64p  || 8m13  || 8h46 || 
    22 || 0.5 deg || 64p  || 24m02 || 25h33 || 
     19|| Forcing file resolution || Recommended number of processes || Real computing time / 10 years || Total computing time / 10 years || 
     20|| 2 deg   || 31 MPI orchidee + 1 MPI XIOS server  || 43 m     || 22 h || 
     21|| 1 deg   || 63 MPI orchidee + 1 MPI xios server  || 1 h 20m  || 85 h || 
     22|| 0.5 deg || 63 MPI orchidee + 1 MPI xios server  || 4 h      || 260 h || 
    2323 
    2424Note: The total(requested) computing time should be used in the demand for computing resources at TGCC and IDRIS.  
    … …  
    2626The total cpu time is calculated by multiplying the real time with the numbers of MPI used.  
    2727 
    28 === XIOS (using restart files) === 
     28=== Details === 
     29==== XIOS (using restart files) ==== 
    2930 
    3031Reference simulation 
    … …  
    3738*: Recommended number of processors 
    3839 
    39 === XIOS (start without restart files) === 
     40==== XIOS (start without restart files) ==== 
    4041 
    4142XIOS only writes the history outputs files. 
    … …  
    4647||halfdeg    || Not possible || Mem limit || 3h33 || 2h03 || 1h25 || 1h05 || 1h02 || 
    4748 
    48 === IOIPSL (using restart files) === 
     49==== IOIPSL (using restart files) ==== 
    4950 
    5051||           || 1 MPI || 4 MPI || 8 MPI || 16 MPI || 32 MPI || 64 MPI || 128 MPI || 
    … …  
    5354||halfdeg    || >16h40 (max limit) || 6h23 || 3h50 || 2h08 || 1h17 || 50m34 || 36m45 || 
    5455 
    55 === IOIPSL (start without restart files) === 
     56==== IOIPSL (start without restart files) ==== 
    5657 
    5758Only IOIPSL enabled. 
    … …  
    6465 
    6566 
    66 == Table 2: Orchidee XIOS scalability (IO) - TODO -  == 
     67== Orchidee scalability to the output level - TODO -  == 
    6768In 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. 
    6869The first line describes different levels of output.