Main menu

Navigation

load balancing of I1850CLM45CN

1 post / 0 new
ncepulee@...
load balancing of I1850CLM45CN

Hi,all.Recently i have been trying to run CLM4.5 at a fine resolution of my own surfdata(360 gride cells * 450 grid cells) and want to get a most effecective pe layout before taking a long run.But, i don't konw the rules to tune the pe layout.Then I post the ccsm_timing file  to turn for help. From the file , it seems like the CPL COMM is the most limited fator.

COMPSET:I1850CLM45CN

NETCDF:pnetcdf

compute nodes:24 cores/per node * 150

---------------- CCSM TIMING PROFILE ---------------------

  Case        : CN-640x360_china_I1850CLM45CN
  LID         : 170405-171740
  Machine     : cern
  Caseroot    : /home/cern1426/cesm1.2.2/scripts/CN-640x360_china_I1850CLM45CN
  Timeroot    : /home/cern1426/cesm1.2.2/scripts/CN-640x360_china_I1850CLM45CN/Tools
  CCSM User   : cern1426
  CCSM Tag    : cesm1_2_2  (best guess)
  Curr Date   : Wed Apr 5 18:01:02 CST 2017

  grid        : a%CLM_USRDAT_l%CLM_USRDAT_oi%CLM_USRDAT_r%null_m%reg_g%null_w%null
  compset     : 1850_DATM%QIA_CLM45%CN_SICE_SOCN_RTM_SGLC_SWAV
  run_type    : startup, continue_run = FALSE (inittype = TRUE)
  stop_option : nyears, stop_n = 1
  run_length  : 365 days (364 for ocean)



  component       comp_pes    root_pe   tasks  x threads instances (stride)
  ---------        ------     -------   ------   ------  ---------  ------
  cpl = cpl        3600        0        3600   x 1       1      (1     )
  glc = sglc       1           0        1      x 1       1      (1     )
  wav = swav       1           0        1      x 1       1      (1     )
  lnd = clm        3600        0        3600   x 1       1      (1     )
  rof = rtm        1           0        1      x 1       1      (1     )
  ice = sice       1           0        1      x 1       1      (1     )
  atm = datm       1200        0        1200   x 1       1      (1     )
  ocn = socn       1           0        1      x 1       1      (1     )

  total pes active           : 3600
  pes per node               : 24
  pe count for cost estimate : 3600

  Overall Metrics:
    Model Cost:            2438.61   pe-hrs/simulated_year
    Model Throughput:        35.43   simulated_years/day

    Init Time   :      88.508 seconds
    Run Time    :    2438.608 seconds        6.681 seconds/day
    Final Time  :       0.073 seconds

    Actual Ocn Init Wait Time     :       0.000 seconds
    Estimated Ocn Init Run Time   :       0.000 seconds
    Estimated Run Time Correction :       0.000 seconds
      (This correction has been applied to the ocean and total run times)

Runs Time in total seconds, seconds/model-day, and model-years/wall-day
CPL Run Time represents time in CPL pes alone, not including time associated with data exchange with other components

    TOT Run Time:    2438.608 seconds        6.681 seconds/mday        35.43 myears/wday
    LND Run Time:    1025.150 seconds        2.809 seconds/mday        84.28 myears/wday
    ROF Run Time:       0.000 seconds        0.000 seconds/mday         0.00 myears/wday
    ICE Run Time:       0.000 seconds        0.000 seconds/mday         0.00 myears/wday
    ATM Run Time:     655.780 seconds        1.797 seconds/mday       131.75 myears/wday
    OCN Run Time:       0.000 seconds        0.000 seconds/mday         0.00 myears/wday
    GLC Run Time:       0.000 seconds        0.000 seconds/mday         0.00 myears/wday
    WAV Run Time:       0.000 seconds        0.000 seconds/mday         0.00 myears/wday
    CPL Run Time:       8.728 seconds        0.024 seconds/mday      9899.18 myears/wday
     CPL COMM Time:   1769.940 seconds        4.849 seconds/mday        48.82 myears/wday

---------------- DRIVER TIMING FLOWCHART ---------------------

   NOTE: min:max driver timers (seconds/day):
                            CPL (pes 0 to 3599)
                                                OCN (pes 0 to 0)
                                                LND (pes 0 to 3599)
                                                ROF (pes 0 to 0)
                                                ICE (pes 0 to 0)
                                                ATM (pes 0 to 1199)
                                                GLC (pes 0 to 0)
                                                WAV (pes 0 to 0)

  DRIVER_CLOCK_ADVANCE        0.004:   0.005
  DRIVER_LNDPREP              0.002:   0.012
  DRIVER_C2L                     <---->           0.775:   2.279
  DRIVER_LND_RUN                                  1.816:   2.809
  DRIVER_L2C                     <---->           0.007:   1.323
  DRIVER_LNDPOST              0.000:   0.000
  DRIVER_FRACSET              0.000:   0.000
  DRIVER_ATM_RUN                                  1.395:   1.797
  DRIVER_A2C                     <---->           0.003:   3.890
  DRIVER_ATMPOST              0.000:   0.000
  DRIVER_RESTART              0.007:   0.008
  DRIVER_HISTORY              0.000:   0.000
  DRIVER_TSTAMP_WRITE         0.000:   0.000
  DRIVER_TPROF_WRITE          0.000:   0.000
  DRIVER_RUN_LOOP_BSTOP       0.000:   0.001


More info on coupler timing:


  DRIVER_LNDPREP              0.002:   0.012
  driver_lndprep_atm2lnd      0.002:   0.011
  driver_lndprep_mrgx2l       0.001:   0.001






  DRIVER_LNDPOST              0.000:   0.000








  DRIVER_ATMPOST              0.000:   0.000


Who's new

  • rlove@...
  • afox
  • shanru@...
  • yongxiao@...
  • terry.mcguinness@...