[RegCNET] Two questions for regcm4 running

[黄河]

Dear Craziano and others,
 
Thanks for your suggestion. The question of the NNRP2 data has been settled. But the other question is lingering. An error occured in mpirun. There is no such problem when carring out the serial computing.
 
Thank you very much! Looking for your reply.
Huanghe  
----------------------------------
[huanghe at water Bin]$ mpirun -np 4 /home/huanghe/regcm/Bin/regcmMPI regcm3.in
application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0
  This is RegCM version 4
   SVN Revision: exported compiled at: data : Oct  6 2010  time: 15:16:08
 : this run started at    : Fri Oct  8 11:12:48 2010
 : it is submitted by   :
 : it is running on     : water
 : it is using          :            4   processors
 : in directory         : /home/huanghe/regcm/Bin
 param: RESTARTPARAM READ IN
 param: TIMEPARAM READ IN
 param: OUTPARAM READ IN
 param: PHYSICSPARAM READ IN
 param: SUBEXPARAM READ IN
 param: EMANPARAM READ IN
  param: dtau =    25.0000000000000        50.0000000000000
  nrec =       289276
  param: IDATE1, IDATE2, dtmin, ktaur =   1989050100  1989060100   1.66666666666667                0
 READING HEADER FILE:/home/huanghe/simulation/input/LAT+16LON067_160x192.INFO                                                                                                                                                                   
 param: DIMS -1610612736 -1073741824   301989888
 param: DOMAIN  7.3709743E-39 -5.2156269E-17  4.8966263E-14 -5.2156269E-17  4.8966263E-14 -3.3390954E-37
 param: PROJLAMCON
 param: SIGMA   0.000000     -4.2844355E+08 -4.2949213E+08  2.0717004E-32  5.6505108E-20  3.9646442E+11  1.7648097E-26 -3.6050446E+14 -4.2791933E+08 -4.5276890E-11 -1.9150387E-29  1.7597612E-26  3.3774611E-02  2.0741079E-32  7.7069599E+35 -1.9175039E-29   461778.0     -5.2147792E-17  4.6006030E-41
 param: PTOP  5.7486868E-41
 param: OUTPUT    16777216    16777216
  param:  SET IN regcm.in:  IY=         160  JX=         192  KX=          18
  param:  SET IN TERRAIN: IYY= -1610612736  JXX= -1073741824  KZZ=   301989888
  -------------- FATAL CALLED ---------------
  Fatal in file: param.F90 at line:    779
  IMPROPER DIMENSION SPECIFICATION
  -------------------------------------------
rank 0 in job 21  water_33111   caused collective abort of all ranks
  exit status of rank 0: return code 1

---------------------------------

--- 10年10月8日，周五, Graziano Giuliani <graziano.giuliani at aquila.infn.it> 写道：


发件人: Graziano Giuliani <graziano.giuliani at aquila.infn.it>
主题: Re: [RegCNET] Two questions for regcm4 running
收件人: regcnet at lists.ictp.it
日期: 2010年10月8日,周五,上午2:54


This IMHO can be caused by improper specification of ibigend and/or the use of compiler flag to force big endian ordering of I/O or wrong ibyte related record length on direct access I/O and for NNRP2 failure, this can also be due to a non correct check if a file was successfully opened before read.

So here my suggestions:

1) FIX ibigend to 1, fix ibyte to 4, fix iotyp to 1
2) Check compiler flags in Makefile.in for a consistent I/O (i.e. for intel -convert big_endian -assume byterecl) are active.
3) For NNRP2, check all needed files are present with correct names in the input directory.

Ciao!
_______________________________________________
RegCNET mailing list
RegCNET at lists.ictp.it
https://lists.ictp.it/mailman/listinfo/regcnet



      
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.ictp.it/pipermail/regcnet/attachments/20101009/5209a5ff/attachment.html>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: regcm3.in
URL: <https://lists.ictp.it/pipermail/regcnet/attachments/20101009/5209a5ff/attachment-0002.in>