Welcome, Guest
Username: Password: Remember me
  • Page:
  • 1
  • 2

TOPIC: Parallel Setup on Windows EXE Version

Parallel Setup on Windows EXE Version 13 years 3 months ago #1928

  • joysanyal21
  • joysanyal21's Avatar
I am running TELEMAC 2D in WINDOWS with the executable version that is available in the website.

I want to distribute the work in multiple cores of my PC. I have already installed MPICH2. I understand that now I have to change 3 things:

1) Specify a value for the keyword PARALLEL PROCESSORS

2) Change something in the file mpi_telemac.conf which is located in C:\TELEMAC\V6P0\install\win in my PC.

3) Change something in systel.ini

Now assuming I have 2 cores in my PC and my host name is 'xyz' could anyone please tell me what I should change in mpi_telemac.conf and systel.ini

Relevant part of mpi_telemac.conf in its default form is like this
# Configuring MPI for TELEMAC
#
#
# Number of processors (same value in the parameter file)
2
#
# For each host:
#
# Name_of_the_host number_of_processeur_on_this_host
#
distel-2000 2




Relevant part of systel.ini in its default form is like this

#---- MPI2
#
FC_MPI="ifort "
LK_MPI="link /debug /MACHINE:IX86 /subsystem:console /incremental:no /stack:67108864 /nodefaultlib:libc.lib /out:<EXE> <OBJS> <LIBS> "
LIBS_MPI="""C:\Program Files\MPICH2\lib\fmpich2.lib"
# les 3" au début sont important pour contourner le traitement perl qui supprime systématiquement 2"
# solution bricolée en raison du problème posé par l'espace de "Program Files"
#
#- PC multiprocesseurs
RUN_MPI="mpiexec -localonly <N> <EXE>"
- réseau de PCs
RUN_MPI="mpiexec -file mpirun.txt"
RUN_MPI="mpiexec -logon : -machinefile mpirun.txt -n <N> <EXE>"
#- PCs en réseau avec solution de connection simplifiee
#RUN_MPI="smpd -machinefile mpirun.txt"
#
What if I have only one core in my PC and I just want to test if the parallel setup is running? What should be the numbers!!

Thanks and regards,

Joy
The administrator has disabled public write access.

Re:Parallel Setup on Windows EXE Version 13 years 3 months ago #1930

  • sebourban
  • sebourban's Avatar
  • OFFLINE
  • Administrator
  • Principal Scientist
  • Posts: 814
  • Thank you received: 219
Hello Joy,

You have the correct procedure indeed.

Please note that for Step 2 (change mpi_telemac.conf) that file is provided as an example, but you do have to midify it to include the name of your computer(s) and the number of cores (or processors) you wish to use. This number has to be the same as what you put your CAS file under PARALLEL PROCESSORS. Moreover, that file mpi_telemac.conf needs to be in your project directory, where your CAS file is.

In your exmaple of mpi_telemac.conf, you should modify the name of the computer (i.e. distel-2000):
2
xyz 2

Note that in the new version of TELEMAC, which is coming soon (its Python version will not require that file anymore).

If you do not need to recompile TELEMAC, then you can leave FC_MPI, LK_MPI .. as they are. You can update LIBS_MPI="""C:\Program Files\MPICH2\lib\fmpich2.lib" so it points to your installation of MPICH2.

As for the RUN_MPI command, please only choose one (comment the one as follows:
#- PC multiprocesseurs
RUN_MPI="mpiexec -localonly <N> <EXE>"
#- réseau de PCs
#RUN_MPI="mpiexec -file mpirun.txt"
#RUN_MPI="mpiexec -logon : -machinefile mpirun.txt -n <N> <EXE>"

To answer your last question, it does not matter how many cores or processors you physically have on your computer(s). Windows will sort out the memeroy swap between your job as activated by running TELEMAC in parallel. So the numbers are up to you. Of course, if you say you want your simulation on two cores but you only have one, it will be just as fast (slow) as runing your simulation in scalar mode.

Hope this helps,

Sébastien
The administrator has disabled public write access.

Re:Parallel Setup on Windows EXE Version 13 years 3 months ago #1931

  • joysanyal21
  • joysanyal21's Avatar
Fantastic! Thanks so much Sebastien for your clear instruction. I will give it a try and see how it goes.

Joy
The administrator has disabled public write access.

Re:Parallel Setup on Windows EXE Version 13 years 3 months ago #1938

  • joysanyal21
  • joysanyal21's Avatar
Hi Sébastien
I tried to follow your instructions but the program stopped giving the following message:


**** MPI MACHINE******

MPI Machine OK <with 2 processors>

*****RUNNING*******
MPI launcher: mpiexex -localonly 2 out30212_winexe
'mpiexec' is not recognized as an internal or external command, operable program or batch file.


Any idea what might have cause this problem!

Thanks very much for your help.

Joy
The administrator has disabled public write access.

Re:Parallel Setup on Windows EXE Version 13 years 3 months ago #1939

  • sebourban
  • sebourban's Avatar
  • OFFLINE
  • Administrator
  • Principal Scientist
  • Posts: 814
  • Thank you received: 219
Hello Joy,

Two things:

1.- Your error message say "MPI launcher: mpiexex -localonly 2 out30212_winexe
I presume it is not mpiexex but mpiexec.

2.- If mpiexec is not recognised, it means that you might not have a valid installation of MPICH2. Check that mpiexec (<where-ever-you-installed-MPICH2> mpi\bin\ ) is in your PATH, and add it to your PATH if not. You can check that mpiexec exist indeed at that locations.

Hope this help.

Sébastien.
The administrator has disabled public write access.

Re:Parallel Setup on Windows EXE Version 13 years 3 months ago #1953

  • joysanyal21
  • joysanyal21's Avatar
Hi added the MPI path in the PATH variable of my system and that appatently solved the previous problem of not recognizing 'mpiexec' as a command.

BUT now I get the following error after MPI lauches (It asked for my password!) with 2 processors:
forrtl: severe (157): Program Exception - access violation
Image PC Routine Line Source
out2052_win.exe 00445ECE Unknown Unknown Unknown
out2052_win.exe 004619BB Unknown Unknown Unknown
out2052_win.exe 00401724 Unknown Unknown Unknown
out2052_win.exe 006C036C Unknown Unknown Unknown
out2052_win.exe 006A3969 Unknown Unknown Unknown
kernel32.dll 76021194 Unknown Unknown Unknown
ntdll.dll 76F5B495 Unknown Unknown Unknown
ntdll.dll 76F5B468 Unknown Unknown Unknown
forrtl: severe (157): Program Exception - access violation
Image PC Routine Line Source
out2052_win.exe 00445ECE Unknown Unknown Unknown
out2052_win.exe 004619BB Unknown Unknown Unknown
out2052_win.exe 00401724 Unknown Unknown Unknown
out2052_win.exe 006C036C Unknown Unknown Unknown
out2052_win.exe 006A3969 Unknown Unknown Unknown
kernel32.dll 76021194 Unknown Unknown Unknown
ntdll.dll 76F5B495 Unknown Unknown Unknown
ntdll.dll 76F5B468 Unknown Unknown Unknown

job aborted:
rank: node: exit code[: error message]
0: west166: 157: process 0 exited without calling finalize
1: west166: 123
Duration of job : 25 seconds ( 0:0:25 ) (system=0 sec)
______________________________________________________________________________
*** FILES DELIVERY ***

- RESULTS FILE : NoLakeResult4.slf

(merged from 2 processors)

________________________________________________________
Execution finished: telemac2d.bat
________________________________________________________
No compilation/linking/file errors detected.
No execution errors detected.
Missing braces on \N{} at delete_cas2052.bat line 27, within string
Missing braces on \N{} at delete_cas2052.bat line 36, within string
Missing braces on \N{} at delete_cas2052.bat line 37, within string
Missing braces on \N{} at delete_cas2052.bat line 38, within string
Missing braces on \N{} at delete_cas2052.bat line 39, within string
Execution of delete_cas2052.bat aborted due to compilation errors.
Returning exit status 0

=========================================================
Telemac System Freeware 6.0 - Perl scripts version V6.0
=========================================================
...stopping.


The message of MPI launch was as follows:
ALLOCATION OF USER FILES ***

- STEERING FILE : cas
- DICTIONARY : telemac2dv6p0.dico

- GEOMETRY FILE : Geo267.slf
(split for 2 processors)
- BOUNDARY CONDITIONS FILE : cas.conlim

(split for 2 processors)
- PREVIOUS COMPUTATION FILE : condInit.ser

(split for 2 processors)
______________________________________________________________________________
*** MPI MACHINE ***
MPI machine ok (with 2 processors).
______________________________________________________________________________
*** RUNNING ***

MPI launcher : mpiexec -file mpirun.txt
User credentials needed to launch processes:
account (domain\user) [GEOG\tpcv24]:
password:
MASTER PROCESSOR NUMBER 0 OF THE GROUP OF 2
EXECUTABLE FILE: C:\TELEMODEL\NoLake\cas2052_tmp\A.EXE
BARRIER PASSED

I don't understand I went wrong? Why is it saying access violation!!

Any help will be greatly appriciated!
The administrator has disabled public write access.

Re:Parallel Setup on Windows EXE Version 13 years 3 months ago #1955

  • jmhervouet
  • jmhervouet's Avatar
Hello,

I see no listing showing the beginning of an execution (a line TELEMAC-2D LISTING followed by a banner), so it could still be a computer or installation problem. However what is strange is that the program seems to run during 25 s, does in work in scalar mode, do you have any specific Fortran file that could raise problem in parallel, e.g. addressing points with their global number (which is changed in parallel) ? To have more information on listing you can add DEBUGGER = 1 in your parameter file.

Hope this helps,

Jean-Michel Hervouet
The administrator has disabled public write access.

Re:Parallel Setup on Windows EXE Version 13 years 3 months ago #1973

  • c.coulet
  • c.coulet's Avatar
  • OFFLINE
  • Moderator
  • Posts: 3722
  • Thank you received: 1031
Hi
Are you running on a 64bits machine under Seven?
If so, you should know that the distributed version is only a 32 bits which is not comaptible with MPICH2-64bits.
We cannot distribute a 64bit version on windows because we found a bug in the version 6.0 which generate memory error.
the next version will normally solve this bug.

regards
Christophe
The administrator has disabled public write access.

Re:Parallel Setup on Windows EXE Version 13 years 3 months ago #1987

  • joysanyal21
  • joysanyal21's Avatar
Hi,
Although I am running Windows7, it is a 32-bit version OS with core2 duo and I installed a 32-bit version of MPICH2 but still unable to run it in parallel mode.

I have another question:

In a 64-bit WINDOWS7 machine with 2*4=8 cores, If I install the EXE version of 32-bit TELEMAC 2d and 32-Bit version of MPICH2 will it run in parallel mode. Ofcource TELEMAC won't be able to access more than 3 GB but otherwise will it run?

Thanks for your previous reply,

Joy
The administrator has disabled public write access.

Re:Parallel Setup on Windows EXE Version 13 years 3 months ago #1989

  • joysanyal21
  • joysanyal21's Avatar
The Latest Error for running the executable version of TELEMAC 2d in parallel mode in a 32-bin WIN7 system is al follows:

Telemac System Freeware 6.0 - Perl scripts version V6.0
=========================================================
starting...

HOSTTYPE : win
PROJECT : C:\TELEMAC\V6P0
BASE DIRECTORY : C:\TELEMODEL\NoLake
LAUNCH DIRECTORY : C:\TELEMODEL\NoLake
WORK DIRECTORY : C:\TELEMODEL\NoLake\cas1680_tmp
PARAMETER FILE : cas


*** Using default configuration file :
C:\TELEMAC\V6P0\config\systel.ini ***



*** Using CUSTOM MPI configuration file :
C:\TELEMODEL\NoLake\mpi_telemac.conf ***


*** TELEMAC2D ON STATION ***


*** Interactive mode ***


*** RELEASE v6p0 ***

________________________________________________________
Steering file : cas
________________________________________________________

________________________________________________________
Starting execution: telemac2d.bat
________________________________________________________
______________________________________________________________________________
*** DEFAULT PARALLEL EXECUTABLE ***

C:\TELEMAC\V6P0\telemac2d\tel2d_v6p0\win\telemac2ddv6p0_MP.exe
DEFAULT EXECUTABLE NOT FOUND
______________________________________________________________________________
*** ALLOCATION OF USER FILES ***

- STEERING FILE : cas
- DICTIONARY : telemac2dv6p0.dico

- GEOMETRY FILE : Geo267.slf
(split for 2 processors)
- BOUNDARY CONDITIONS FILE : cas.conlim

(split for 2 processors)
- PREVIOUS COMPUTATION FILE : condInit.ser

(split for 2 processors)
______________________________________________________________________________
*** MPI MACHINE ***
MPI machine ok (with 2 processors).
______________________________________________________________________________
*** RUNNING ***

MPI launcher : mpiexec -file mpirun.txt
User credentials needed to launch processes:
account (domain\user) [GEOG\tpcv24]:
password:
launch failed: CreateProcess(C:\TELEMODEL\NoLake\cas1680_tmp\out1680_win.exe) on
'west166.geog.ad.dur.ac.uk' failed, error 2 - The system cannot find the file s
pecified.

launch failed: CreateProcess(C:\TELEMODEL\NoLake\cas1680_tmp\out1680_win.exe) on
'west166.geog.ad.dur.ac.uk' failed, error 2 - The system cannot find the file s
pecified.

Duration of job : 48 seconds ( 0:0:48 ) (system=0.016 sec)
______________________________________________________________________________
*** FILES DELIVERY ***

- RESULTS FILE : NoLakeResult4.slf

ERROR : RESTITUTION FILE NoLakeResult4.slf
________________________________________________________
Execution finished: telemac2d.bat
________________________________________________________
No compilation/linking/file errors detected.
No execution errors detected.
Missing braces on \N{} at delete_cas1680.bat line 27, within string
Missing braces on \N{} at delete_cas1680.bat line 36, within string
Missing braces on \N{} at delete_cas1680.bat line 37, within string
Missing braces on \N{} at delete_cas1680.bat line 38, within string
Missing braces on \N{} at delete_cas1680.bat line 39, within string
Execution of delete_cas1680.bat aborted due to compilation errors.
Returning exit status 0

=========================================================
Telemac System Freeware 6.0 - Perl scripts version V6.0
=========================================================
...stopping.



Why is it telling that the 'Default executable not found'? I checked that file is there in the ...\win folder.

It will be of really great help if someone can tell me what is going wrong here? I believe there is no problem with my MPICH2 installation!

Thanks
The administrator has disabled public write access.
  • Page:
  • 1
  • 2
Moderators: borisb

The open TELEMAC-MASCARET template for Joomla!2.5, the HTML 4 version.