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

TOPIC: Error MPI config gfortran/openMPI

Error MPI config gfortran/openMPI 6 years 9 months ago #28997

  • j.dasilva
  • j.dasilva's Avatar
Hi everyone,

I come here for ask some help regarding the installation on Linux.

For the story, I use telemac system since few years now. I have already installed (scalar and MPI) on several computers the code but only for one processor with the configuration gfortran and openMPI.

Actually, I trying to install (with the same configuration gfortran & openMPI) the telemac system on a workstation with 2 processors of 22 cores each.

The installation of Metis seems ok. I have completed the compilation without error. The scalar mode works but not with the MPI. I put in attachment my systel config file and below I have copy/paste the error when I try to run an example of test case.

If you have some advices it would be really appreciated.

Thank you for your time.

julien

... modifying run command to MPI instruction

... modifying run command to PARTEL instruction

... partitioning base files (geo, conlim, sections, zones and weirs)
    +> /home/c9p7ndn2/TELEMAC_MASCARET/v7p3r0/builds/susgfopenmpi/bin/partel < PARTEL.PAR >> partel_T2DGEO.log
   Current memory used:           0 bytes
   Maximum memory used:           0 bytes
***Memory allocation failed for CreateGraphDual: nptr. Requested size: 158879430250632 bytes

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
#0  0x7F986F48BE08
#1  0x7F986F48AF90
#2  0x7F986EBA54AF
#3  0x40BEE3 in partel_ at partel.F:1037
#4  0x42860F in MAIN__ at homere_partel.f:320
Segmentation fault (core dumped)
runPartition:
   |runPARTEL: Could not split your file T2DGEO (runcode=139) with the error as follows:
   |
   |... The following command failed for the reason above (or below)
   |/home/c9p7ndn2/TELEMAC_MASCARET/v7p3r0/builds/susgfopenmpi/bin/partel < PARTEL.PAR >> partel_T2DGEO.log
   |
   |      You may have forgotten to compile PARTEL with the appropriate compiler directive
   |        (add -DHAVE_MPI to your cmd_obj in your configuration file).
Attachments:
The administrator has disabled public write access.

Error MPI config gfortran/openMPI 6 years 9 months ago #28999

  • c.coulet
  • c.coulet's Avatar
  • OFFLINE
  • Moderator
  • Posts: 3722
  • Thank you received: 1031
Hi
the problem comes from the partitioning step (partel).
But the error message shows a problem of memory allocation in CreateGraphDual which is a metis source...

So you should check the compilation of metis and if possible test it
Then check again the parallel compilation of Telemac.

In yoru config file I found strange to use gfortran for the compilation and mpif90 for the link. We only use mpif90 on our computer

regards
Christophe
The administrator has disabled public write access.
The following user(s) said Thank You: j.dasilva

Error MPI config gfortran/openMPI 6 years 2 months ago #31376

  • Yunhao Song
  • Yunhao Song's Avatar
  • OFFLINE
  • Senior Boarder
  • Posts: 118
  • Thank you received: 9
Hi Mr Coulet,

When I tested the aed2 examples on Ubuntu 18.04 the partitioning error also occurred, please check the info below
... partitioning base files (geo, conlim, sections, zones and weirs)
+> /mnt/e/TELEMAC_Code/v7p3r1/builds/ubugfmpich/bin/partel < PARTEL.PAR >> partel_.log
STOP 1
runPartition:
|runPARTEL: Could not split your file (runcode=1) with the error as follows:
|
|... The following command failed for the reason above (or below)
|/mnt/e/TELEMAC_Code/v7p3r1/builds/ubugfmpich/bin/partel < PARTEL.PAR >> partel_.log
|
| You may have forgotten to compile PARTEL with the appropriate compiler directive
| (add -DHAVE_MPI to your cmd_obj in your configuration file).
|

I also tested some other t2d/t3d examples in parallel mode and no error occurred. Could you give me some advice?

Kind regards,
Yunhao
The administrator has disabled public write access.

Error MPI config gfortran/openMPI 6 years 2 months ago #31381

  • c.coulet
  • c.coulet's Avatar
  • OFFLINE
  • Moderator
  • Posts: 3722
  • Thank you received: 1031
Hi
Could you give us more information?
There is a problem during the partition step but we can't help you with only this message.
We need at least the partel.log file

Regards
Christophe
The administrator has disabled public write access.

Error MPI config gfortran/openMPI 6 years 2 months ago #31386

  • Yunhao Song
  • Yunhao Song's Avatar
  • OFFLINE
  • Senior Boarder
  • Posts: 118
  • Thank you received: 9
Hi Mr Coulet,

The partitioning error emerged when running example 'waq3d_aed2' in parallel mode (2 cores), please see the two partel log files attached.

When running in serial mode, error occurred as 'Segmentation fault - invalid memory reference', please check the .sortie file attached.

Thank you for your consideration! (sorry about the re-editting as the attachments failed to be unloaded at first time)

Yunhao


File Attachment:

File Name: partel_.log
File Size: 1 KB


File Attachment:

File Name: partel_T3DGEO.log
File Size: 3 KB


File Attachment:

File Name: sortiefile.txt
File Size: 9 KB
The administrator has disabled public write access.

Error MPI config gfortran/openMPI 6 years 2 months ago #31387

  • c.coulet
  • c.coulet's Avatar
  • OFFLINE
  • Moderator
  • Posts: 3722
  • Thank you received: 1031
I can't access to the log file.
Rename them .txt before joining them

Regards
Christophe
The administrator has disabled public write access.

Error MPI config gfortran/openMPI 6 years 2 months ago #31388

  • Yunhao Song
  • Yunhao Song's Avatar
  • OFFLINE
  • Senior Boarder
  • Posts: 118
  • Thank you received: 9
Please check the files below, sorry for the inconvenience.

File Attachment:

File Name: partel_.txt
File Size: 1 KB


File Attachment:

File Name: partel_T3DGEO.txt
File Size: 3 KB


File Attachment:

File Name: sortiefile.txt
File Size: 9 KB
The administrator has disabled public write access.

Error MPI config gfortran/openMPI 6 years 2 months ago #31389

  • c.coulet
  • c.coulet's Avatar
  • OFFLINE
  • Moderator
  • Posts: 3722
  • Thank you received: 1031
Hi
The geometry file is well partitionned.
But it seems there is another partionning step for another file which doesn't work.
Could you join the steering file?
Christophe
The administrator has disabled public write access.

Error MPI config gfortran/openMPI 6 years 2 months ago #31390

  • Yunhao Song
  • Yunhao Song's Avatar
  • OFFLINE
  • Senior Boarder
  • Posts: 118
  • Thank you received: 9
Hi Mr.Coulet,

I barely changed the steering file, only added 'PARALLEL PROCESSORS = 2'. Please see attached.

Regards,

File Attachment:

File Name: t3d_waq3d_aed2.txt
File Size: 3 KB
The administrator has disabled public write access.

Error MPI config gfortran/openMPI 6 years 9 months ago #29003

  • j.dasilva
  • j.dasilva's Avatar
Thank you Mr coulet for your advice.

i have checked the configuration of metis and compiled again.
i have just changed the configuration options inside the Makefile and it works !!

Thank you again

regards

Julien
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.