[mesa-users] Test Failure on Supercomputer

Ben Brown bpbrown at astro.wisc.edu
Wed Jun 5 10:29:16 EDT 2013


Jared,
     Could you also let us know what machine (which supercomputer) you're 
trying to compile on?  Also, have you successfully compiled other codes on 
this machine yet?

In my experiences, some of these machines require architecture specific 
libraries and compilers, especially when using mpi and openmp.  Depending 
on the machine, the vanilla MESA SDK may not work on these systems.

--Ben

> Date: Tue, 4 Jun 2013 19:28:41 -0500
> From: Richard Townsend <townsend at astro.wisc.edu>
> Subject: Re: [mesa-users] Test Failure on Supercomputer
> To: Aaron Dotter <aaron.dotter at gmail.com>
> Cc: mesa-users <mesa-users at lists.sourceforge.net>,	Jared Brooks
> 	<jcbrooks at physics.ucsb.edu>
> Message-ID: <AB613A75-6B33-4867-A939-87B20CA18D0B at astro.wisc.edu>
> Content-Type: text/plain; charset=iso-8859-1
>
> As an addendum: Jared, can you post the output of gfortran -v? This gives more information than gfortran --version. (I've updated the SDK website to specify -v instead of --version).
>
> On Jun 4, 2013, at 7:22 PM, Aaron Dotter wrote:
>
>> Hi Jared,
>>
>> Is it possible you installed the 32-bit version of the SDK rather than the 64-bit version?
>>
>> We saw some issues with allocating tables a few months ago, but this was specifically related to 32-bit OS (and therefore SDK).  It is possible to run 32-bit programs on 64-bit machines, but then you may run into such issues unnecessarily.
>>
>> That problem ought not to occur in a 64-bit OS with a 64-bit compiler unless the supercomputer has some strict settings on resources available to a process (check with 'ulimit -a').
>>
>> Aaron
>>
>>
>> On Wed, Jun 5, 2013 at 10:15 AM, Richard Townsend <townsend at astro.wisc.edu> wrote:
>> That's a pretty old SDK. If nobody suggest a better option, you could always try upgrading it.
>>
>> On Jun 4, 2013, at 7:11 PM, Jared Brooks wrote:
>>
>>> I'm trying to install mesa (4942) on a Linux based supercomputer using the mesa sdk.  The installation runs fine until it hits the star test cases and fails.
>>>
>>> This is the screen output:
>>>
>>> first file is short
>>>
>>> /home/jbrooks/mesa_4942/star/test
>>> TEST FAILED -- compare test_output to tmp.txt
>>>
>>>
>>> /home/jbrooks/mesa_4942/star
>>> ./build_and_test FAILED
>>>
>>> I went into star/test and looked in log to find this:
>>>
>>>  saved initial &star_job inlist values: star_job_namelist.out
>>>                                          version_number        4942
>>>  Load1_eosDT_Table ierr        5014           1           3   0.0000000000000000        4.0000000000000001E-002
>>>  Load1_eosDT_Table ierr        5014           2           1  0.20000000000000001        0.0000000000000000
>>>  Load1_eosDT_Table ierr        5014           2           1  0.20000000000000001        0.0000000000000000
>>>  Load1_eosDT_Table ierr        5014           1           1   0.0000000000000000        0.0000000000000000
>>>  Load1_eosDT_Table ierr        5010           1           1   0.0000000000000000        0.0000000000000000
>>>  Load1_eosDT_Table ierr        5010           1           1   0.0000000000000000        0.0000000000000000
>>>  Load1_eosDT_Table ierr        5010           1           1   0.0000000000000000        0.0000000000000000
>>>  Load1_eosDT_Table ierr        5010           1           1   0.0000000000000000        0.0000000000000000
>>>  finish_load_model: failed in set_vars
>>>  failed in finish_load_model
>>>  star_load_zams ierr          -1
>>> WARNING: allocate failed for eos tables
>>>  do_load1_star ierr          -1
>>> WARNING: allocate failed for eos tables
>>>
>>> Here is my mesa sdk troubleshooting info:
>>>
>>> [jbrooks at tscc-login1 mesa_4942]$ uname -a
>>> Linux tscc-login1.sdsc.edu 2.6.32-279.22.1.el6.x86_64 #1 SMP Tue May 14 17:25:16 PDT 2013 x86_64 x86_64 x86_64 GNU/Linux
>>> [jbrooks at tscc-login1 mesa_4942]$ gfortran --version
>>> GNU Fortran (GCC) 4.7.2
>>> Copyright (C) 2012 Free Software Foundation, Inc.
>>>
>>> GNU Fortran comes with NO WARRANTY, to the extent permitted by law.
>>> You may redistribute copies of GNU Fortran
>>> under the terms of the GNU General Public License.
>>> For more information about these matters, see the file named COPYING
>>>
>>> [jbrooks at tscc-login1 mesa_4942]$ echo $MESASDK_ROOT
>>> /home/jbrooks/mesasdk
>>> [jbrooks at tscc-login1 mesa_4942]$ echo $PATH
>>> /home/jbrooks/mesasdk/bin:/projects/itp-group/mesasdk/bin/:/opt/openmpi/intel/ib/bin:/opt/intel/composer_xe_2013.1.117/bin/intel64:/usr/lib64/qt-3.3/bin:/usr/local/bin:/bin:/usr/bin:/usr/local/sbin:/usr/sbin:/sbin:/opt/ganglia/bin:/opt/ganglia/sbin:/opt/gold/bin:/opt/gold/sbin::/usr/java/latest/bin:/opt/maui/bin:/opt/torque/bin:/opt/torque/sbin:/opt/torque/bin:opt/torque/sbin:/opt/pdsh/bin:/opt/rocks/bin:/opt/rocks/sbin:/home/jbrooks/bin
>>>
>>>
>>> Any help would appreciated.
>>>
>>> Thanks,
>>>
>>> Jared




More information about the Mesa-users mailing list