gilgamesh: root access while I am gone

[Jason Larkin]

Right, I actually had John make a new module (or update an existing) 
before he left.  So, module load intel/intel64 now has appropriate 
environment variables for the Intel compilers/libraries.  Intel 
compilers are installed at:

/opt/intel/Compiler/11.1/072/bin/intel64

The Intel MKL at:

/opt/intel/mkl/10.2.5.035/lib/em64t

Intel MPI wrappers:

/opt/intel/impi/4.0.0.028/intel64/bin

The Intel Compilers are the non-commercial versions (not sure what that 
means, John installed them). The MPI library is the evaluation version, 
so it will expire Sept. 03 if we don't pay for the lisence (so if you 
want to compiler with mpiifort/mpif90, you better try it out now). Just 
FYI for those who want to try the Intel stuff.

Thanks for the script file,

Jason


Craig Maloney wrote:
> Hi Jason,
>
> Important points:
>
> 1) make sure you build your mpi  code after a "module load 
> openmpi-psm-gcc" command.
> 2) put that same command into your pbs script right before the "cd 
> $PBS_O_WORKDIR" command
> 3) make sure to put a cput limit and the node/process number request 
> on the command-line when issuing the qsub command.  it looks like pbs 
> is ignoring any "-l cput" or "-l nodes=:ppn=" directives in the script 
> file, but I'm not positive about that.
>
> Here's the script I am using where "../lmp_generic" is the executable 
> I built with the appropriate environment and "../input-10" is a 
> parameter file I'm sending to ../lmp_generic on stdin.
>
> ----------
> #!/bin/bash
> cd $PBS_O_WORKDIR
> module load openmpi-psm-gcc
> mpirun -np `cat $PBS_NODEFILE | wc -l` ../lmp_generic < ../input-10
> ---------------
>
> Typical qsub command:
> qsub -l cput=12:00:00 -l nodes=2:ppn=16 ../ljMelt-20
>
> where "../ljMelt-20" is the pbs script.
>
> Good luck,
> Craig
>
>
> On Aug 5, 2010, at 11:00 AM, Jason Larkin wrote:
>
>> Hi Craig,
>>
>> This is Jason, one of Alan's students.  I see that you have some jobs 
>> running on the ChemE cluster.  Could you send me the queue script 
>> file you used to submit these jobs?  I'm having trouble getting my 
>> job to submit to the queue system.  Thanks,
>>
>> Jason
>>
>>
>> Alan McGaughey wrote:
>>> Spencer's email is rommel at cmu.edu <mailto:rommel at cmu.edu>.
>>>
>>> You can look up anyone at CMU here:
>>>
>>> http://directory.andrew.cmu.edu/
>>>
>>> -----Original Message-----
>>> From: Jason Larkin [mailto:jmlarkin at andrew.cmu.edu] Sent: Thursday, 
>>> August 05, 2010 10:15 AM
>>> To: Alan McGaughey
>>> Cc: 'Minyoung Lee'
>>> Subject: Re: FW: root access while I am gone
>>>
>>>
>>> Alan,
>>>
>>> How can I get in touch with Spencer?  Do you have his email?  I need 
>>> him to do something only the root can,
>>>
>>> Jason
>>>
>>> Alan McGaughey wrote:
>>>  
>>>> fyi
>>>>
>>>>
>>>> From: johnrkitchin at gmail.com [mailto:johnrkitchin at gmail.com] On 
>>>> Behalf Of
>>>> John Kitchin
>>>> Sent: Wednesday, August 04, 2010 5:50 PM
>>>> To: Alan McGaughey; Craig Maloney
>>>> Cc: Spencer Miller
>>>> Subject: root access while I am gone
>>>>
>>>>
>>>> Hi Alan, Craig,
>>>>
>>>> I am heading out of the country tomorrow afternoon, and will be out of
>>>>    
>>> email
>>>  
>>>> contact until 8/15. I think the cluster is setup so you can do most of
>>>>    
>>> what
>>>  
>>>> you need without root access. If something comes up and you think it is
>>>> essential that it be done before I am back, I have left the root 
>>>> password
>>>> with Spencer.  He isn't planning to use it for anything, unless 
>>>> there is
>>>> some emergency.
>>>>
>>>> John
>>>>
>>>> -----------------------------------
>>>> John Kitchin
>>>> Assistant Professor
>>>> Doherty Hall A207F
>>>> Department of Chemical Engineering
>>>> Carnegie Mellon University
>>>> Pittsburgh, PA 15213
>>>> 412-268-7803
>>>> http://kitchingroup.cheme.cmu.edu
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>      
>>>
>>>
>>>
>>>
>>>  
>>
>>
>
> -----------------------------------------------
> Contact info:
> http://www.ce.cmu.edu/People/faculty/maloney.html
>
>
>