1. Introduction

The QuantumCube computer resource is an 8 processors system runing the Linux operating system.

2. Hardware

• 4 - PQS nodes
• 2 - AMD Opteron 1.8GHz processors per node
• 4 GB of memory per node (2 GB of memory per processor)
• 10/100/1000 networking
• 845 GB raw disk space
  • 37 GB /home
  • 7.4 GB /usr/local
  • 116 GB /scr (server)
  • 202 GB /scr (other 3 nodes)

  Note that each node has a local scratch space. There is no shared scratch space.

• Configuration:
  • 1 node: file server/interactive node ( /home filesystem, /usr/local/, mail )
  • 3 nodes: compute nodes

3. Software

• Operating System : Fedora Core 2
• GNU Compilers

4. Operations

• How to login ?

  To connect, login to quantum.msi.umn.edu. The prefered method is to use ssh if it is available at your site.
  

                ssh -l username quantum.msi.umn.edu
                

  You only need to use -l if your quantum login name is different than the login name you use on your own workstation.
  

  If ssh is not available on your machine, please contact your system administrator for an installation. You can find more information about ssh at our webpage : www.msi.umn.edu/ user_support/ssh/

• How to compile a sequential code?
  • Fortran: g77 <options> mycode.f
  • C : gcc <options> mycode.c

• How to run jobs interactively?

  There is one node set aside to run jobs interactively.

  just run the executable. e.g.: ./a.out

• Parallel Jobs and PVM
  
  • To run a job in parallel you will need to set up the parallel virtual machine (PVM)
    • setup PVM by typing : pvm
    • add the slaves at the prompt: add n2
    • exit PVM and leave the configuration intact: quit
    
    
  • Important : When using PVM, various files are created in /tmp directory, including pvml.XXX and pvmd.XXX (where XXX is a number). Sometimes these files may be left behind on a job crash and their presence will prevent you from properly setting up the parallel virtual machine subsequently.

    The command pvm_reset -f all

    will remove these files on both the master and the slave and should allow you to reset PVM properly.

• How to run jobs in the queue?
  
  The QuantumCube includes a queuing utility called Sun Grid Engine (SGE). This has many features in commong with DQS. All the basic DQS commands still function under SGE, however there are differences in some of the set up commands compared to DQS.
  
  • Example script to submit a serial job:

    	 #$ -S /bin/csh
    	 #$ -cwd
    	 ./a.out
    
    

    If you copy a submission script file from another system, please make sure it is in Unix text format by running dos2unix myscript.

  • Example script to submit a PVM parallel job to run on 4 processors

            #$ -S /bin/csh
    	#$ -cwd
    	#$ -pe pvm 1-4 
    	#$ -masterq n1.q
    	#$ -q n1.q,n2.q,n1.q,n2.q
    	pqs histidine 4
    
    

  • SGE submission commands:
    • To submit a job: qsub
    • To display the current status of the queue: qstat -f
    • To delete a job from the queue: qdel
    • Manual page: man qsub
    • There is an interactive graphical queue monitoring system that can be invoked by typing qmon at the terminal prompt; this allows full monitoring and interaction with the parallel job queue graphically.

5. Software

Software and related documentation supplied by Parallel Quantum Solutions with your QuantumCube are all in the directory /progs/PQS This includes:

• pqs.x : PQS Ab initio program executable single processor version
• pqs.pvm.x : PQS Ab initio program executable parallel version
• sqm.x : SQM program for scaled quantum mechanical force field
• binrdr_new.x : Bin sort program used during parallel MP2 calculations
• pqs : shell script for running PQS jobs
• tidy : shell script for deleting old PQS files
• archive : shell script for archiving PQS files
• GAUNTLET :directory in /progs/PQS contains input and output files and run scripts for an extensive series of test jobs. The current GAUNTLET directory has six subdirectories:
  • SMALL : contains primarily small jobs to test various program options. Should be run in single-processor mode only.
  • MID : Larges jobs testing both program functionality and performance in parallel mode. Should be run in 4 processor mode.
  • BIG : As MID, but longer jobs. Run in 4-or 8-processor mode.
  • XL : Very large jobs to test program performance. Run only in 4- or 8-processor mode
  • EXAMPLES : All the test examples described in the program manual. Most of these use the PQS input style but test jobs 13-16 user the Pople-style input.
  • POPLE : Further Test jobs using the Pople-style input.

6. Documentation

Online documentation (for PSI3, NBO, PQS, etc.) can be found at http://quantum.msi.umn.edu.

7. Getting Help

Send your questions by email to help@msi.umn.edu or call (612) 626-0802.

This information is available in alternative formats upon request by individuals with disabilities. Please send email to alt-format@msi.umn.edu or call 612-624-0528.

HOME | QUESTIONS | FEEDBACK Events | Links | People | Programs | Publications | Support | Welcome

URL: http://www.msi.umn.edu /sdvl/hardware/quantum/quickstart.html This page last modified on Wednesday, 23-Apr-2008 11:06:56 CDT   Please direct questions or problems to help@msi.umn.edu   Website related questions or problems should be directed to webmaster@msi.umn.edu The Supercomputing Institute does not collect personal information on visitors to our website. For the University of Minnesota policy, see www.privacy.umn.edu.