December 28, 2004

Conversation with Vijay Pande

o The importance of SMP resources is increasing, make sure we install easily and scale well in this environment. What is the current level of support for threads in GROMACS? How does GROMACS running threaded compare in terms of performance to F@C? The CCG will locate SMP boxes that we can test on.

(Postscript: There is only partial support for threads in GROMACS currently (non-bonded interactions), F@C, ie MPI, scales pretty well on SMP boxes.)

o Ease of installation, what can we do to make it as simple as possible to install and use F@C? LAM is a hurdle in this respect. Simple how-to documents for ssh (assumed), key exchange, LAM, using shared file systems, other topics?

o Does hyperthreading do anything for F@C?

(Postscript: Not really. HT is duplicated processor state resources which facilitate very fast context switches between threads in a single process. Since GROMACS doesn't use threads F@C can't leverage this.)

o Windows port, Vijay has a person that can work on this with us.

Posted by charliep at 02:43 PM | Comments (0)

December 17, 2004

Meeting Minutes - December 17, 2004

F@C

  • JoshM will setup an environment for Charlie to do testing on integration
    bug. When this is fixed and checked-in a number of todo items come into
    play.

    Plumbing

  • In addition to Gaussian/Linda Lori also needs the following:

  • Super (and you should hear with that word, "wow, things are easier
    here")--please install Molden, version 4.2 (instructions at
    http://cmbi1.cmbi.kun.nl/~schaft/molden/howtoget.html--they are for
    version 4.0 (just change the 4.x in the ftp site) but I think 4.2 is out;
    if not, 4.0 or 4.1 is fine) and Moldenogl which goes along with it (at
    http://cmbi1.cmbi.kun.nl/~schaft/molden/opengl.html). The link to
    download it for Linux is about halfway down the page. You might need MESA
    too if it's not already on the system--there's a link for that at the top
    of the opengl page.

  • Molekel (current version 4.3) at http://www.cscs.ch/molekel/ (the
    website has annoying frames, but if you click on distributions you see the
    downloadable executables and libraries. There are several platforms--I'm
    guessing you want the Linux one. Brief instructions are under FAQ; longer
    ones under Installation.

  • JoshM reports that the images for bazaar and Cairo are making progress.

  • John needs to do a serious disk cleaning of scheajo and test*. We need
    disk space.

  • Everyone else needs to go cleaning as well (see previous meeting notes).

    Numerical Methods

  • NAMD is making progress, see JoshM's MT entry. He will put together
    documentation on building and running (in software/doc with an MT pointer).

    Other

  • Recruiting, Eric may be interested, will get back to us in January. We
    need two or possibly three depending on their backgrounds.

    Posted by charliep at 01:33 PM | Comments (0)

    • update

    • Plumbing
      • Gaussian requires TCP Linda. Although this is payware, it is included in the price of Gaussian. I could see no other dependencies in their docs.
      • I spent a large chunk of yesterday making coasters out of cdrs. Could someone else try burning the yellowdog images? I've been using the burner on admin and on my machine with no luck even with md5 verified downloads.
      • NAMD source installation is progressing well on cairo. I'm currently installing the prereqs (charm++/convers, VMD molfile plugins, TCL) The user documentation was informative for basic running info. Still looking into converting gmxbench to a format NAMD can understand.
      • Cleaned admin's hdd up a bit. /cluster is 93% full and could use some love.
    • General
      • Mary Lou caught me and I helped her for a bit with SIAM information. Signs point to a stay in a Disney resort.
    Posted by mccoyjo at 12:22 PM | Comments (0)

    December 14, 2004

    Meeting Minutes - December 14, 2004

    Next meeting Friday at 1p, JoshM and Charlie in Richmond, John in Chicago, JoshH?

    NAMD - Found source, haven't found molecular systems yet. JoshM will try building on bazaar and Cairo. We need to find either molecular systems for NAMD that are comparable to the ones from GMXBENCH or figure-out how to adapt GMXBENCH systems for use with NAMD.

    Imaging - New SystemImager installed, pulled one image from existing server

    Plumbing - Disk cleaning on admin or possibly a new bigger disk. Maybe the last of the HHMI money?

    Plumbing - /cluster needs to be cleaned as well. All of us should take a tour and clean old runs, etc. where possible.

    Lori's software - JoshM will track-down Gaussian/Linda (from Lori) for the new bazaar image.

    Toby's software - JoshM will track-down distcc and ccache for the new cairo image.

    Midwife - No report.

    F@C - Charlie needs to look at the error JoshM forwarded to him. Charlie will contact Vijay and see how it's going.

    SIAM - Charlie will make conference and hotel reservations next week for all of us.

    Recruiting - Application due-date is tomorrow.

    Posted by charliep at 11:57 AM | Comments (0)

    December 09, 2004

    update

    • Folding@Clusters
      • Have been answering John's question's about the midwife.
      • Problems with making mdrun a stand-alone binary. Charlie is helping out.
    • Plumbing
      • Successfully downloaded SUSE 9.1 and Yellowdog 4.0. We need more CD-Rs to use the iso's.
      • Installed the latest systemimager on admin and successfully pulled an image to it. Need to re-read and understand the problem with systemimager and dhcp 3.
    • Numerical Methods
      • Installed NAMD on bazaar in /cluster/bazaar/software/NAMD_2.5_Linux-i686/ (need to find the source; I could only find a precompiled linux-i686 version). There was no linux-ppc version, but there was on for OSX.
    Posted by mccoyjo at 07:15 PM | Comments (0)

    December 03, 2004

    Meeting Minutes - December 3, 2004

    Scheduling


      Busy two months coming-up. JoshH loose between Dec 18 - late January.
      Charlie gone Dec 8-12 and 14-16, working on this stuff otherwise until
      classes start. John working remotely starting after exams.

      JoshM busy this week and part of next on graduate school stuff. Couple
      more coming up in Jan/Feb.

      John is behind on everything (Charlie is too but doesn't admit it).

      Towards late January/early February we all (four) need to gather for a
      couple of days.

    SIAM CSE05


      Hotel, airplane, conference registration. (charlie)

      Two posters and one educational session to design and assemble.

    Numerical Methods


      NAMD - nothing yet (joshm). Once it's installed John and Charlie will
      figure-out the recipe for running our molecules with it (or theirs).

      Charlie has a paper that he will copy and distribute.

    Folding@Clusters


      New testing recipe, John will start writing mother.conf and using the ability
      to specify nodes on the mpirun command line. lam still needs to start
      and stop between each run.

      a1 release fails miserably with poly-ch2. Try straight-up GROMACS before
      we spend a lot of time on this. (john)

      testing results for release a1 can be found here:
      http://cluster.earlham.edu/detail/home/test2/results-cairo.html
      http://cluster.earlham.edu/detail/home/test2/results-bazaar.html

      Code changes - divorcing mdrun in test now. New developers documentation
      and changes to makefile.am, tabs and 80
      columns still need to be completed. (joshm)

      The To-Do list needs freshening. (charlie)

      When to do our major re-write? Before or after beta, may depend on the
      nature of the feedback we get.

    Plumbing


      Bazaar upgrade - SuSe 9.1 (including ccache and distcc). Maybe some
      help from Dawit.

      Cairo upgrade - YellowDog 4.0 (including ccache and distcc).

      SystemImager upgrade.

      As part of all this upgrading we should be able to finally have a complete,
      accurate list of all the 0th node issues.

      Node list - Toby is using c4, JoshM using c2 for the golden client, John using
      c5-c15 for testing. John using b5-b15 for testing.

    General


      Lunch today for the applied and research groups, need 2? Update - looks like
      we will have funding for 2 students next summer. Should we try to do
      recruiting now or wait until classes start in January?

      Gaussian - Lori Watson is ordering the right release, we'll run it on Bazaar.

    Posted by charliep at 10:17 AM | Comments (0)