Software Coordinating Committee Conference Call
		       August 21, 2003 3:00 PM EDT
			  Recorder: C. DeTar

Present:
  Osborn, DeTar, Brower, Mendes, Holmgren, Edwards, Mawhinney, Watson

** action items
=====================================================================

#1. SciDAC extension software budget 
    --------------------------------

I need to make up a software budget for SciDAC extension for June '04
to June '06 period. Tell me what you think we need and why? (In reality
we will have at best flat funding in '04 and '05 is not know but flat is
a good guess!)

**  Brower: Need input before early next week.


#2. Tests and Benchmarks. 
    ---------------------

We made a lot of progress in the last 10 days (thanks to Robert,
Carleton, James, Steve G et al) BUT there are still some loose ends
and more to do.

Look at the page

    http://physics.bu.edu/~brower/tests

to see what has changed and what is missing!

    Brower: We need FNAL and JLab numbers for Robert's
            SSE improved Wilson, but including the clover term
    Edwards: I don't have clover SSE.  It will take some time.
    Brower: Can we get at least Wilson/SSE on FNAL as well as JLab?
**  Edwards: Yes.
    Brower: Is any group in the US using clover dynamical at present?
    DeTar:  Not that I know of.
    Brower: Then perhaps we can simply drop that from the table.
    Brower: I would like performance numbers from real production
            runs.
    Edwards: This can be done for domain wall code on JLab.
    DeTar:  We are running small specialized projects at FNAL
            such as topological susceptibility, so the numbers may 
            not be very useful.
    Brower:  Anything we have that we can brag about would be good.
    Brower:  I would like details of how the code was compiled and run.
    Holmgren: I can just copy over whole directories of code to
            make a snapshot of the benchmark environment.
**  DeTar:  OK, I'll work with Don.
**  Edwards: This can also be done with SZIN.   


#3. Lessons learned runing at JLab/FNAL: 
    -----------------------------------
 
Let's take opportunity (while it is fresh in our minds) to make hit
list of difficulties encountered by our "experts" running code at JLab
and FNAL.  We should have a plan to make the 2 enviroments easier to
use AND as nearly as possible IDENTICAL. (I would like to be able to
move application codes with their Makefiles and shell scripts
unchanged, recompile and submit it. Maybe not even recompile!)

    Brower: We need to make the environments look identical.
    Watson: Don Petravik will be visiting JLab soon to discuss coordination.
    DeTar:  What can be done about moving files directly between JLab
            and FNAL?  I couldn't get Kerberos to work.  I believe
            that was because of the firewall.
    Watson: In the future everything will be done on a grid.  We are
            working on JPARS for transferring files.  Jie is doing the
            coding. It is Java based with PKI credentials, and is
            being tested by MIT/JLab.  For SciDAC it should be mounted
            from BNL.  If someone wants to try it, contact me and Jie.
    DeTar:  JLab also needs a user info web page with compiler, MPI,
            PBS, etc information.
    Watson: We are working with Don on run time environment.  Any comments.
    DeTar:  Staging large files to the batch nodes is an unsolved
            problem.
    

#4.Response to QCDML draft ver-4.0 (July 8, 2003)
   ---------------------------------------------

See Carleton's Email Sun 17 Aug 2003 "Comments from MILC" (posted on
my webpage of course)

    DeTar:   [Summarized the "Comments from MILC" document.]
    Edwards: For ILDG we need to be specific about what we want.
    Brower:  Norman suggested a mechanism for putting configurations
             in the archive, but accessible only to a group of
             collaborators.
    DeTar:   MILC hasn't felt the need for it, but I can understand
             that some collaborations may want it.  It should probably
             be done only for a short time.  ILDG should discuss it.
    Brower:  Can we send the document on to ILDG?
    DeTar:   Let's let Robert edit the document to something
             that says specifically what we want, run it by me,
             and then we send it on.
**  Edwards: OK


#5.I/O draft completion and message to ILDG for Comment.
   ----------------------------------------------------

    Brower:  We'll postpone this to next week when Jim Simone returns.    

=====================================================================
Committee conference concluded at 4:40 PM EDT.  
Next plenary call August 28 at 3 PM EDT.