Software Coordinating Committee Conference Call
		       September 30, 2004 3:00 PM EDT
			  Recorder: C. DeTar

Present:

  Holmgren, Simone, Dru Renner, Ying Zhang, DeTar, Francesca Maresca,
  Brower, Mawhinney, Osborn, Watson, Jung

Absent:

  Edwards
  

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

   5.) Status of Grid node to share SciDAC lattices --- Don/Chip/Tom Schlagel

     Holmgren:  Don't have any news.  I'll check with Tom Schlagel

   6.) When should we return to "uniform Queuing/Scripting/FileFormats
       etc" conventions for BNL/FNAL/JLab user environment?

     Holmgren:  We need to create a new testing regime.
        I would like to get something out by mid-October
     DeTar: If you need something from me, let me know.

   1.) Implementation of QIO/Host interface  --- Carleton
        (Vote on Partitioned I/O support: Yea ___  Nay___   
         See Carleton's message appended below)

     DeTar: Explained the choices
     Simone: I would favor keeping the two new QMP calls
     DeTar: That would be OK as long as QIO is not required
       to support the case QMP_io_node(node) != node
     Brower: I would worry that we'll need the PARTFILE for
       machines like Blue Gene L which may have a specialized I/O
       node
     Mawhinney: Surely BGL supports MPI, but that doesn't imply
       support for each node 
     DeTar: Can we find out whether BGL will need PARTFILE?
       We will wait to hear.

   2.) Status of QMP Test Suite  Celso Mendes/Ying Zhang

     Ying: I am getting accounts on various machines
       I am building MILC code on one of the specialized NERSC machines
       Whom can I contact about testing the build?
**   DeTar: There are sample input and output files.  But comparing
       output files is not straightforward.  You can ask me when
       you have one.
     Osborn: For benchmarking, you will want one of our
       benchmarking scripts.  I have one, Carleton and Steve Gottlieb
       also have one.
**   Brower: Why don't you contact both James and Carleton for these scripts?

   3.) Status of Asqtad Level 3 Code -- Chulwoo
         (Odd axes/Single Precision/interface/User Documentation)

     Brower: Chulwoo reminded me that he is doing QMP first
        as per committee request.
**   Brower: We need documentation about API.
     Brower: How is QMP coming along?
     Jung: I am working on the basic parts.  I am working on
       allowing faster nonblocking NN and slower non NN working
       simultantously - would allow it to use faster when possible

       Status DW Level 3 Code --- Andrew

      Brower: Andrew is not on the call.  Still a few discrepancies
        between his and Robert's code to be resolved.

       Mobius DW algorithm

      Brower: We have been having good success with our new
        DW algorithm and will want to implement it on clusters
        and QCDOC.  Kostas and Robert are implementing it in chroma.
      Jung: Taku is showing some interest in a QCDOC implementation.
 
   4.) Register usqcd.org -- Robert E.
  
      Brower: We should register our new name.
**    Watson: Can do.
   
=====================================================================
Committee conference concluded at 4:15 PM EDT.  
Next plenary call October 7 at 3 PM EDT.