Software Coordinating Committee Conference Call
		       July 7, 2005 1:00 PM EST
			  Recorder: C. DeTar

Present:

	DeTar, Maresca, Holmgren, Simone, Pochinsky, Zhang, 
        Efstathiadis, Osborn, Brower, Jung, Edwards, 

Absent:
   
   Mawhinney, Renner, Sommervoll
   Petrov, Gottlieb, Watson 
   
    
** action items
=====================================================================

1.) Status of US QCDOC Hardware -- QOS 2.6.0/ Disk Array/etc.

   Jung:  The new 4K still being shaken down -- weeding out bad nodes
          8 8 8 2 2 2 cabling
          The older 4K will soon be split.  Delayed by staff 
          vacations.
   Edwards: What is the status of QOS 2.6.0?
   Jung:   I have put a new version of qmp that runs under 2.6.0 under
           /qcdoc/sfw/qos/devel/v2.6.0-CJ/aix5.2f
   DeTar:  Should we be trying 2.6.0 now?
   Jung:   Yes. I have recompiled the level 3 inverter also.
   DeTar:  Should we expect bit-for-bit identical results?
   Jung:   Yes.
   Edwards: Should I not be messing with the .qrb?
   Jung:   It is best to run "qset_reset_boot -w file" to get the
           the default qrb file appropriate for the machine.
           Then you can edit it.
   Edwards: What about gmp (Gnu multiprecision)?
   Jung:  It is installed./qcdoc/sfw/packages/gmp

2.) Code Optimization and Production plans  Asqtad on QCDOC
	     (Dru and  Carleton)

   DeTar:  We talked about three approaches.  Coax the xlc compiler into
           emitting efficient assembly code for the gcc compiler.
           Next, to take the CPS single mass term and write a wrapper
           as the inverter was done.  Third, we could try Bagel.
   Brower: It would be good to find out how UK is doing it.
   DeTar:  I can contact them.
   Jung:   I don't think they will be using anything other than
           CPS. So there may be nothing new there.
   Edwards: Bagel-QDP was used to create vectorized QLA instructions.
   Brower:  How can we tell what Bagelized routines are available?
   Edwards: They are on JLab CVS.
   Osborn:  The new gcc compilers have a build-in prefetch.

3.) Code Optimization and Production plans  Chroma on QCDOC
            (James and Robert)

    Edwards: Chroma is running at about the same speed as CPS
       with DWF.  Of course, there are many shared routines.

4.) ILDG compliant File Archives at BNL/FNAL/JLab

    Brower: I'll be talking to you, Don, about the US
      status of ILDG archiving.  I am preparing a presentation
      for Lattice 2005.

5.) Update of  Chip's Run Time Env task list.
        
	http://www.jlab.org/~watson/lqcd/RTEworklist.htm

        Pairwise tests by Steve Gottlieb?

    [ Chip is away traveling.  Steve sent remarks separately. ]


=====================================================================
Committee conference concluded at 2:15 PM EST.  
Next plenary call July 14 at 1:00 PM EST