Software Coordinating Committee Conference Call
		      April 14, 2005 1:00 PM EST
			  Recorder: C. DeTar

Present:
   Brower, DeTar, Maresca, Mawhinney, Renner, Osborn, Holmgren, 
   Pochinsky, Zhang, Jung,

Absent:

    Efstathiadis, Edwards, Eric Blum, Simone, Watson,

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

QCDOC/MILC memory problem

   DeTar:     Dru ran into a memory problem when he moved from
              the 64-node mother board to the 1024-node rack.  
   Mawhinney: The Columbia MoBos have 256 MB DIMMs, whereas the
              BNL racks have 128 MB.
   DeTar:     Chulwoo was telling us that after the OS is loaded
              we are left with only 32 MB out of 128 MB.
   Renner:    I found that a  40^3 x 96 lattice with basic MILC + Level 3 
              fits with some tweaking, but 48^3 x 144 does not.
   DeTar:     Can we do anything about the QOS extravagence?
   Mawhinney: We will try to recover more memory from the QOS.  I will 
**            speak with Peter Boyle.
   Jung:      There are two heaps, so we may be able to play with them
              recover some memory.
   Mawhinney: I don't understand why the MILC code requires so much
              memory.  The lattice is 1.7 GB and with 32 MB per
              processor and 1024 processors, that's 32 GB.
   DeTar:     I don't either. We need to investigate more.

   Brower:    We worry about inverting the DWF on the 40^3 x 96
              lattices.
   Mawhinney: It should fit just fine.
   Jung:      I can change the malloc call to report the amount of memory
              allocated on each call for diagnostic purposes.
   DeTar:     If our own method doesn't work, I'll ask you for yours.

QCDOC Rack testing:

   Renner:    I hope I can use the rack first to be sure the MILC code
              is at least working before running the test suite.

   Brower:    Is there another rack available for James?

Upcoming project review

   Brower: On May 24 there is a DOE project review.  Documents will be
           needed some weeks prior to the review.  It would be good to
           be able to report successful physics running.


=====================================================================
Committee conference concluded at 1:45 PM EST.  
Next plenary call April 21 at 1:00 PM EST