Software Coordinating Committee Conference Call March 10, 2005 1:00 PM EST Recorder: C. DeTar Present: DeTar, Brower, Mawhinney, Osborn, Renner, Petrov, Pochinsky, Watson, Holmgren, Simone, Efstratios Efstathiadis (Stratos) Edwards, Jung Absent: Maresca, Zhang, ** action items ===================================================================== Osborn: We need CVS and wget on the qcdhost machine. ** Stratos: I'll pass it on 00.) News from UK Stratos: Balint is working on the Chroma code and will use it at BNL with the new QOS. They are running at a lower clock speed 400 MHz and trying a lower cooling water temperature. Single Anacapa box per crate, so two NFS servers per rack. They are running diagnostics and checking reproducibility. They are happy. 0.) Status of US-QCDOC Construction: When will hardware be ready? -- Bug reporting Webpage? Mawhinney: The site is ready to go. I need to set up the categories still. https://sfwcts.phys.columbia.edu [Bob gave a demonstration of the system] ** I need everyone to send their Unix username:password combinations so I can paste it in. We have used this sytem for several years on the QCDSP. Brower: Don and Chip, would this system work for you? It would be good to have only one place to report problems. If we have separate bug reporting pages, there should be links to those sites. QCDOC/BNL status Mawhinney: All 12 racks are in place. 45 MBs are at BNL 49 MBs are promised by the end of this week. First 1000 nodes were put together Thursday. It takes 2 - 4 days to do basic tests on 1000 nodes. There are enough for 4 - 5 of these. Beyond the basic tests are burn-in physics tests. That may take a couple of weeks. The alpha users should be able to get on the machine soon (week or two, but it will be rougher than what they have seen so far.) The second 100 MBs are supposed to be arriving in two weeks, due to a delay in the Woburn plant. Brower: From previous experience, how long would it take after the initial burn in? Mawhinney: We have to weed out problem nodes by gradually increasing the clock speed. It is hard to predict. Some troublesome nodes may take a couple of weeks. 1.) Progress on Asqtad Level 3 tests/interface to MILC? (Dru, Chulwoo and Carleton) Renner: I now have the double precision inverter files. Jung: I am still working on tests of the odd number of site and then single precision. Renner: I have modified the MILC code so it will interface with Level 3. I also have the MILC/QDP version running an got the expected better performance. The next step is to slot in the double precision Level 3 inverter. Osborn: We should be able to run qsetup and get the compatible. Jung: We have to be careful about not compiling the code against one OS version and running against a different one. Mawhinney: The alpha users should recompile their codes to the most recent versions. Osborn: We should be able to choose the version as we do at FNAL. Holmgren: At FNAL on login you inherit the default setup. You can get a different version when you run qsetup. We can set up the syntax so it is compatible with JLab. DeTar: What about Asqtad fermion force? Jung: The CPS version has been checked against MILC. See the CPS information page. Start from http://qcdoc.phys.columbia.edu. The code is there. DeTar: I'll speak with you off line about this. I have written a multidirectional QDP version of the fermion force and would like to compare it with yours. 2.) Progress for Chroma on QCDOC (James and Robert) Edwards: We are ready 3.) QIO Petrov: What are the restrictions on the number of machine dimensions specified in the file conversion? DeTar: The number of machine dimensions must be no greater than the number of space-time dimensions on the lattice. ===================================================================== Committee conference concluded at 12:30 PM EST. Next plenary call March 17 at 1:00 PM EST