Software Coordinating Committee Conference Call January 25, 2007 1:00 PM EST Recorder: C. DeTar Present: Brower, DeTar, Levkova, Gottlieb, Watson, Mawhinney, Zhang, Pochinski, Scholz, Khoriaty, Holmgren, Simone, Osborn, Joo, Fowler, Absent: Efstathiadis, Clark, Renner, Jung, Edwards ===================================================================== ** Action items -1. BU Wiki Brower: http://twiki.org/cgi-bin/view/ It doesn't support LaTeX, but a LaTeX'ed document can be uploaded. I am still evaluating it. Holmgren: We don't allow you to log in to the FNAL Wiki, since you could modify files at FNAL without a Kerberos login. With this Wiki It's too easy to lose content through loss of organization. Brower: We would initially have it for the Software Committee, but open it read-only to the rest of the community. 0. Fantastic! usqcd.org NOW LINKSTO NEW page at usqcd.fnal.gov (Don, Chip ????) Holmgren: It is still reached by redirection, but the new page is up. We are ready to go. Watson: We are also ready to go. 1. Visualization (Massimo if he can is able to make the call) [ Massimo was not on the call. ] 2. Workflop document (prototype?)Wiki page: http://lqcd.fnal.gov/workflow/ (Jim, Don) Simone: We are working with IIT. We are still evaluating options. Please see the main workflow project pages: http://lqcd.fnal.gov/workflow/WorkflowProject.html Under "OngoingWork" there is a link to a document describing the old FNAL ACPMAPS workflow, but that was written in Perl and doesn't have the sophistication we need. Links to workflow related projects are collected here: http://lqcd.fnal.gov/workflow/WorkflowRelatedProjects.html See "WorkFlowRelatedProjects" for a variety of cases we are considering. The workflow requirements document is here: http://lqcd.fnal.gov/workflow/attachments/WorkflowProject/FuncReq.doc [ Jim reviewed the above document. ] Fowler: The GFAX stuff from Indiana will take a vanilla app and wrap it into something griddable. Brower: How portable would this be. We need to be able to run on our machines as well as BG/L. Holmgren: These are layered systems so the API's provide portability. Gottlieb: Would the workflow server run on a single site so you wouldn't need to also run it on an alternate site? Simone: Yes, it could. Holmgren: I would rather focus on just transferring files at first. Fowler: At RENCI we put in failure analysis and fail-over options to make things more fault tolerant. Joo: it would be good to be able to detach the system and let the project keep going. Simone: Also we don't want to the system to depend on being attached to a GUI. It should run in background and be possible to reattach a GUI later. Joo: Under data provenance, do I need to specify a schema for each file? Holmgren: This is used for determining whether a step is complete. "Schema" here is abstract. You will find an example lqcd workflow and how such a workflow might look in karajan here: http://lqcd.fnal.gov/workflow/EvaluationOfWorkflowLan Simone: I am using our heavy-light project as a test case to evaluate some systems. There is a cogkit example xml document for the Karajan system. ** Simone: You are all invited to go through examples and comment. Joo: I have been looking at Triana for building software. Fowler: We have a couple of workflow projects for benchmarking including code building. It captures output from the builds and the test runs and sends it to a database. We are working on examples using quantum chemistry. ** Simone: Rob, it would be good to have a workflow spec for compilation so we can add it to our document. [ The rest of the agenda was postponed. ] 3. News on Asqtad/RHMC on QCDOC? (Chulwoo, Carleton) 3. News on performance on DW Level 3 codes for BG/L (Andrew, Chulwoo, James,Balint) 4. Plan for miminal CRE implementation (Chip/Balint, Don, Statos) 5. When do we start to prototype and design API for multi-core? (Robert,...) Committee conference concluded at 2:00 PM EST. Next call Feb 1 at 1:00 PM EST ======================================================================