GlueX Offline Meeting, May 4, 2010
GlueX Offline Meeting
Tuesday, May 4, 2010
2:00 pm EDT
JLab: CEBAF Center F326/7
To connect by telephone:
- 800-377-8846 : US
- 888-276-7715 : Canada
- 302-709-8424 : International
- enter participant code: 77438230# (remember the "#")
- ESNet: 8542553
- Meeting URL: http://evo.caltech.edu/evoGate/koala.jnlp?meeting=eleBevvlvDatasaeIl
- Phone Bridge ID: 56083
Talks can be deposited in the directory
/group/halld/www/halldweb/html/talks/2010-05 on the JLab CUE (you have to be a member of the "halld" Unix group). This directory is accessible from the web at https://halldweb.jlab.org/talks/2010-05/ .
- Project Management: Offline Computing Change Request: re-do
- Review of minutes from the last meeting: all
- Collaboration Meeting Agenda: Mark
- Tests of a SIMD version of DVector3: Simon
- Action Item Review: all
- IU: Ryan Mitchell, Matt Shepherd
- JLab: Craig Bookwalter, Mark Ito (chair), David Lawrence, Simon Taylor, Elliott Wolin, Beni Zihlmann
- Mark reported that the change request, which was recently submitted to the project, for Offline Computing will have to be redone. The current version references last year's baseline BIA schedule. The request must reference the current revised schedule.
- Elliott told us that Yelena Prok from CNU has summer students looking for projects. He would like to get a feeling for our need for students by the end of the week. Please contact him if you have ideas.
- On a related note, David expressed concern that our plan to assign work to a new collaborator on the calibration database might result in a product incompatible with the current framework. He agreed to write a specification for the system to guide prospective work. Mark agreed to set up a meeting with David and Eugene Chudakov to discuss this issue.
- David has successfully run our reconstruction on MacOS 10.6 (HDGEANT is still problematic on this platform). Changes to support this have been checked in on the trunk.
- On a topic possibly related to the offline, Elliott reports that Labview talking on EPICS appears to work.
- Beni reported on a new system to edit an existing relational database, using MySQL and PHP.
- Elliott urged caution using this combination; he has heard reports that it is vulnerable to cyber attack.
Review of Minutes from the Previous Meeting
- We agreed that Mark should make a new release of sim-recon.
Collaboration Meeting Agenda
Mark walked us through the Offline and Physics section of the agenda for next week's Collaboration meeting. This agenda was developed consultation with Ryan, David, Simon, Elton.
Tests of a SIMD version of DVector3
Simon described his recent re-write of the DVector3 class to take advantage of the SSE hardware present on many x86 processors. See his wiki page for details. Speed increases of a factor of two to three were seen for most member functions. The goal is to increase the speed of charged particle tracking.
Items from the discussion:
- Matt mentioned trying the Intel compiler. It takes advantage of the SSE hardware functionality and does not require custom code.
- Craig argued against investing too much time in optimizing matrix operations before consulting experts in the field. A lot of research has been done in this area.
- Mark noted that optimization might be different for different choices of our standard scientific library.
- Ryan wondered if there was a way to save the results of tracking so that work devoted to that step could be re-used while working on other parts of the reconstruction. Elliott noted that the DANA-to-EVIO package could be modified to save tracking information. Mark pointed out that unfortunately, currently data written in EVIO format cannot be used as input to the reconstruction code without significant development (although Elliott contested the word "significant" in this point). David agreed to modify the data model and the relevant input/output routines for HDDM to accommodate summary information from charged particle tracking.
Action Item Review
We went over the list.
- Elliott volunteered to install an instance of Mantis for us to try out (after the Collaboration meeting).
- David reported that the script to run the "b1pi" analysis is done. Mark will incorporate it into a weekly cron job.
New Action Items
- Write a specification for the calibration database. -> David
- Arrange a meeting with Eugene to discuss database work. -> Mark
- Make a new release of sim-recon. -> Mark
- Write out summary tracking information in HDDM -> David
- Run b1pi script in a weekly cron job. -> Mark
- Install Mantis -> Elliott