Software Review4 Meeting, September 9, 2016

Present:

• CMU: Curtis Meyer
• JLab: Eugene Chudakov, Mark Ito, David Lawrence
• W&M: Justin Stevens

We met to kick-off planning. What follows are notes on ideas that we should consider including in our presentation, both in talks and in a written document.

Curtis:

• by the time of review, Fall 16 run calibrations should be underway
• the various launches give numbers for CPU and disk usage

Eugene:

• should estimate number of collaborators who can use the software to do analysis

Mark:

• Launches, each type of launch has different aspects that can be emphasized, all give computing resource use data
  • Monitoring
  • Reconstruction
  • Analysis
• REST file distribution a la Matt
  • what is our official plan?
• Conversion to Git
• Work on Analysis Library since last review
• Geant4: where are we?
• Simulation runs for collaboration-wide use
• Profiling: formal program, has not been done
  • Studies done on optimal number of threads to run for particular tasks
  • David: real data processing provides important numbers

Justin:

• we should highlight Physics Analysis Workshop
• representation of user perspective
  • agenda from Vancouver DNP session on GlueX
  • add 1 or 2 of the 10 minute talks to the agenda?

???:

• work on error matrices, tracking, showers

David:

• translation table system needs development for ease-of-use
• Sean has done work recently on EventStore
• Richard has contributed improvements of the HDDM API
  • includes multi-threaded I/O

Curtis:

• JLab and CMU will jointly monitor GlueX data transfers to the university

David:

• CPPSim has full geometry from HDDS, hits from many of the detectors
  • can generate pictures of events
  • in addition to anything that Richard may have

Mark:

• OSG
  • current status vis-a-vis GlueX

Curtis:

• Nathan's Hall D Package Manager (HDPM)
  • [Mark: and Mark's build scripts]
• C++ 11 features have been adopted
• "Paper Review" materials from early 2016 should be included
  • resource usage spreadsheet was not update, maybe now it the time to do that

Eugene:

• comparison of CPU usage for regular vs. high-intensity data taking
• also comparison of wide data window vs narrow-window data for regular intensity data

Curtis:

• What are the "automatic" calibration processes we have developed

Eugene:

• construct a timeline-like presentation of the time required for the various steps on the way to physics results:
  • calibration
  • reconstruction
  • analysis
  • etc.

Action Items:

• Curtis will produce an outline of a written document to present to the committee
• Mark will circulate his notes from this meeting
• David will talk to Rolf about how much we should be emphasizing online data acquisition in general and high-intensity running in particular.
• We will meet again in a week. Same day and time.