Developer's meeting

Wednesday 23 Nov 2016, 16:00 → 18:00 Europe/Zurich

Summary of the "data flow" discussion from last Monday's DAQ meeting (Giovanna):

• Kurt's summary now available as https://indico.fnal.gov/getFile.py/access?contribId=13&resId=0&materialId=minutes&confId=13369
• items relevant for the FELIX BoardReader:
  • need to be able to handle the situation that the trigger arrives before the data
  • need to send empty fragments when no data available
  • will drop data that are "never" requested
• "single BoardReader per card" requirement not to be taken too strictly

Data compression studies (Milo):

• still struggling with the fact that reverse-engineering the data format is not simple (especially with zero-suppressed data)
• Giovanna: Kurt offered to take care of the reverse-engineering. Discussion (with Milo) not quite finished as yet?
  • should talk to Kurt again
• Giovanna: another option is to use the 35-ton data

Running the artdaq demo (Frank, Kevin):

• Frank has built and run the demo successfully under SLF6.7, but is struggling to do so under SLF7 or CentOS7
  • Kevin has posted instructions here: https://github.com/FELIXDAQ/ARTDAQ-Instructions/wiki (NB: work in progress)
• Eric has built and run it even under SLF7, and has also been able to make it work with HTG710 fake DMA fragments
• Kevin has created several repositories under https://github.com/FELIXDAQ/
  • these will be moved to gitlab.cern.ch once Kevin, Eric, and Lynn have CERN accounts
• Giovanna: artDAQ should (be able to) run under CentOS7; issue to be raised with Kurt

Data format (Giovanna):

• identification of data to be done using full timestamp in raw data
• now a revised proposal (also, stream ID will disappear); Eric Hazen will update his DocDB-1701 note to reflect this

17:00 → 17:20 PNNL contributions

Speaker: Lynn Wood (Pacific Northwest National Laboratory, USA)

PNNL_FELIX_status_23-Nov-2016.pdf