DUNE DAQ Core Software Meeting

Wednesday 29 May 2024, 09:00 → 10:20 US/Central

Kurt Biery (Fermilab), Roland Sipos (CERN)

Zoom:  https://fnal.zoom.us/j/91792149723

09:00 → 09:20 Round table on activities

• Appfwk Review status and next steps
• OKS
• TPG
• Perf. Evaluation

Speakers: Danaisis Vargas Oliva (University of Toronto), Eric Flumerfelt (Fermilab), Kurt Biery (Fermilab), Matthew Man, Roland Sipos (CERN)

AppfwkReview-followup.pdf

Performance_team.pdf

Appfwk: Material reflects the separation of self-contained independent tasks that can go on with marginal effort. Documentation is critical (with proposed structural changes). 

OKS: No news, more details tomorrow at OKS meeting.

Perf.ev: Some struggles with kernel isolation, to be followed up. Move of 002. 

TPG: Many developments and updates. We will hear an in-depth presentation on an upcoming DUNE DAQ General Meeting.

09:20 → 09:40 Developments

Planned and ongoing development on:
- EHN1 operations
- Appfwk
- TPG

Speakers: Eric Flumerfelt (Fermilab), Kurt Biery (Fermilab), Roland Sipos (CERN), Wesley Ketchum (Fermi National Accelerator Laboratory)

A few notes from Kurt

• Are there action items that we should be (and/or are) working on for the issues that have been seen recently at EHN1?
  • lost Fragments when the number of time slices and data sources is large (long-window readout)
  • TP troubles into or out of the tp_datahandler?
  • issues with the SkipListLatencyBuffer noticed in CRT work?
  • PDS folks seeing repeated trigger_times in successive TriggerRecords (with 10+ Hz)
  • reiterate on high rate runs to look for dropped packets (request handler threads are not named)
  • other?
• Where do we stand with np04-srv-005?  performance, aliases for data disks
  • William:  still need aliases; need to check that a reboot doesn't adversely affect the RAID
  • it would be great to have a future talk with information about the performance, etc.
• Why isn't TRACE set up when one runs dbt-setup-release fddaq-v4.4.1-a9?
• Mention that libusb is now required on DAQ computers...
• what else?

Eric: branch in iomanager, two connections to the same channel

Ivana: setting the TPG algorithm per plane (coming soon); AbsRS algorithm and its configuration, hard to configure, bug in configuration (scale factor, memory factor); trying to avoid floating point operations in AVX.  Would like to retest after changes are included.   Accumulator value in TPG to make emulation more predictable.

Marco:  Configuration for DAPHNE should be done.

09:40 → 09:55 AOB

• other ongoing activities
• request and questions
• next week's agenda

Speakers: Kurt Biery (Fermilab), Roland Sipos (CERN)

Next week's agenda:

1. Appfwk Review follow-up
- Documentation updates
- Independent task updates
- Short technical discussion on ActionPlan and ThreadedDAQModule

2. Performance evaluation activity status
- Gantt updates
- 005 results
- 400Gb RU updates

3. Developments
- EHN1 related
- Framework and other developments