present: Enrico, Roland, Eric, Milo, Frank, Junbin, Kai
EHN1:
Roland: BR committed to the repository. Check EHN1 minutes for commit details.
- BR application: implementation in artdaq v3 done, tested with full internal FSM (start/stop/shutdown)
- trigger matching added, works in emulation mode (should be straightforward to have it work with real timing information)
- could e.g. start tomorrow with "data quality" tests -- even if the throughput issue will limit these to several seconds' worth of data
- focus for the near future is on improvements to the throughput
- not far from the required (single-link) throughput: 6 Gb/s instead of 7 Gb/s
- now working with ATLAS FELIX experts (especially Jörn) on improvements. Already using some packet coalescence but can push this further
- using Infiniband as an alternative is still on the table, however, it comes with some side effects (e.g. felixcore crashes) that would need to be debugged, and is not the first priority
- Jörn is now on vacation, expect effort to resume shortly
- apparently the option of having a buffer per input link is being discussed -- this ought to help us as well
BNL:
- (see slides)
- nice diagrams explaining also channel counts etc.
- full hardware setup at BNL, including FEMB, WIB, FLX709
- see summary slide for next steps
- discussion:
- on the topic of developing code to analyze the FELIX data: Milo pointed out that such code exists already
- it is thought to be very helpful for BNL to be able to help with the data validation
- from-host GBT mode communication relevant for us? No, but used in ATLAS
- according to Roland this functionality may actually be removed in order to obtain required to-host throughput
Next meeting: tentatively planned for April 4
There are minutes attached to this event.
Show them.