present: Milo, Enrico, Roland, Eric, Frank, Giulio, Francesca, Abdallah
EHN1 integration:
- no update on WIB-FELIX transmission errors
- 100 Gbps Mellanox card installed in FELIX host, and can be seen properly
- but the actual network connection is yet to be tested
WRM:
- (see slides)
- WRM is an ASIC that should be able to do fast feature recognition, and could be very interesting to DUNE
- scope for now is limited to indicating, on a frame-by-frame basis, whether the data contain more than noise
- there are a few aims:
- validate the technical feasibility of pushing 60 Gbps of data to the PC hosting the WRM and its associated FPGA
- test the algorithmic aspects on ProtoDUNE
- see how this project can fit DUNE
- Q1: can FELIX (in ProtoDUNE) provide sufficient throughput?
- not easy to see that this will work with the full rate (which will mean duplicating the output from felixcore?)
- but it should be OK to run short tests sending data to WRM instead of to the FELIX BoardReader
- somewhat lengthy discussion about how the data come in (subscription to 10 NetIO connections?), and whether the data will be strictly ordered in time
- Q2: where will the output go?
- in ProtoDUNE, it seems most appropriate to write a separate BoardReader, so that the output data can be included in event building
- discussions are ongoing concerning the DAQ/trigger strategy in DUNE. It would make a lot of sense to talk to the DAQ Consortium at large (h/w wg?) soon
- agree to exchange more information on the technical aspects, and tentatively aim for this to be presented on January 4 to the h/w wg
Conferences:
- proposal is to submit an abstract to both CHEP and IEEE
- Frank, Roland and Enrico will write an abstract
- abstract for CHEP is due before the end of December
- there will also be an ATLAS FELIX abstract submission to CHEP, need to make sure not to clash
Next meeting: tentatively planned for January 10, 2018
There are minutes attached to this event.
Show them.