[Halld-trigger] L3 meeting minutes posted
davidl at jlab.org
Thu Jun 16 10:01:34 EDT 2016
Minutes from last Friday’s L3 trigger meeting are now posted and copied below:
Attendees: David L., Adesh S., Mike W., Cristiano F., Sean D.
David Showed some benchmarking results for the EVIO parsing code with various levels of associated object linking enabled
Mysterious rate drop observed when no linking is done from 12kHz to 8kHz.
Mike noted that they had to empirically determine the optimal number of threads for the LHCb HLT
David will work on mapping our the performance for various settings in the parsing threads and processing threads (benchmarking was always done with both set to 32)
sim-recon algorithm benchmarking
Cristiano continues to work on mapping the performance of various algorithms using janadot
Some discrepancies between real data and simulated data performance
Will look into effect of LED triggers on results (particularly time spent matching tracks to calorimeter clusters)
Sean noted that there are lots of PS triggers in the real data that should be filtered
Adesh is currently running 10M bggen events with incident γ energy down to %pi; threshold (global use sim1 data set only goes down to 3GeV)
Farm jobs expected to be done by Monday morning
Size of REST files produced is about 10 times smaller than "normal"
Justin told Adesh this is expected due to the dominance of low energy events
Wiki documentation is forthcoming
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Halld-trigger