Speedup and cleanup of LArDigitalTriggMonAlg
As reported in ATLASRECTS-7798, the monitoring algorithm LArDigitalTriggMonAlg
is horribly slow. This MR brings the execution speed down to about 100ms/event, mostly because the call to fill
was pulled out of a tight loop over all supercells
On top of the speedup, the code was streamlined and lots of dead/unnecessary code was removed.
I expect the histogram output to change because the current version had at least two obvious bugs:
- The event-size counter started with -1 instead of 0
- The
LArOnlineID
helper class (for regular readout) was used instead of theLArOnline_SuperCellID
helper
There are a few more suspicous details we should address, therefore this MR remains a draft for now.
cc @ekay