You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
prescales events based on (( e.event() % ps ) == 0), where ps is the prescale factor. This works fine for MC eventsi for which events are are in order and there are no missing event numbers; that is the environment in which it has been tested so far.
In the trigger farm most process will see about 1/800 of the events. Moreover the distribution of event numbers across processes will have short term structure. This could lead to unbalance rates of prescalled events coming out of each processes. It's possible that the structure is so short term that the imbalances will be short lived enough that we don't care. Be we don't know that with any certainty.
Does it make more sense instead to prescale on )( _nevt % ps ) ==0 ) where _nevt is declared at:
Trigger/src/PrescaleEvent_module.cc , at line:
Offline/Trigger/src/PrescaleEvent_module.cc
Line 92 in c4a4b22
prescales events based on (( e.event() % ps ) == 0), where ps is the prescale factor. This works fine for MC eventsi for which events are are in order and there are no missing event numbers; that is the environment in which it has been tested so far.
In the trigger farm most process will see about 1/800 of the events. Moreover the distribution of event numbers across processes will have short term structure. This could lead to unbalance rates of prescalled events coming out of each processes. It's possible that the structure is so short term that the imbalances will be short lived enough that we don't care. Be we don't know that with any certainty.
Does it make more sense instead to prescale on )( _nevt % ps ) ==0 ) where _nevt is declared at:
Offline/Trigger/src/PrescaleEvent_module.cc
Line 65 in c4a4b22
The text was updated successfully, but these errors were encountered: