Hi,
In order to sell our new library to everybody, especially the
integration of charge.C feature into the library, I made the following
improvement on the head version of BlastLib2.
in additional to save all the ntuples to file before the whole run is
crunched, so that people can look at them and do analysis. charge file is
generated along with the autosave of ntuples.
in detail, for every 5000 triggers that are crunched, (every new line of
those dots), all lr/flr/dst ntuples will be saved to correspondent files.
Since the save is done by ROOT command Autosave, the ntuples are written
to disc with a root key therefore they are readable already. In the
meantime, charge-####.chg is generated too. the charges logged therein are
the integrated charge from the first physics event to the last event that
is saved.
charge file will be repeated over written whenever ntuples are saved.
with this setup, full analysis is possible well before the entire run is
crunched. This is highly desirable when crunching a single run now takes
quite a few hours.
again, a quick check shows that version 2.96 currently handles 6-7
triggers a second, the new version does >=8 triggers a second on spuds.
the new one also does flr and charge, so all in all, it could be a 25%
save of crunching time.
I also agree that ep-elasitc cut is more volatile than the library in
nature. the inclusion of it into the library is only for the preservation
of format of charge file which has in its last line a Elastic Event number
entry.
We can discuss to move this elastic ep counter feature out of charge file
and the library.
Chi
This archive was generated by hypermail 2.1.2 : Mon Feb 24 2014 - 14:07:30 EST