"Aha" I exclaim, "there's the problem!"
I get Alexandre to make a new peds file from run 3266 and install it, take a new run -- and get the same 31% dead-time as before (no difference at all)...
I really expected *some* change -- can someone confirm that pedestals really are being suppressed for ALL crates with ADCs? I know it used to work, did the ped-file generating code get borked so now peds aren't loading correctly (but there are no errors)?
Alexandre and/or Sergey: please confirm correct pedestal suppression operation for all ADCs and write a halog entry (ie. look at GeN config run vs. GeN_ped run -- is it working the way you think it's working).
Supporting the borked code theory is the fact that the ped-file
generation script also created non-zero pedestals for bbvme1 (ROC23,
which has the v792 ADCs). That would be fine except that the v792s are
utterly disabled in the crl (and in the resulting C code -- I checked) --
so there are no v792 ADC data in the recorded datafile. Is the analyser
not zeroing some array properly?
A copy of this log entry has been emailed to: feuerbac@jlab.org, camsonne@jlab.org, aserg@jlab.org