Bug #63
openScaler event count wrong in end-of-run counter summary
Added by Ole Hansen over 7 years ago. Updated over 7 years ago.
10%
Description
I get this with analyzer-1.5.25:
Normal end of file /home/ole/tutorial/data/g2p_3132.dat.0 encountered
End of file
Counter summary:
314292 events read
314292 events decoded
313476 physics events
314172 scaler events
115 slow control events
5 other event types
313476 physics events analyzed
314292 events accepted
314172 is not the number of scaler events, but "all"-"slow control"-"other". There should be 696 scaler events. Looks like a bug.
Updated by Ole Hansen over 7 years ago
This is not a bug in THaAnalyzer, but it looks like it could be one in the decoder. Maybe Bob can help:
THaAnalyzer counts a scaler event whenever fEvData->IsScalerEvent() is true. As implemented in THaEvData.h, this function is true for event type 140 OR if member variable evscaler 1. A comment says that "a scaler event can also be a physics event". I assume that these "dual" events are the ones with evscaler 1.
evscaler is set to 1 in THaCodaDecode::vme_decode whenever the event contains data from at least one ROC that is defined as type "scaler" in the crate map and that has at least 16 bytes of data:
if (fMap->isScalerCrate(roc) && GetRocLength(roc) >= 16) evscaler = 1;
This is true even if none of the scalers in this crate actually have any data. The crate map used for the replay above contains three such crates:
==== Crate 10 type scaler "evright" # slot model clear header mask nchan ndata 1 3801 1 0xceb00000 0xffff0000 32 32 2 3801 1 0xceb10000 0xffff0000 32 32 3 3800 1 0xceb20000 0xffff0000 32 32 4 1151 1 0xceb30000 0xffff0000 16 16 5 1151 1 0xceb40000 0xffff0000 16 16 6 1151 1 0xceb50000 0xffff0000 16 16 7 560 1 0xceb60000 0xffff0000 16 16 8 560 1 0xceb70000 0xffff0000 16 16 ==== Crate 11 type scaler "evleft" # slot model clear header mask nchan ndata 1 3801 1 0xabc10000 0xffff0000 32 32 2 3801 1 0xabc20000 0xffff0000 32 32 3 3800 1 0xabc30000 0xffff0000 32 32 4 3800 1 0xabc40000 0xffff0000 32 32 5 3800 1 0xabc50000 0xffff0000 32 32 6 3800 1 0xabc60000 0xffff0000 32 32 ==== Crate 12 type scaler "ev3arm" # slot model clear header mask nchan ndata 1 3801 1 0xddd10000 0xffff0000 32 32 2 3801 1 0xddd30000 0xffff0000 32 32 7 775 1 0xda00bdc0 0xffffffff 32 2048 9 792 1 0xda00adc0 0xffffffff 32 2048
Models 1151, 560 and 3801 are handled in vme_decode, but 3800 is not implemented. (What is it?) 775 and 792 are a TDC and a QDC, respectively, presumably not present in the g2p setup. (Are they?) In any case, it looks like scaler crates can have both scaler and non-scaler modules. (Can they?) If so, then I wonder if evscaler should only be set if an actual scaler module reports data in the event, otherwise THaAnalyzer::ScalerAnalysis will be called for nothing, and too many events with scaler data are reported.
Bob: How do you explain the scaler event count in the above analysis? Does every physics event really have data from scalers in the scaler crates?
Updated by Ole Hansen over 7 years ago
From Bob Michaels 30-Jan-2014:
The simplest answer is that for some experiments, a fraction of our scaler channels were read for every event. The scaler events were evtype 140, occur every 4 sec approx and were typically 200 channels of scalers integrated for 4 sec or so. The scalers in events were evtype 1 - 7 normally, typically at 1 kHz and with perhaps 16 chan. I don't think there were "bugs" but this will get reorganized this year as a byproduct I'd what I'm doing. There are parts of ThaEvdata related to scalers which actually don't get used, and the work done by THaScaler instead. In the future I plan to have a THaScalerEventHandler for the evtype 140, hall C can have their own EvHandler, and if there are "scalers in Physics triggers" they will be handled like another VME module.
Updated by Ole Hansen over 7 years ago
From Ole Hansen 30-Jan-2014:
I guess at the minimum we need to make the functions
THaEvData::IsPhysicsTrigger(), THaEvData::IsScalerEvent() etc. more
adaptable since their behavior is currently hardcoded. Perhaps they
should report bits from an internal bitfield or similar, with each bit
corresponding to a certain event class. The event type handlers then can
set those bits as appropriate.
In your new decoder scheme, do you still want scalers to be analyzed
within THaAnalyzer, using something similar to THaScaler? I suggest we
come up with a universal design for handling scalers, regardless of
whether they are read in "evtype 140" or physics events. It sounds like
you want the "evtype 140" scalers to be handled by an event type handler
and scalers in physics events by something (some "scaler apparatus"?) in
THaAnalyzer? Is that the plan? Why not let the scaler event handler take
care of them as well?
BTW, what type of module is model "3800"?
Updated by Ole Hansen over 7 years ago
From Bob Michaels 30-Jan-2014:
The 3800 is a type of scaler (we have 4 types in Hall A,
each with different properties).
I'll try to come up with some working code for scalers soon
and then we can discuss it.
I think the "scaler events" should be handled by a "event type
handler", while a "scaler module" is a kind of module. The
same module class used by scaler events can be used in
physics events if there are scalers in the event stream.
Updated by Ole Hansen over 7 years ago
From Ed Brash 30-Jan-2014:
I think the "scaler events" should be handled by a "event type
handler", while a "scaler module" is a kind of module. The
same module class used by scaler events can be used in
physics events if there are scalers in the event stream.
That logic makes sense to me.
Updated by Ole Hansen over 7 years ago
From Ole Hansen 30-Jan-2014:
Makes sense to me, too. My point is: there shouldn't be two places in
the analyzer where scaler modules are processed. In fact, maybe the
distinction of "scaler events" could even be dropped. Maybe scaler
events are just like physics events, except that ALL modules read in the
scaler event are scaler modules. Physics events may contain a mixture of
physics modules (ADCs, TDCs etc.) and scaler modules. Data from scaler
modules are always processed by the same code in the same place
(presumably somewhere in the decoder), regardless of the event type.
Would that work?
Also, how do we make scaler data available to the user? If they should
go into a separate tree in the output ROOT file, maybe processing scaler
modules in THaAnalyzer is the way to go after all?