EventBuilderConfigurations.tex

% Event builder configurations

We do not want to build complete events on one node; rather we want to
make sure that the event fractions from each FED is always shipped to
the same filter unit.

To accomplish this the network fabric of the event 
builder will need to be reconfigured for each pixel
calibration run. Note that this is likely to be done
between each LHC fill. (This does not mean that we
will determine all constants at this rate, but that
we will want to monitor quantities such as address
levels frequently.)

Freya Blekman has developed the algorithms that analyze 
the pixel raw data for the purpose of doing gain calibrations,
pixel alive, and Scurves. Initially this work is 
done using data recorded in plain files. To progress
towards the final goal we should first try to run this
with a filter unit that is fed data from a source
that reads files. This will allow us to test and develop
the application in the form it will run in the HLT. The
second step is to actually run this with data produced 
from a FED. This can be either real data of playback 
generated in a DAQ test stand. We have discussed this 
with Franz Meier. Freya will follow up on this in
August and September.

One important question to settle here is if we can generate
the triggers we need for this using our LTC. 

We should also test and deploy code so that we can take 
data using local DAQ (VME) and send it out to an event builder.
Jim Hunt wrote some of this code. This has not been used
recently, but we need to deploy this.