User name R. Michaels
Log entry time 05:03:28 on April30,2000
Entry number 40282
keyword=Online ESPACE (using ET)As I told the ESPACE community (halla_espace@jlab.org),
I've implemented ET system to obtain data online in real-time.
Below is a plot showing how the observed rate of ESPACE
analysis varies with DAQ rate. Most of the data was taken
last night with semi-realistic simulation of DAQ (mostly random
scint trigger but no beam). The circles shows a prediction based
on adding the latencies of DAQ and ESPACE. A couple curious
features: 1) Around 500 Hz, there is some extra latency presumably
from ET; and 2) Above 1200 Hz, the observed rate over-shot my
prediction. But this was probably because I used T8 to bring the
rate above 1200. Those triggers are 'empty' and I guess ESPACE
ran faster.
Note: ESPACE runs on a datafile at 150 Hz on adaql1, the max in
the curve. It is slow because of a lot of histograms.
A rule of thumb emerges: For rates above 1kHz, for which running
ESPACE on the current data file would cause deadtime, it pays to
run ESPACE using ET system. It will run about as fast as on a data file.
But for lower rates, it is better to run ESPACE on a data file. By the
time you start ESPACE you've accumulated some data, and results
will come quicker. Below 1 Khz, reading a datafile causes no deadime.
FIGURE 1
