HLT_DA_status071107 - RHIG AT YALE

boorishadamantΤεχνίτη Νοημοσύνη και Ρομποτική

29 Οκτ 2013 (πριν από 3 χρόνια και 8 μήνες)

66 εμφανίσεις

Current status of
PHOS Online
Processing

Øystein Djuvsland, Per Thomas
Hille.

07 Nov 2007

Requirements: Calibration data
during ordinary runs


Energy distribuiton for all channels


35840 histograms


Update frequency: All events (1 KHz)


HLT or DAQ


TOF distribution for all channels.



35840 histograms


Update frequency: All events (1 KHz)


HLT or DAQ


Accumulated energy for all channels,


one histogram.


Update frequency: All events (1 KHz)


HLT or DAQ


Live channel map


one histogram


Update frequency: All events (1 KHz)


HLT or DAQ


Summary:


71682 histograms must be updated at 1 KHz


Status: Implemented and sucessfully tested in HLT



Requirements: Pedestal
calibration run


Average Baseline measurements


35840 values


35840 histograms


HLT or DAQ


RMS of pedestals


35840 values


35840 histograms


HLT or DAQ


Distribution of per channel mean pedestals


35840 Values


35840 Histograms


HLT or DAQ


Frequency


Can be adjusted to available CPU power

Requirements: Data Quality
Monitoring


The lesson learned from real data taking is that the quality of the data has to be
verified on event by event basis for each individual channel. It is not enough to
base the DQM on statistics accumulated during calibration run.


Some channels is noisy for all events.


Some APD are bad and give signals that is not due to any external events
(breakdown), however


These channles also measure physical events


One has to discriminate between pysical and non physical events channel by channel
and event by event.


For some channels there are bit flips during readout.


Should be corrected online in HLT prior to analysis by intelligent algorithms.


Some channels has corrupted or bad signal for some events but are ok for
others.


Summary:


The DQM has to run in HLT because there are no mechanism to correlate corrupted
single events between HLT and DAQ.


Status: Intelligent DQM algorithms has been implemented and sucessfully tested in the
HLT.


Requirements: Calibration data
for cosmic runs


The requirements for Energy, Live channel map and accumulated energy
is simlar too those for Ordnary runs but the updtae frequency is less, only
20


100 Hz.


Data compression & selection.


Selection of signals above noise.


HLT


Discarding bad data


HLT


Triggering on Isolated MIPS


HLT


Status:


The Online Data selection and compression has been implemented and
successfully tested online in HLT.


The comsic MIP trigger has been successfully tested offline on HLT output data.

Preliminary conclusion


Calibration algorithms for ordinary runs and
baseline calibration runs can run either in
HLT or DAQ.


Calibration algorithms for cosmic calibration
runs has to run in the HLT.


Needs more CPU time than can be provided by
DAQ.


DQM has to run in HLT

Calibration DAs for ordinary
run: HLT vs DAQ discussion


Case 1: The DAs run in the HLT


The data has to be made available for analysis
(decoding etc..) regardless of where the DA is
running. The overhead of running the DAs in HLT
is relatively small compared to the CPU time
needed for reconstruction.


Case 2: The DAs run in DAQ


Decoding, mapping etc... Has to be done twice.
Once in HLT and once in DAQ

Calibration DAs for calibration
run: HLT vs DAQ discussion


The DAs can be run in either HLT or DAQ

Some real life examples

APD in continious breakdown mode

Noisy APDs due to missing HV

Bit flip during readout

Cosmic event

Pi0 ?

HLT calibration display