Lessons learned from large number of modules at Liverpool and Oxford have been incorporated.
Transparancies describing Release 4 at [[IDWEEK]]
What was proposed for this release
comments added 12/2/2005 by AlanSome of this will now move to SctRodDaqRelease5?
The focus should be the archiving and some of the performance increases. Robustness and control features should also be considered and of course, any user feedback.
- Complete archiving system including a database, retrieval support from GUI, re-analysis support
- reanalysis available, retrieval available, DAQ database in progress
- Greater robustness of FittingService, AnalysisService and ArchivingService by supporting a "recover from crash" operation, so that they can pick from where they left off if something happens.
- Done for FittingService - not yet Analyis or Archiving
- Review of the control logic (may be necessary for the above anyway)
- still required***
- Review of status logic - e.g. status of Scan, Test, Sequence. On a per-module basis?
- still required - and the end of the RollerCoasterAnalysis? required
- Remove ROOT:
- Think carefully and define a 1d data format
- Add support for this format to the Fitter
- Add evaluation support to FitObjects?
- Define a 2d data format for raw data
- Write a Projector class and an Occupancy class (rather than the single OccupancyProjector?)
- Migrate all TF1 objects to FitObjects?
- Replace any other ROOT objects that may be used (e.g. TGraphs - should be relatively easy)
- still to do
- Add support for binomial log-likelihood fitting (shouldn't be too hard)
- Allow multi-threaded apps (as in thread safeness? don't most things have threads already?)
- A seperate RunController - should be a seperate package, and probably linked to SctApi. (This is probably essential for RCD and TestBeam?)
- CalibrationController is now a RunController
- As a minimum, rename all the binaries (and the RunController) so they won't conflict with other sub-detectors
- Greater DCS integration
- largely done
- Move to online-00-21-00. Dependant on DCS
- DCS have requested this as a matter of priorty so this is now happening in a branch of CVS. Currently ChrisLester and AlanBarr working on this. See OmniMigration.
- actually moved to 00-22-00
- Read out, display and analyse the nhits/event histograms
- done
- Change queues from FIFO to FILO - this should give a 10-15% performance improvement with large numbers of modules.
- perhaps may be confusing have instead added ServiceStatusGUI
- Support multiple analysis chains, or allow load balancing and multiple services.
- Alternative is to migrate some code to the ROD
- Currently long term, though fitting has been thought about a little
- This loses a lot of flexibility so definitely shouldn't be exclusive
- multiple analysis chains possible through different arguments to services not tested
- Alternative is to migrate some code to the ROD
- Alternative to NFS - needs discussion with Online Software people.
- Binary version of IS is apparently imminent
New Scan / Test List
- none of these added yet
- AccumulateTest? - quick test of the accumulation mode of ABCD (used during xray alignment)
- Scan mask (the 01 and 10 settings) and see if the two together comes out with 100% occupancy?
- Set threshold to a little above 50% and histogram 1 trigger per bin? Time between bin will vary with number of modules though...
- Should be possible currently to run a scan like this, unless it requires a fixed time between triggers
- DutyCycleTest? - measure duty cycle of clk/2 signal returned by module, send single bit to flip BPM phase, repeat duty cycle measurement of clk/2. This information can be used to determine, and hence tune, the duty cycle of the BPM clock. Requires adjustment of fine RX delay. (Something like that anyway...)
- CrosstalkScan? with two triggers in one serial stream
- How is this different from DoubleTriggerNoiseTest? Separation of triggers?
- CrosstalkScan? with single triggers in each of two streams
- Also referred to elsewhere as Dual Triggers