CERN QMUL
L1Calo Software Video Conference at RAL/HD on 17 December 2001
Present: Bruce, Eric, Gilles, Murrough, Norman, Oliver, Steve, Tony.Actions from previous meetings
- Murrough and Bruce made a first integration of the run control prototype with the module services developments. A lot more work needs to be done...
- Other actions are carried over...
Oliver: progress at Heidelberg
Oliver reported on some hardware developments at Heidelberg: the ASIC is nearly, nearly ready for submission, the MCM will need a thermal redesign for the higher than expected ASIC power consumption.On software, Karsten Penno has made progress in programming the video RAM. Some synchronisation problems observed will hopefully be fixed with a new video driver. Oliver is starting work on the RemFGPA receiver station. He expects this work to continue into January. From February through to April he expects to work on implementing HDMC parts required for the hardware which will arrive soon, eg MCM test board, etc.
Oliver has also installed the Together tool in Heidelberg and has found it useful in analysing the existing HDMC code base. He was encouraged to put the Together analysis up on the web.
Bruce: module services developments
The run control and module services integration test showed up some problems with the direct part dependencies created by the L1CalDaq "inheritances". These have now been changed to use the SubModule part (which didnt exist when the first version of the inheritances was developed). Bruce has also worked on extending the syntax, and it has now been renamed as "composites".He stated that he planned to upload the changes to the HDMC CVS repository and update the documentation before christmas. Part of this work includes sample code and parts files which will serve as an example of using the composites syntax.
He remarked that the present scheme for binary and library directories (eg bin/Linux) isnt sufficient to cope with having different compiler versions on the same shared filesystem. The Online software have a more extended scheme and we should move to something similar for HDMC.
Gilles: CPM parts files
Gilles has been creating the parts file for the CPM and its submodules. The newly extended HDMC "composites" syntax will be helpful for the Serialise and CP chip submodules.Steve: simulation
Steve has made a start on documentation for the simulation package. He also reported that the two Stockholm diploma students have started work on the JEM simulation.The next steps will be integration with the DAQ. It was suggested that we should try to set an internal milestone for further integration steps.
Norman: progress with CMM and readout studies
Norman has updated the CMM documentation and created the HDMC parts and configuration files for the CMM. He has now started thinking about the CMM module services and also about defining the various database requirements of the CMM.At the request of Dave Francis, he has also rechecked the data volumes coming from the CP and JEP RODs. He showed two plots giving fragment size versus tower occupancy (0-10%) for different readout options (1, 3 or 5 slices). The Slinks from the CP system are not saturated up to 10% occupancy, but the JEP data exceeds 160% of the link capacity with 5 slices. We need to look at other zero suppression or compression schemes for the JEM. The CMM data is rather bulky.
We had a short discussion about Normans use of an Excel macro to generate the CMM parts file and the outline of the configuration file. Should the Excel be treated as the source and kept in CVS? If not, might there be conflict between Excel spreadsheets and output HDMC parts and configuration files stored in CVS and subsequently edited. Do we want to add non-Unix steps in the generation of our code?
It was felt that since the module documentation is generally in Word and we wanted to keep documentation in sync with register names in the code that there was such a step anyway. Norman will send his "short, simple" macro to those who are interested.
Murrough: run control
Network problems at QMW over the weekend meant that no one had a copy of Murroughs slides on run control and databases.The main change to the run control, apart from lots of name changes consequent on the integration with module services, was that the crate database is now split off into a separate class, so that other standalone programs can easily get the configuration of the crate and access the DaqModule objects.
Murroughs slides also summarised the status of the database. A library for the trigger menu has existed for some time, also a very sketchy first attempt at CPM calibration data. Prompted by the recent integration efforts, Murrough has started work on run parameters and run type dependent options.
Bruce commented that we will also soon need a scheme for the run control to communicate with "kicker" programs, such as that in the present "looper" scheme for generating bursts of triggers and restarting the DSS playback during the running state. Either threading or a shared memory message passing will be needed. We may be able to dispense with such programs for the DSS (via suitable DSS firmware updates) but they may still be needed for other purposes, eg hardware monitoring.
Bruce: Database organisation and HDMC
Bruce raised the subject of the medium/long term plans for the HDMC database (ie present parts manager and configuration store syntax). Even if we cant undertake a major HDMC upgrade project in the short term, we should define the API we want. We should also try to achieve a single consistent database scheme which describes not only the hardware but which is also the basis for the organisation of calibration and other data required by the hardware.Oliver said that he expected to have more hands on experience with HDMC starting around February next year and suggested that we should discuss this issue at the next software joint meeting in Heidelberg in March. Meanwhile perhaps Bruce, in consultation with others, could make some concrete proposals or suggestions.
Schedule
Murrough reported at the last (UK) software meeting on a very first preliminary software schedule (Gantt chart). This has been imported into MS Project but no further work has been done. It should include what Oliver reported as the Heidelberg schedule. It also needs many cross links between interrelated parts of the software project and also to the hardware timescales.Next meeting(s)
In the UK we will have another software meeting on Wednesday 16 January (the day before a general UK meeting). This could include a telephone conference with non-UK sites. Following that we expect another video conference in early/mid February.Actions
Actions from this meeting:- Bruce: send suggestions/proposals for new HDMC database to Oliver
- Norman: find/define remaining data formats
- Steve: documentation for simulation package
- Steve: consider organisation of test vectors
- Bruce/all: investigate software development tools
- Bruce: complete Module Service document
- Murrough: complete run control design document
- Murrough: update HDMC changes document with recent suggestions and include the proposed long term strategy
- Murrough: check D32 and D16 access with VMEBusEltec on QMW CPU
Last updated on 18-Dec-2001 by Murrough Landon