CERN QMUL
L1Calo Software meeting at Heidelberg on 14 March 2002
Actions from previous meetings
- Steve has now written user documentation for the simulation package.
- Module services and database documentation is still required...
- Stockholm and Heidelberg have favourable experience with Together. We should examine the possibility of getting Purify on at least one accessible system.
- Norman has done further work on data formats. See below.
- Other actions are carried over...
Murrough: overall status and schedule
Murrough introduced the meeting with a brief overview of the status and timescale. Modules are starting to arrive (eg CMM, CPM) but we still have many weeks of work to do on the software.Bruce: module services
Bruce presented a didactic update on the module services package. He explained the use of the "composites" syntax he has added to the HDMC parts files. This allows a more hierarchical structure where the register content of submodules can be defined once and then replicated automatically when the parts file is read.The documentation is still in preparation, but he has already provided an example module (Bbm) and submodule (BpChip) as both parts file and code as an illustration.
The module services code will shortly be moved from the HDMC CVS repository to the RAL respository. We expect to have one library (maybe even one CMT package) per module to ease the development of the different modules in parallel by different people.
The existing CPROD test code is gradually moving into the new module services scheme. The next steps include another integration round with the run controllers and the database. This will allow an implementation of a bigger system.
The timescale for completion of work on module services is about two weeks plus time for documentation.
Steve: test vectors
Steve gave a short preview of his longer talk to be given on saturday, concentrating on test vector generation.He listed the test vector generators he knew about. We will clearly need more test vectors, covering larger and more diverse systems (eg with test vectors loaded into DSS, CPM and PPM at once). He also touched on the proposed scheme for test vector and system test organisation as discussed at our recent brainstorming meeting.
Apart from test vector generators, further work includes using the scheme to generate L1As in the simulation as well as implementing the test organisation scheme. All this will take at least a month.
Sam reported that the Stockholm students had found the documentation very useful in starting work on simulation of the JEM.
Murrough: run control, databases, etc
Murrough reported on the status of work on databases. Some aspects, eg trigger menu, have been stable for some time. The configuration database has been extended recently to include descriptions of cabling. The calibration data is partly implemented but needs more work. For the benefit of module services classes, it will also be useful to introduce a layer which combines all the sources of data for each type of module and submodule. This has not yet been done. Probably another months work is required in the database area.The run control area has been fairly stable but needs some more work (a week or so each) on the sequencer program for multistep tests and the DSS "kicker" program to keep the DSS looping during tests.
Panels have been added to the IGUI to set L1Calo specific run parameters. It should be easy to add new parameters if required.
As reported in more detail at the last meeting, the run control and database packages are now succesfully using CMT.
Oliver: software for the PP system
Oliver said that recent effort at Heidelberg had gone into computing infrastructure improvements. The focus for the next few months will be developing firmware and software for the ASIC and MCM tests. This will be done within HDMC. New parts are required as well as code to generate and interpret test vectors. The VME driver also needs to be installed for the "Heidelberg home brew" CPU.Gilles: software for the CPM
Gilles has been using the module services example code from Bruce to start developing the parts file and module service code for the CPM. The parts file still needs description of the serialisers, while the code is waiting for definition of the database. At least a months work is still required.Norman: software for the CMM
Norman reported that the CMM software was in a similar state to the CPM. HDMC parts file exists, but little has been done on module services code.There is also no simulation code yet written for the CMM. The algorithm part should be straightforward, but the readout code is more complex - it also needs agreement on the data formats (seebelow). There may be some overlap with readout code for other modules, especially in the handling of the L1A scheme for the tests. Code developed for one module should be propogated to others.
Thomas: software for the JEM
Thomas said that initial tests on the JEM had used only VMELinux driver calls to test the VME interface. However the latest version of HDMC has now been installed and code for the DSS obtained from Bruce. This will be used to test the realtime path.On the firmware side, Andrea has been in contact with Stockholm regarding incorporation of the Jet algorithm block into the main JEM FPGA code.
Bruce: ROS for Level 1
Bruce reported on developments in the ROS area. We have been promised a "beta++" version of the official ROS distribution by the beginning of April. When we get it we should test it.So far Bruce has only used a private development version of the ROS code and hasnt really used the ROS function at all. However the Slink test code has proved useful in testing the ROD by providing a random source of Xon/Xoff. He has passed back comments to the ROS developers.
Bruce also mentioned that Markus Joos at CERN had started doing tests of industrial PCs with many PCI slots. These are ongoing but look promising.
Norman: DIG training
Norman mentioned the forthcoming DIG training sessions. The first one will repeat the Online software training which a few of us have already been on. The second will concentrate on ROD crate DAQ aspects (eg monitoring programs and histogram transport) and ROS configuration. This should be very useful. We should book on these courses via Norman.Norman: data formats
Norman has updated his compendium of data formats (which is available via the technical information notes page). He has added and/or updated various Slink formats, especially for the JEM and CMM. The CMM formats still need confirmation with Ian. The document will then be fairly complete apart from the preprocessor formats. Apparently these are defined in Bernd Stelzers diploma thesis, but should be incorporated.Thomas: calibration
Thomas reported that he has made contact with Christian Zeitnitz, the LAr Online software contact. The LAr groups will have a full ROD crate test in autumn 2003. Meanwhile the Annecy group is working on standalone calibration but without any input from us so far.We should try, in the near future, to set up a meeting between two or three interested people from each side to discuss the many issues in this area.
Normans report from the LAr week, later in this meeting, also touched on the calibration issue.
HDMC issues
We had a short discussion on issues relating to HDMC. Murroughs talk mentioned a few items relating to libraries and some new parts, common to all our modules, which would be useful to develop soon (eg TTCrx and FPGA loading).Bruce also mentioned some of the important, but longer term aims, eg for access control and a proper upgrade of the parts file syntax to integrate with the configuration database.
Norman reported that HDMC had been mentioned in the ROD crate DAQ task force. We need to be clear that while HDMC is open source and available to others that we dont have the effort to support outside users.
Addendum: brainstorming session on 16 March
At the end of the joint meeting we had a short extra informal brainstorming session, covering a few issues.Calibration: Thomas will try to setup a meeting with the LAr people during the next LAr week. Attendees from our side will be Thomas plus Norman and/or Murrough.
DIG training: probably three of us will attend the first session. A different three or four will go to the second session. For the latter session we hope for experience in using and configuring the ROS, using the event monitoring framework and histogram tranport package.
We discussed the overall design for the slice test DAQ. It is clear that we need to document this overall idea in one place.
It will be useful if we can establish a system, eg at RAL or perhaps at QMUL, where all of us can login and find all our DAQ type software installed and configured. This will allow people to play with it before installing it (or without having to install it). An alternative to having a system people can login to would be installing the software properly on an AFS accessible directory.
It would also be convenient if we could move all our software to a single CVS repository, rather than having HDMC separate. Either RAL or CERN would be suitable. To be investigated.
Next meeting
To be arranged later.Actions
Actions from this and previous meetings:- Bruce: complete Module Service document and submodule syntax description
- Bruce/Murrough: define required HDMC improvements
- Murrough: write database user guide
- Murrough: complete run control design document
- Murrough: setup standard system for external users to login to and use default Online/L1Calo DAQ system
- Murrough/Bruce/Oliver: merge HDMC and other L1Calo CVS repositories?
- Murrough/All: write overall guide to L1Calo DAQ software design (especially regarding slice tests)
Last updated on 18-Mar-2002 by Murrough Landon