CERN : R. Blair (Argonne), M. Abolins (Michigan)
NIKHEF : J. Vermeulen
1. Paper modelling: The comparison of computer model (simdaq) and paper model results is almost finished. It resulted in the removal of a number of minor bugs from the paper model and a few corrections to the configuration files used for running simdaq. The complexity of the paper model is causing this comparison to be a time-consuming procedure. With the knowledge available now it would be possible to set up a better organised spreadsheet, but this is still a lot of work. It was regarded to be doubtful whether this is worth the effort. An updated version of the document on paper model results is still to be finished and then to be submitted as DAQ note. Also the new computer model results need still to be documented. The consequences for modelling of the "ROS on a ROD" issue and the possible absence of multi-casting were discussed. The "ROS on a ROD" issue can be studied in a straightforward way with the current paper model. M. Abolins and J. Vermeulen will look into possible scenarios for a system where multi-casting is not possible. The spreadsheet has to be modified to study these scenarios
2. ROS modelling: As reported in the last meeting, implementation of the models based on the UML description can start. It was discussed whether it would make sense to start to do this in parallel in at2sim and simdaq. As also working in this way is tested and in view of efficiency, a first implementation in only one program - at2sim in view of the plans of and work done by R.Cranfield - seems to be natural.
3. Data Collection modelling: P. Golonka as well as K. Korcyl could not attend the meeting, but K. Korcyl had submitted a short status overview . The work is continuing now in Cracow, where Ptolemy and at2sim have been installed. It is the intention to concentrate on the "strawman architedure" presented by R. Dobinson in the Data Flow meeting of 15 October. The work on UML diagrams still needs to be started, it is intended to start with the switch and parameterized TCP models.
4. HLT modelling: S. Wheeler could not be present due to a clash with the HLT meeting. Via email she had informed J. Vermeulen that no progress could be reported. For the UML diagrams she is waiting for the availability of code. However, she does not expect to have time to work on the UML diagrams and hopes that the Data Collection group can provide these. J. Vermeulen suggested that the new Norwegian student mentioned in the email by K. Korcyl (see also the next point) can help here and will contact K. Korcyl about this.
5. Ptolemy: K. Korcyl and S. Wheeler had written that there were no further developments to report. A new Norwegian student (already mentioned in the previous point,) who will be at CERN from the beginning of next year, is becoming involved in the work.
6. Simdaq: Since last meeting no modifications to the program have been made, recent results have been used for cross-checking the paper model results (see 1.)
7. "Chiba City": R. Blair reported that improvements in drivers used in the system give rise to the hope that the problems experienced can be solved. He is not yet prepared to give up and hopes to arrive at a conclusion at the time of the NIKHEF week. K. Korcyl had reported that the modelling activities were postponed due to the absence of calibration data. A spreadsheet with results should already have been received earlier by K. Korcyl, R. Blair gathered that it was not sent, so this is to be sent to K.Korcyl
8. Workplan and status in relation to workplan: An updated task list had been made available earlier by R. Blair and was discussed.There was some discussion on how the item "Full system model with rough components implemented" should be interpreted. It is completed in so far that in at2sim the model has been set up using the models for the testbed components. For simdaq it is a matter of producing an appropriate configuration file, but this has not been done. A serious study, consisting of running the model(s) and comparison to paper model results would be a natural next step. At this stage, with changing hardware and software, it seems that the effort needed is not justified. Considering this and in view of the state of the current computer models it was concluded that this item should be considered to be "completed". Another item discussed was the comparison of computer and paper model results. This requires a substantial amount of effort but is very useful for detecting and correcting bugs in the paper model. In view of the amount of effort it makes sense to add "comparison of paper and computer model results" as an additional task to the list. As there are two reviews of the paper model (one last summer and the next one in March) there should also be two tasks, one with deadline in November, the other one one with deadline in July of next year. The Phase 2C system model results are foreseen to be available only after the date of the TDR, but this is not considered to be a problem. It seems now that the milestone for the availability of the UML diagrams for Data Collection and the Higher Level Triggers of December of this year is too early, so this has to be shifted into 2002.
9. Next meeting: The next meeting is in the NIKHEF workshop at Tuesday 13 November. As there is no ISDN system in Cracow, for meetings after the November meeting we may have to switch to VRVS, the quality of which seems to be quite acceptable.
10. AOB: As reported in last meeting, R. Cranfield is acting as modelling representative in the "Connect Forum" (QA group).
J. Vermeulen, 29 October 2001