1. Paper modeling issues : M. Abolins has studied the note by S. Gonzalez (ATL-COM-DAQ-2000-025 ), "Further Studies and Optimisation of the Level-2 trigger" and is looking into how to implement the procedures and results described in this note in the spreadsheet model.
2. ROS modeling : R. Cranfield is looking into the Ptolemy code of B. Rensch, modeling phase 1 is the starting point for ROS modeling. An UML description will be produced, which can serve as an example UML description for Data Collection and HLT models. J. Vermeulen has been looking into what UML tool to use, see point 5. He has also collected relevant parameters and will pass these to R. Cranfield.
3. Data Collection modeling : K. Korcyl reported that new measurement results are available on an 8 port Gigabit Ethernet Titan T5 switch. He is working on an updated switch model, which may be finished at the end of the week. In a short discussion the items on modeling in the work plan of the Data Collection group was mentioned and that work items could be proposed to the Data Collection group. R. Blair will circulate an updated work plan on modeling via the modeling mail list for discussion, and to be passed later to the groups.
4. HLT modeling : S. Wheeler is the only person involved at the moment and intending to start working on it in the beginning of the new year.
5. UML Tools : J. Vermeulen reported to have looked at ArgoUML (free Linux tool) and Together (commercial tool from TogetherSoft). Together is recommended by the ATLAS off-line group. Academic licenses (for educational and academic work) are available to academic groups without cost. The tool seems to be a good choice. It is written in Java and Linux, Windows and Mac OS (but only for an older version of Together) are supported. It requires a >= 400 MHz CPU and >= 128 MByte of memory. J. Vermeulen reported that reverse engineering of part of the simdaq code worked well.
6. Simdaq : J. Vermeulen has set up the full system model using the component models used for the pilot project testbed application model. This seems to work, results have to be checked against the paper model.
7. Ptolemy : K. Korcyl has looked again at the
application testbed model in order to resolve the discrepancies between
measurement and model results for the measurements not reported in ATL-DAQ-2000-039.
With the additional parameter used in the simdaq model and by tuning one
of the parameters better agreement than with the simdaq model is now obtained,
but now there is a discrepancy with the results used for the DAQ note,
so further work is required. J. Vermeulen asked whether the usage of 300
and 400 MHz processors was taken correctly into account. This has to be
looked into.
K. Korcyl mentioned that a Polish PhD student will work
on modeling at CERN from the beginning of the new year. He may devote a
significant fraction of his time on HLT modeling.
8. "Chiba city" : the work started on using the system, further work is required.
9. Next meeting : Tuesday 16 January 16.00, telephone conference. In the week of January 29 there probably is a dataflow meeting of a few days at CERN, wich may offer an opportunity for a face-to-face meeting.
10. AOB : Up to 3 persons from a Rumanian group are interested in modeling. They need self-contained tasks, a possibility is development of a Ptolemy model of the "Chiba city" system, R. Cranfield intends to help, R. Blair can provide data.
Notes by J. Vermeulen