Argonne : R. Blair
CERN : P. Golonka, K. Korcyl, S. Wheeler
Michigan : M. Abolins
NIKHEF : R. Scholte, J. Vermeulen
UCL : R. Cranfield
1. Paper modeling : M. Abolins reported that he changed the low luminosity menu in the spreadsheet for obtaining the reduced scan rate of 5 kHz. He also scaled the processor speed with a factor of 4 and changed data transport speeds from 80 to 160 MByte/s. He said that he would send detailed information to J. Vermeulen and ask for comments / suggestions via email.
2. ROS modeling : R. Cranfield mentioned the document with UML diagrams discussed in the last meeting. He has now offically access to the DAQ-1 code and received additional diagrams from J. Vermeulen. In a telephone discussion the day before the meeting it was agreed that R. Cranfield extends the diagrams. He is still thinking about how to describe the Ptolemy code of B. Rensch in terms of diagrams and found that complicated activity diagrams are necessary to describe the details of the model. Problems with "Together" running under Linux were mentioned : the program is sometimes almost unbearably slow, while the windows often needed may appear iconized in stead of normally opened on the desktop. These problems have been observed on a 128 MByte desktop machine and, to somewhat lesser extent, on a 192 MByte laptop. K. Korcyl and S. Wheeler had the experience at CERN that the program was running very slow on the ATLAS cluster with display on their local machines. PCs to be delivered soon should make it possible to run the program locally. J. Vermeulen did not see the problems of R. Cranfield under Linux (PC based and a version running on a Macintosh Powerbook, i.e. PowerPC based) but said that he did maybe not use the program intensively enough in the Linux environment. He asked R. Cranfield whether he used the newest release of Java, the answer was that this should be so. R. Blair suggested that Win4Lin could provide a Windows environment under Linux in which the program could run more reliable. R. Cranfield also mentioned VMWare as alternative, Win4Lin however could be faster.
3. Data Collection modeling : K. Korcyl and P. Golonka have continued with the model for the TCP/IP stack, for which use is made of measurement results obtained with the setup in bldg. 513. Results are now available for version 2.4 of the Linux kernel, an improvement of 20 microseconds in the offset for sending a message has been seen, as well as an improvement in the behaviour for multi-segment messages. It is estimated that a first version of the model will be available at the time of the T/DAQ week in April. A document with a description of the model and its parameters and the measurement results, including results obtained with the new version of the Linux kernel, will be produced.
4. HLT modeling : H.P. Beck, K. Korcyl and S. Wheeler have had a discussion on modeling which led to the plan that S. Wheeler will look after the Event Builder model, as initially implemented by G. Lehmann. The start of this activity is waiting for a new PC to arrive. P. Golonka will look after the processing model for the LVL2 trigger, again the start of this activity is waiting for another new PC to arrive.
5. Ptolemy : R. Cranfield reported that the multi-step processing seems to be working correctly, but that additional refinements (sequential processing of the calorimeter data was mentioned) are necessary and comparison to existing results needs to be done. P. Golonka gradually will take over the multi-step menu code.
K. Korcyl and J. Vermeulen had continued to look into the remaining discrepancies between the Ptolemy and Simdaq testbed models. J. Vermeulen reported on the possibility to describe the testbed results with the help of simple formulas and showed some plots illustrating this. He had almost completed a description of his findings, which is intended to be part of a larger document that also should include new results from the Ptolemy and Simdaq models. An earlier version of this description had already been sent to K. Korcyl, a new and much improved version would be sent to him for comments the following day. It is intended to make this partial document available on a short time-scale, as soon as it is more or less stable.
Finally R. Cranfield pointed out that there is a URL for snapshots of the at2sim code, there is a link on the Ptolemy modeling page, the link will also be placed on the modeling page.
6. Simdaq : J. Vermeulen had nothing to report, apart from the necessity to modify the code somewhat so that inter-frame gaps as used by the Ethernet link protocol are taken into account. This is necessary for the testbed model.
7. "Chiba City" : R. Blair is now close to being able to run, after experimentation required due to a lack of up-to-date documentation of the system. The "two IP addresses problem" has been resolved (but it had anuther cause than that implied by the name given to the problem). The system is running kernel 2.4, so the behavior will differ with respect to communication via the network from the testbed studied at CERN. The visit of a member of the Rumanian group interested in modeling the system to CERN has been delayed and is foreseen now to take place in 2 - 3 weeks. In view of this time scale it was remarked that it makes sense to plan for a visit during the April trigger/DAQ week. In view of his findings on the testbed model J. Vermeulen asked for doing measurements for more than three event fragment sizes, for example for 64, 256, 512, 768 and 1024 Bytes.
8. Workplan and status in relation to workplan : J. Vermeulen concluded from the present status that the work is progressing reasonably in line with the milestones specified in the workplan, although some items (work on Chiba City and on interpretation of the testbed model measurement results) proved to require more effort than originally anticipated.
9. Next meeting : The next meeting is to be held at CERN during the trigger/DAQ week in the first week of April, a time slot of 2 hours has been requested from A. Lankford. A video meeting near the end of April is foreseen, the date and time are to be decided on in the next meeting.
10. AOB : There were no further items to be discussed.
Notes by J.Vermeulen, 16 March 2001