Notes of the meeting on modelling at CERN, March 3 1998.

Present : S. George, P. Clarke, R. Highes-Jones, K. Korcyl,
          J. R. Hansen, J. Vermeulen

Notes by J. Vermeulen, March 10, 1998

The workplan for the pilot project was discussed in detail. The result can be found below.

The draft document for the TPR had not yet been read by majority of the participants before the meeting and was therefore not discussed in detail. Please send comments via email to JV.

The meeting concluded with a presentation of Peter Clarke on his ideas of how to specify sequential processing for SIMDAQ-C++. He proposes to start with specification of the trigger menu items in the trigger menu file together with a keyword indicating the trigger strategy. The possible strategies would be hardcoded in the program. A strategy is then chosen according to the keyword. Relevant reduction factors can be specified in the trigger menu file. Peter and Gordon Crone will proceed to implement this. Where to specify processing times was also discussed. JV mentioned that these are now specified as parameters for the various processors, but that for the model C processors this is not feasible, as there are different types of feature extraction and also global processing to be executed. In the physics event data structures there are already variables available for storing  this information and JV is working on a modification of SIMDAQ-C++ so that the necessary information can be read from the configuration file by the PhysicsEventGenerator object. For the processors then only a multiplication factor needs to be specified that takes into account the type of processor used (i.e. relative computation speed).

Next meeting : probably end of May (most likely date is now Wednesday May 27) at UCL
 
 

Pilot project modelling workplan

Goals

1. input to decision making,
2. aqcuisition knowledge about the factors controlling system behaviour and resource requirements,
3. understanding relevant technologies with respect to behaviour and resource requirements if applied in the ATLAS environment,
4. provision of help in understanding results obtained with test set-ups, with as side effect "confidence building" with respect to modelling results.

It should be noted that unforeseen results are possible, which could make it desirable to revise the workplan.

Workplan for 1st half year, milestones are June ATLAS week and October workshop, to be discussed again in May meeting :

Paper modelling

Item                                Who           Implementation done        Results available

Review of parameters               AA,SG,JV           June                      June
  in new version of DAQ note 70

Update of paper model              AA,SG,JV           June                      June
  (trigger menus,
   different # of ROBs, etc)
 

Generic model, SIMDAQ-C++

Item                                Who           Implementation done        Results available

Specification of sequential        PC,GC              June                      October
  processing

Implementation of sequential       KK                 June                      October
  processing

Model C processors                 AA                 October                   > October

Pull model B + RSIs                JV                 June                      October

SFIs                               AA                 October                   > October

"generic model of test setups"
  DS-link                          JV                 June
  SCI                              JV ?
  ATM model C                      AA ?
  (Macrame)
  pilot project                            -- Specify in October --
  HPCN

Other

Item                                Who           Implementation done        Results available

Technologies
  SCI                              RHJ
  ATM                              AA
  Gigabit Ethernet                 RHJ,KK
  Supervisor                       BB
  ROB complex                      BR,PC,GC

Test setups

Full model                                           June 1999
 

AA  = A. Amadon
SG  = S. George
JV  = J. Vermeulen
PC  = P. Clarke
GC  = G. Crone
KK  = K. Korcyl
RHJ = R. Hughes-Jones
BB  = B. Blair
BR  = B. Rensch

NB : part of the work, in particular with respect to understanding technologies, may be done using Ptolemy.