This is the log book for the MSU project started 28-NOV-1994 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 19-JAN-1995 phone conversation between Jim Proudfoot and Philippe - Jim will leave for CERN January 20th. John Dawson is on vacation until 3-feb. John will go to DAISY mid-March, and can/will hop over to CERN. I think I understood that John asked him (but only AFTER leaving for vacation) to become a contact while he is at CERN and collect information from the UK guys. John didn't get a chance to tell him much, just to try and call Philippe....! Jim is part of the Calorimeter effort. (by the way he thinks this isn't too hard, and that they will have time left over to think about other stuff). - Expose the idea of the involvment of the US with respect to the T2 supervisor. explain the UK T2 Testeam 94, and the plans for 95. - Tell Jim to find the 2 documents by Yuri Ermolin (t2 Sup + RoI builder) on the Web. Give him the names of J.Strong, and F.Wickens. - Forward to Jim, the notes from the ANL meetings, and the notes from the December CERN meeting. - His address is Proudfoot@anlhep.hep.anl.gov Phone (708) 252-4357 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 16-JAN-1995 Minutes of 11-Jan-1995 meeting at ANL ----------- Minutes from the ATLAS Trigger Meeting 11-JAN-1995 4PM at ANL with video link to Irvine taken by Dan Edmunds In Attendance: Maris Abolins, Larry Price, John Dawson, Bob Blair, Teresa Fuess, Andy Lankford, Dan Owen, Dan Edmunds 1. Note not everyone has received a copy of the minutes from the European half of the meeting on 16-DEC-94 during which the audio link did not work. 2. Hardware: We need to have a specification for the L2 Supervisor. We need from the Europeans: 1. The hardware interface specification. 2. The specification of the minimum functionality for the L2 Supervisor. The we have some flexibility about adding functionality above the minimum. John Dawson goes to DESY for 2 months starting 22nd of February. He would like all the hardware for the L2 Supervisor ordered before he leaves. The next CERN test beam run for triggering is scheduled for 15-30 Sept 1995. Bernard Pope will be at CERN starting this summer for one year. 3. Software: The program ATLAS trigger simulation programs are now running on the SGI at ANL. No actual code changes were required to get these programs working. It looks like CERN is starting to support these programs on machines besides the HP. The newest version of Geant 3.21 is required for things to operate properly. Now that the software problem is solved what are we going to do? There is no plan yet. The new SUN workstation is running at MSU and MODSIM 2 is being installed. Can MODSIM 2 be run remotely? Need to find out. 4. Money: No formal information yet about the DOE emergency request. There is no NSF emergency pot of money. Maris is going to try a supplemental request to get "emergency" money for this year. 5. General Items: There is a modeling meeting sometime in March. We should have someone go to it to present something that we have done. There is a ATLAS Trigger Steering Committee meeting on January 31st. Maris will attend via a video link from Fermi. Maris is going to send a note to Livio Mapelli to let him know that Larry Price will be at CERN at this time and would like to attend the Trigger Steering Committee meeting. This will be the second meeting of the Trigger Steering Committee. MSU is getting a VTEL tele-conferencing system. Schedule for our next meeting is Wednesday February 1st at 1PM at ANL with a video link to Irvine. We hope to have L2 specifications from the British people by then. Bob and Teresa will focus on what to simulate and will communicate any interesting ideas via Email. Maris with a MSU colleague or student will attempt to get the SIMDAQ package running at MSU Maris will distribute the minutes of the 16-DEC-1994 meeting at CERN during which the audio part of the link to ANL did not work. (Done) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 16-DEC-1994 Summary of ANL meeting from 16 Dec 94 The video link to participate to the L2 meeting at CERN failed. We still had an informal (local) meeting: Bob Blair, Theresa Fuess, Larry Price, John Dawson, Maris, Dan, Steve, Philippe Bob and Theresa appear to be interested in alorithm studies. Bob had tried porting the ATLAS software (written on an HP) to an SGI, failed, and gave up. ANL might already be getting some HP stations from other sources, and/or might considering buying one. Steve will check with the MSU CASE center about the possibility of using some of their HP workstations. Some of out HEP physicists might/should be interested in participating in algorithm studies. Maris mentioned that Bernard would be on sabbatical at CERN starting next fall Setup the next ANL/MSU meeting (with possible link to UCI) at ANL for Wedn, 11 Jan '95 at 4pm. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 15-DEC-1994 Summary of ANL meetings from 15 Dec 94 pre-meeting 3pm --------------- present: John Dawson, Larry Price, Maris Abolins, Steve Gross, Philippe Laurens Maris presented MSU's lack of funding for ATLAS T2 work for 1995, and MSU's decision towards a major commitment to the D0 trigger upgrades for Run II. Larry Price acknowledged, and asked J.D. if he still thought that this was a doable project for ANL, given the small contribution level from MSU. John said yes. There was a discussion about the relative merits of signing up for this project. Everyone agreed that this would be 100% throwaway work, and that this prototype (and the current UK T2 Test Beam Architecture) would not have much to do with the final thing. It was also noted that it wasn't certain that this UK group would end up building the L2 system. It wasn't clear how much one would really learn from this exercise, but that it was still important to try to learn. The general consensus was that we couldn't pass this offer without losing credibility with respect to the US involvement. And that we could at least get to know some people, and exercise a work relationship with our European colleagues. Larry asked if other types of involvement were possible, and Steve + Philippe mentioned the possibility of defining/designing a complete Level 2 system. Larry + John had the impression that the collaboration didn't need/want this type of involvement from the US group(s). Steve + Philippe expressed the opposite feeling, based on earlier private conversations with N. Ellis and L. Mapelli. Anyway, it was clear to everyone that neither ANL or MSU have the resources to undertake such an effort. The conclusion was that this US group should go ahead with the project, and confirm its intentions in the CERN meeting of Friday (via video link). meeting 4pm ----------- present: John Dawson, Larry Price, Maris Abolins, Steve Gross, Philippe Laurens with video link to UCI: Andy Lankford Maris: Repeated MSU's limited resources. Andy: Stated that UCI's conditions and position were identical to MSU's. John: Not much to learn from the T2 Sup prototype, except that it would expose us to some of the problems, and help one think about the real thing. We need to pick up this project to gain/maintain credibility. Saying NO would probably have grave consequences. John: answers a question from Andy about resources necessary for this prototype: - Some money from ANL for hardware ($15k) - Some % of John's time (design + hardware) - Some % of MSU's time (design + software) - Go there to install + test - Take shifts at the Test Beam Andy: repeats that, like MSU, he cannot involve his people unless some funding is on the way. UCI is already involved in pixel readout for silicon tracker, but has no money for T2 supervisor work. Crucial to stake out some territory for this US involvement He senses that ATLAS puts a great emphasis on Test Beam activities "We" need to do something, please do it, and save room for me. Larry: We should also save time to work on the next generation of the Y.Ermolin definition of the T2 Sup, RoI Builder,... It is important that US names start being associated to it. Andy: The real T2 Sup project seems like a hard job (100 kHz and lots of transactions). This is an important point that needs to be studied. By the way, what are the rates at the Test Beam. Phil.: Bursts of 150 Hz, or 1kHz for some detectors. The UK TB system might only be able to handle the 150 Hz level in '95. Maris: Our involvement must start to become concrete/visible to gain leverage at CERN and for funding from the US agencies Andy: How much work is needed on the real/final T2 Sup in '95? Is it more important to understand the real T2 Sup in '95 or build the T2 Sup TB prototype? What are the priorities? Phil.: Schedule might help us: 1) build prototype by September 2) work on next revision of Ermolin's notes. 3) Back it up with some MODSIM simulations Must force a simple design with minimal custom hardware, simple software Resist all temptations/pressure to make it any fancier than necessary. Use same modules/environment as the UK so that they can operate and maintain this thing. Larry: What is the UK philosophy about the final device and development path. How does their '94 and '95 Test Beam effort fit into this? --> good question for tomorrow's video conference. There was a discussion between Maris+Andy about funding difficulties from DoE/NSF, and about likelihood of LHC approval. Andy: Is it more Software than Hardware? Phil. We must enforce that the software aspect stays simple and doesn't explode out of proportion if we want to be successful, get it done on time, and have it supported by the UK guys. Steve: Answer Larry's question: UK TB code is all written in C. We should thus do the same thing, using the same real time OS. There were questions about a separate T2 effort that was being conceived at CERN: a T2 test platform based on the DS link protocol, and for evaluation of processors and algorithm. This effort is independent of the UK effort, and emerged from a CERN R&D program. The name attached to this is John Hansen, and he is on the agenda for tomorrow's video conference. There was a discussion between Maris and Andy that concluded that no other US groups were involved or interested in the L2 Trigger. Andy: Could we do modelling only? Phil: Some aspects are already well covered Simulate what? new ideas, or somebody else's proposal? There is a lot of space to explore It is quite time consuming, and uses a lot of CPU resources. Larry: wants to ask tomorrow what the next milestone will be for the Trigger studies, now that the Technical Proposal rush is over. Andy: Are people really studying algorithms, does one exist? Remind of the constraint of 100kHz L1 accept => 10us per event in L2. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 28-NOV-1994: Summary of information from DAQ and L2 meetings of 18 to 23 Nov 94 DAQ Organization (Mapelli) ---------------- There is NO excess of resources. DAQ is worse off. 4 Areas: Overall system design Develop Central DAQ Develop detector specific interfaces Prototypes for Test Beam No single institution can build the whole DAQ. CERN can maybe do 25 % of Central DAQ. Technical Proposal ------------------ Trigger/DAQ is at: http://atlasinfo.cern.ch/Atlas/GROUPS/DAQTRIG/TPBACKUP/tpbackup.html or [Atlas Welcome] -> [Notes/publications] -> [Trigger/DAQ chapter] printed copies of some parts Test Beam --------- Test beam line is at H8 1994 6 groups were present : TRT LAr Tiles MSGC UK-L2 (L1 CAL?) with various readout capabilities 150 Hz trigger during bursts L2 was "asynchronous" (off tape, to tape) 1995 Integrate most systems and readout schemes Add busy control Central run/tape control Trigger ID Run Database Electronic Logbook Still fixed detector readout selected per run Up to 10kHz burst for TRT Still NOT all purpose, but not local DAQ VERY BUSY SCHEDULE at H8 for 1995 No room for development --> start another beam line (at H6) Saclay: 100 Channel Test Cal Readout over next 3 years compare different shapers, analog/digital pipelines... measure energy/position resolution, cross talk, calibration Integrate L2 Simulations -------------- WWW: ftp://sunsci.cern.ch/simulation/atlas_daqsim/www/modelling.html Geometry / segmentation: LeDu + Bock DAQ note data input: random from 179 GEANT events ( 2.2 RoI/event ) SIMDAQ = ATLAS package of source code of simulation language MODSIM runs of 5-10k events randomly picked from these 179 events. i.e. 50-100ms of beam time 5-8 hours of SPARC 10 Goals: evaluate performance of various switching technologies for ATLAS relate algorithm length vs. # of nodes vs. latency vs. # of buffers Check model on small switches from CERN, Saclay... to validate extrapolation ATM (S. Hunt) SCI (A. Bogaerts) various topologies, CMOS, AsGa http://www1.cern.ch/RD24 Fiber Channel / HiPPI These are new results, people are still learning/working/tuning Need longer runs, integrate L2-L3 L2 Muon algorithm (S. Falciano, Italian group) ----------------- Sharpen Thresholds Read 0.6-4kB out of 100kB of barrel data 500 us of FORTRAN measured on R3000 Data Driven L2 -------------- "Enable Machine" = FPGA "processor" Currently: 100 kHz of RoI of TRT 2nd Generation: general purpose for all detectors programmed in a C-like syntax (Enable ++) test next summer/fall L2 Test Bed project (new): -------------------------- John Renner Hansen L2 test rig based on DS links and switches Something CERN already has, needs to modify for C40 RoI Unit: --------- New document by Y. Ermolin MSU/ANL Building L2 Sup for UK L2 test beam setup ================================================= PROs ---- ANL seems to want to help in $$ and time (but no experience with C40) Maybe controllable/definable as a single C40 commercial board with little custom glue hardware Maybe simple, cheap way to 1) exercise our joint venture with ANL 2) show involvement / press for $$$ 3) Get foot in door 4) participate in Test Beam 5) "Build something" in 95 CONs ---- L2 sup will NOT look like that, even at the level of functionality UK/J.Strong might not end up building L2 L2 will most likely not look much like that (buffer, switches..) 99% throw away work other commitments, D0, ... The L1 and L2 UK groups might not be the "ideal partners" Septembre = 9 month, project far from defined Hard to support from a distance Lot of work Questions --------- How much money will MSU/ANL get/spend for Hardware in 1995 What TYPE of involvment do we want for 95 What could we buy vs. piggy back (C40 HW, Jtag, debugger, loader..) Is this doable from here Who can we send and for how long to install/support it. Minimal Actions --------------- Decide if this is worth pursuing Write description/definition ASAP (end of january?) Guidelines (short time, no money, throw away, far) ---------- As simple as possible (system, HW, interfaces, software...) can be installed quickly can be left unsupported (supported by the UK guys) look enough like what they use (but this differs from what we know)