ATLAS     Level 1     Calorimeter Trigger     Software    

L1Calo Software Minutes: 17 August 2006

 

Software phone meeting on 17 August 2006

Present: Bruce, Eric, Florian, Gilles, Marianne, Murrough, Norman, Richard, Steve, Victor, Weiming.

Status reports

Birmingham

Gilles requested that we set a date for moving to tdaq-01-06-00. Richard also wants this version to have the new COOL API. We discussed this later in the meeting.

Gilles also mentioned that Chris Curtis wants to put the CAN firmware into our CVS repository.

Heidelberg

Florian has added the crate header to the VME buffers read by the crate readout program as discussed at the last meeting. This should allow more than one PPM per crate to be read (via VME). He has tested that this still works with existing one PPM/crate datafiles. Victor has made corresponding changes to his code. Florian has also made the changes to use ROD fragments instead of VME buffers. All these changes, which are committed but not yet tagged, should be tested by Florian when he comes to CERN next week.

Victor has also done more work on a channel mapping package incorporating suggestions from Steve. This should be added to the CVS repository.

Mainz

Rainer was not able to join us, but sent an email with two main points arising from the JEM test week at Birmingham.

The first concerned incompatibilities between simulation and data unless some jet/energy thresholds were zero. This is an old problem which has not been fixed because no one has taken on responsibility for the jemSim package. Murrough should chase Mainz about this.

The second point was whether we should have finer control over FIO inputs from adjacent modules. At the moment the DB only allows the whole module to be disabled. However with the observed problem of shorts on the CP/JEP backplane at Birmingham, it would have been useful to be able to disable a few individual channels. It was questioned whether we should be working on firmware, software and database to cope with this rather than expecting to fix the backplane. But the feeling was we should add more fine control.

Bruce asked that if we were making requests for JEM firmware changes that it would be useful to once again be able to send jet or energy results to either CMM under software control as this had proved useful in test setups.

QMUL

Paolo was not present, but he has fixed a subtle bug in his package which was preventing it working on lxplus. Bruce requested that he circulate the details.

Stockholm

Now that the gnamL1CaloHisto package is fixed (see above) Marianne has been able to get her new JEM monitoring histograms to work. She would like a further discussion about the future direction.

CERN

Steve has deleted some old code for the 6U ROD, leaving only the capability to handle neutral format as agreed some time ago. The old data types should be removed too if no problems with this are seen. At some point we should remove support for the 6U ROD entirely.

Murrough took the opportunity of being at DESY to do some work with Johannes Haller on integrating the new trigger configuration database into our software. Some problems were identified which Johannes will address.

Bruce has been working on ROD software in connection with the ROD review. He has observed some problems of (lack of) robustness of the roskicker program with large numbers of modules in the system. The long term future should be to convert this program to use the GNAM framework.

Moving to tdaq-01-06-00

We discussed the timescale for moving to the new TDAQ release - though there will soon be a tdaq-01-06-01 update release so perhaps we should wait for that.

We will need to move at CERN before tests with ROS/ROBin and the RoIB (September). Gilles would like to move before starting the CPM production tests (October).

Moving to the new release will require changes to at least some of the drivers, so it will not be simple to run both versions in parallel. We should change at CERN in about two weeks and fix the problems. Then other sites can move.

AOB

Gilles asked if there was a problem with addressing more than one ROD in a crate via VME. Bruce said this should be possible. Later, at CERN though, we found problems with VME address mapping when using two PPMs in one crate if using a VP315 CPU. This is a known issue, with possible workarounds, and should not affect the VP110 CPU.

Bruce reminded us that the hardware schedules were getting shorter and we need to keep an eye on software shedules too.

Next meeting

The next phone meeting will be on Thursday 31 August at 11:00 CET (10:00 GMT).


Last updated on 22-Aug-2006 by Murrough Landon