Software phone meeting on 20 July 2006

Present: Bruce, Damien, DaveS, Eric, Florian, Frederik, Kambiz, Murrough, Paolo, PeteW, Rainer, Steve, Victor, Weiming.

Status reports

Heidelberg

Florian has finished work on the run plan package for the moment. He proposed to start preparing for full preprocessor crate tests at Heidelberg, extending software to read more than one PPM per crate. He intended to use ROD Crate DAQ as this already has support for many modules per crate together with better speed. He wanted to try this under tdaq-01-06-00 as he had found some instabilities with RCD under tdaq-01-04-01.

This plan generated a lot of discussion. Kambiz felt that the full crate test didnt need to read out more than one module, just to read one in a full crate. It was also commented that now that the ROD readout of the PPM at CERN was working, focus should finally move to migrating PPM monitoring/analysis for cable tests to use ROD fragments rather than put too much more work into the VME readout.

Bruce also pointed out the the formatted buffers used in the VME readout program (crate_readout) could have additional crate headers added, making readout from multiple modules per crate possible. The crate_readout program should be changed to do this (requiring corresponding small mods in Florian and Victors code).

The remaining major issue with migrating this code to use ROD fragments is related to the database: retrieving the scan parameters from the DB rather than private VME readout buffers.

A related point is that we need to check that the ROD does correctly update the step number in the ROD fragment header. (After the meeting this was checked and found to work). It also needs to be made available in the RDOs produced by the bytestream decoding. (Also done after the meeting).

In conclusion, although we do want to complete the move to RCD one day, and do also need to prepare the move to tdaq-01-06-00, it was felt that Florian should not spend too much time on those but more on preparing his software for handling ROD fragments. For this it would be useful for him to have a multistep run dataset from the PPM-ROD setup at CERN. (Also done after the meeting).

Meanwhile Victor has done more working mappings between (eta,phi) coordinates and digital channels in PPM and JEM. CPM should come soon. His plan for his week at CERN is to work with Paolo to port his monitoring package to the GNAM framework.

Mainz

Rainer has been writing some small tools to help with the JEM full crate tests. However he wasnt able to be at the tests this week in Birmingham so doesnt yet have detailed feedback from them.

QMUL

Paolo has not yet had time to fix the problem with loading more than one of his new plugin classes in his gnamL1CaloHistos package. He hoped to look at that during this working week at CERN when he will also be helping Victor and Marianne get going with developing monitoring in GNAM.

RAL

Damien has finished integrating Daves latest PPM data compression algorithm into his simulation. For the dataset being used it gives a 70% compression factor.

Dave reported that with help from Ian he now has a gate level simulation of the firmware. This shows an 83% occupancy which is a little high, but more seriously there are timing violations so the algorithm wouldnt actually run in practice. It needs some further tweaking which requires help from experts who are rather busy at the moment.

It was requested that this latest favourite algorithm be documented so that Steve and Pete Faulkner can implement online and Athena based bytestream en/decoders for it. Dave mentioned that the algorithm still needs a small mod to handle disabled channels.

Module services for the new PPM

The new PPM has some changes to the memory map. These are now going to be more extensive than previously advertised. So it now seems unlikely that we can gently evolve ppmServices to handle the new PPM while also being able to handle the existing one, even if the memory map is updated on the old PPMs.

Kambiz was asked to look again at any possible strategies for making the change as easy as possible. However we may need to freeze the ppmServices for the old PPM and have a big bang change to the new PPM.

AOB

Eric asked Kambiz if the PP full crate test was going to look for "collective effects", eg to all LVDS link and all Glinks. The answer was yes.

Bruce raise the issue of the present calibration files which are only named after crate and module. If you need different ones when running different partitions this is a problem. However Murrough felt that it would also be a problem if you wanted the same calibrations in several different partitions. To be discussed further.

Next meeting

The next phone meeting will be on Thursday 3 August at 11:00 CET (10:00 GMT).