OWG Meeting 16-Sep-2009

From GlueXWiki
Jump to: navigation, search

Agenda


Time/Location

1:30 PM Wed 16-Sep-2009 CC F326


Announcements

Next Meeting

1:30 PM Wed 30-Sep-2009


New Action Items from this Meeting

Minutes

Attendees: Elliott W, Dave L, Alex S, Mark I, Chris C, Eugene C, Carl T, Vardan G, Ben R, Hai D, Simon T, Ed J, Beni Z, Dave D, Dave A.


Solenoid Coil Test

Elliott reported it is likely that delivery of the cryo can will be delayed a few months, but with luck this will not seriously delay completion of the coil tests. If turnaround time per coil can be kept to one month the test should complete on schedule. If six weeks are required per coil the test will end one month after the RFE date for Hall D, which should not cause a big problem. In either case we cannot slack off!


Plans for the Trigger Interface (TI) and Trigger Supervisor (TS)

Ed reported on the latest thinking concerning TI and TS development. See his slides for more details:

  • Ed first reviewed material presented at the CNU Electronics Workshop this past summer.
  • Current system can support max 128 crates, mainly due to number of crates per Trigger Distribution (TD) card (8/card times 16 slots/crate).
  • TS will support 32 level 1 trigger bits, each with its own prescale.
  • TS operation is entirely synchronous.
  • Trigger word sent to crates is 12 bits.
  • TS will support two types of partitioning, "simple" and "complex". Implementation requires no hardware changes, just appropriate FPGA firmware.
  • With simple partitioning the two DAQ systems cannot share any hardware other than the TS itself.
  • With complex partitioning the two DAQ systems can share hardware, but of course the two systems must cooperate concerning power cycling and resets.
  • Partitions cannot share TD's. I.e. all crates connected to a single TD must be in the same partition.
  • TI-only DAQ systems will be supported, max 6 crates.
  • TI will include front-panel trigger inputs.
  • Since TI driver chips are expensive, will only populate TI with as many as needed. Only one TI per DAQ system needs more than one driver chip.

A long discussion ensued on whether support for multiple simultaneous DAQ systems was a diversion, a useful feature, or a requirement. Dave D argued we should keep our "eyes on the prize" and concentrate on getting a working TI and TS as soon as possible, as partitioning is not an essential requirement. Others argued that support for multiple simultaneous DAQ running and these TI features are requirements (perhaps not for Hall D, but for other halls and test systems), that the proposed TI is little more than a scaled-up version of the existing and working TI prototype (already has same TD features built-in), and that building such a TI now would be a helpful step in figuring out how to design the final TS.


RootSpy

Dave L reported on the RootSpy package, a remote ROOT histogram viewing package. Simon has been using it for FDC analysis. Dave gave an interactive demo where he viewed histograms from another reconstruction process:

  • Combines ROOT for histogramming and cMsg for interprocess communication.
  • ROOT histogram producer receives requests, then sends histograms via cMsg to ROOT-based client for display or processing.
  • Relies on ROOT object serialization/deserialization technique described at previous Online and Offline meetings.
  • Will be used in online for display of monitoring histograms.
  • Will be used in offline to display analysis histograms as they are being filled, allowing one to stop a long job if enough data processed already, or to abort jobs if an error is discovered.
  • JANA plugin is unintrusive...it finds histograms by itself, so no registration needed.
  • Many improvements are planned, see Dave's presentation.
  • Elliott will investigate cMsg port and firewall issues.
Retrieved from "https://halldweb1.jlab.org/wiki/index.php?title=OWG_Meeting_16-Sep-2009&oldid=15247"