Difference between revisions of "OWG meeting 11-Jul-2008"
m |
m |
||
| (5 intermediate revisions by the same user not shown) | |||
| Line 1: | Line 1: | ||
=Agenda= | =Agenda= | ||
| − | * | + | * Informal discussion on DAQ status - Graham H, Dave A. |
| Line 48: | Line 48: | ||
'''Online Farm in Computer Center''' | '''Online Farm in Computer Center''' | ||
| − | We had a long discussion on | + | We had a long discussion on whether it makes sense to locate the H-D online farm and perhaps other online computers in the CC. |
| + | The idea is to save money on sys admins, spares, and infrastructure. We identified many pros and cons: | ||
| − | * | + | * No savings on sys admins, as we would be charged for "work for others." |
| − | * | + | * But may be more efficient, as many sys admins might share the load. |
| − | * | + | * Have to identify a clear advantage to H-D to do this - none identified yet. |
| − | * | + | * Will we get 24/7 response from CC sys admins? |
| − | * | + | * What would be the role of the H-D sys admin? |
| − | * | + | * What might be located in the CC: farm, EB's and ER's, other stuff? Dave showed two possible architectures: [[Media:online_11-jul-2008_DAQ1.pdf]] and [[Media:online_11-jul-2008_DAQ2.pdf]]. Others are possible. |
| − | * | + | * How to get 3 GB/s from H-D to CC, and what does it cost? |
Elliott will look into cost of transport, Graham will work on manpower issues related to this. | Elliott will look into cost of transport, Graham will work on manpower issues related to this. | ||
| + | |||
| + | Later...networking guys (Bryan and Brent) say that moving 3 GB/s from the CH to the CC will not be a problem at the time we might need it. The network infrastructure at that time will easily handle this data rate with no modifications. | ||
Latest revision as of 11:05, 14 July 2008
Contents
Agenda
- Informal discussion on DAQ status - Graham H, Dave A.
Time/Location
1pm Fri 11-Jul-2008 third floor CC hallway across from Elke's office, near the whiteboard. We ended up in F227.
Announcements
Next Meeting
25-Jul-2008 1pm CC F326
New Action Items from this Meeting
Minutes
Attendees: Elliott W, Graham H, Dave A, Simon T, Carl T, Fernando B, Jim S, Alex S.
DAQ Progress
Graham gave a quick overview of recent DAQ progress. Many DAQ group members have been on vacation recently, though.
- EMU
- tested using simulated ROC input and real data transport.
- shell written in Java, calls C++ routines via JNI.
- receives and builds ok, now needs to work on send module.
- implements runtime choice of in/build/out modules.
- full prototype expected by Sep 2008, on schedule.
- needs to work out strategy for global configuration change.
- 200 kHz test
- hardware coming along - TI, FADC
- readout libs in good shape
- two-crate test by Fall 2008
- CODA 3 ROC is moving along, but slowly
- Manpower needed - admin agreed to new hires but no money yet. No hires means slower progress.
- Linux ROC readout libs - problem with manufacturer's VME library, no Endian swapping! Hope for correct version soon, if not can kludge something up.
Online Farm in Computer Center
We had a long discussion on whether it makes sense to locate the H-D online farm and perhaps other online computers in the CC. The idea is to save money on sys admins, spares, and infrastructure. We identified many pros and cons:
- No savings on sys admins, as we would be charged for "work for others."
- But may be more efficient, as many sys admins might share the load.
- Have to identify a clear advantage to H-D to do this - none identified yet.
- Will we get 24/7 response from CC sys admins?
- What would be the role of the H-D sys admin?
- What might be located in the CC: farm, EB's and ER's, other stuff? Dave showed two possible architectures: Media:online_11-jul-2008_DAQ1.pdf and Media:online_11-jul-2008_DAQ2.pdf. Others are possible.
- How to get 3 GB/s from H-D to CC, and what does it cost?
Elliott will look into cost of transport, Graham will work on manpower issues related to this.
Later...networking guys (Bryan and Brent) say that moving 3 GB/s from the CH to the CC will not be a problem at the time we might need it. The network infrastructure at that time will easily handle this data rate with no modifications.
