Fall 2016 High Intensity/L3 beam test
From GlueXWiki
Plan
High Intensity
Goals:
- Test ability to read out events and build them at high rate
- Get sample of events taken at various luminosities to allow:
- mapping of component data rates vs. luminosity
- testing of L3 algorithms offline using real data occupancies
Issues:
- Single stream over IB limited to 1.5GB/s
- Disk writing limited to ~1GB/sec (depends on block size)
Requires:
- For anticipated data rates we will need multiple streams into SEB.
- To record data using single ER we may need to run ER on same node as SEB
- High rates may require use of EMU sockets between SEB and ER
- requires modification of CODA configuration
- preclude EPICS/BOR from datastream, and online monitoring
Detailed plan:
- Take data as close as is possible to 1x108 but with dumping the data at the SEB level.
- Record event rate, livetime after running for at least 10 minutes.
- Repeat at 80%, 60%, 40%, and 20% beam current
- Repeat above but with writing data to disk (livetimes will likely be very low)
L3
Goals:
- Test L3 infrastructure using real beam conditions
- quantify compute resources required sans L3 algorithm
- Test ability to compress data at ROC level and maintain high rate