Skip to content

More Bandwidth reduction in HLT2 IFT SMOG2 lines

A few changes to HLT2 lines to reduce bandwidth taken in the FULL stream when injecting Argon:

  • Redistribution of lines between turbo and full according to discussions at IFT WG meeting
    • Caveat: calibration in `hlt2_ift_smog2_chargedPID` and Omega&Cascade lines are not sent to Turbo for the moment. Their rate is reduced by filtering on hlt1 trigger lines.
  • Selection and changes in the lines:
    • Increased PID tag requirement in muon id line (to reduce high rate) to 0 and min pt probe to 200. Rate reduced from 0.8kHz to 0.33kHz
    • Eta_c selection adapted to modifications in HLT1 line, changed in this MR Allen!1523 (merged). Rate is reduced from 1kHz to 0.8kHz.
    • Omega&Cascade lines: add filter on `Hlt1SMOG2L0ToppiDecision`, rate of Xi line reduced from 0.47132kHz to 0.35kHz
    • Generic lines: require `hlt1_filter_code` in single track high pT lines (requiring a SMOG2 HLT1 single track line) and in diKS and diLambda lines (requiring the hlt1 ks or lambda
    • Remove persistReco in physics line of phi->KK (saves ~0.2Gb/s of bandwidth)
    • All lines are renamed with the prefix "Hlt2IFTTurbo_" and "Hlt2IFTFull_" so the stream they belong is explicit in the name
  • Add a new minimum bias hlt2 line, passthrough on hlt1 decisions for CEP (low activity lines), filtering now on `Hlt1SMOG2PassThroughLowMult5`. This line is removed from `Hlt2IFT_SMOG2LumiPassthrough` as only `Hlt1SMOG2BELowMultElectronsDecision` will be used in the luminosity determination.

Updated estimations of rate and bandwidth:

The lines in the MR uses HLT1 lines with name changed by Allen!1523 (merged).

Note: global BW are currently not being run on a sample with Argon, so they do not include this contribution to the throughput.

I tag @samarian and @thboettc also as ift convenors so they check these changes.

@kmattiol

Edited by Oscar Boente Garcia

Merge request reports