Skip to content

Global fixes to the bandwidth of the RD stream

Miguel Ramos Pernas requested to merge rd-fix-bandwidth into master

Goals

Implementation of several changes to the selection requirements and the storage of additional output objects in the lines of the RD working group to reduce the overall throughput of the stream to around 0.5 GB/s. Related to #710.

Contributing

If you want to contribute to this MR, please speak first to @mramospe and @matzeni and explain your goals. We should avoid having clashes across different files, so if your line is defined in a module where other lines live, please coordinate among the contributors to the file to avoid conflicts. Once you are given green light to do changes, please do so and include in the description of this MR a little summary indicating the file(s) that you have modified as well as the changes to the lines.

Changes

global (@mramospe)

  • Remove ambiguity on the DOCACHI2 requirement (previously defined as adocachi2cut_min and now defined as adocachi2cut_max).
  • Set the requirements to persist candidates for isolation to F.DR2 < 0.25 in all the RD modules

b_to_v0ll_hlt2.py (@hvc)

  • PID cuts on electrons/muons tightened to PIDe/mu > 0 (previously no electron PID cut applied)
  • IsMuon=true required for all muons
  • Flight distance chi2 of Kshorts tightened to > 50 (from 15)
  • Vertex chi2 of Kshorts tightened to < 15 (from 25)
  • pT cut on electrons and muons tightened to > 500 MeV
  • pT cut on soft-pion from K*+ --> KS0 pi+ decay tightened to > 500 MeV
  • Downstream-Downstream lines added (zero rate on 20,000 Hlt1 filtered minbias MC)

b_to_xtaul_hlt2.py (@tfulghes)

  • Remove b_to_xtaul_spruce.py from the list of spruce lines, ALL HADRONIC TAU DECAYS LINES WILL BE IN TURBO STREAM
  • Remove isolation for K daughters
  • Tight requirements on hadronic tau builder, fix PIDK requirement and change other selection to be aligned with Run2
  • Tight requirements on x hadron builders
  • Tight PID requirements
  • The entire selection is collected in this comment

B_to_ll (@tmombach)

  • removed sprucing lines. At some point it can be investigated if full+sprucing is better to not loose information of the rest of the event but only persist to disk the needed part

btosetau_exclusive.py, btosmutau_exclusive.py (@lmadhanm)

  • implemented minimum mass cuts for particle combiners
  • reduced doca max to 0.12 mm for dihadron combination

btosetau_exclusive_hlt2.py, btosetau_tau_to_e_exclusive_hlt2.py, btosmutau_exclusive_hlt2.py, btosmutau_tau_to_e_exclusive_hlt2.py (@lmadhanm)

  • reduced prescales for background lines with higher rates
  • tighted PIDe > 5 (from 3)
  • Total RD rate reduction by 2 kHz
  • RD bandwidth reduced from 0.5928 gb/s to 0.4788 gb/s

builders/strange_builders.py (@mramospe)

  • Add n_candidates as a monitoring variable for all the lines for strange decays
Edited by Miguel Ramos Pernas

Merge request reports