Adds --digi flag to run_bw_test_jobs -> if args.digi download_digis_locally else download_mdfs_locally.
Changes EVENT_SIZE_UPPER_LIMIT to 400 now that we are using Digis.
Removes process-dependence on N_EVTS as the new sample has more than 1e5 events.
Removes muon_geometry_v2 from Hlt1 as is incorrect for new samples
Output + log (with build-locally) at /eos/lhcb/user/l/lugrazet/public/BWTesters/tmp-14-Nov/
ToDo:
now using .digi again instead of .mdf, need to estimate the correct EVENT_SIZE_UPPER_LIMIT needed.
downloaded the first digi file, containing 1181 events and size of ls -lh <file.digi> = 364M
avg evtsize 308 kB/evt. -> Generously rounding to 400 to account for differences between files should be fine.
Double checking that this completes for N_EVTS=1e5 + -d locally and timing it. Finished successfully (taking 88:30/90:20 to run the job).
HUGE RATES?!
Still using hlt1_pp_forward_then_matching_no_ut_no_gec, decided after some discussion with tim evans about the 'baseline' sequence before some post-Hlt1 BW Division-sequence has been implemented.
The Division should reduce this rate but still this is massive.
Attached screenshot of hlt1__rates_for_all_lines.html for reference.
Review: Ross reviewed, MM channel had no comments on the message.
Luke Grazette authored