Skip to content

Enabling use of cached Views in precision Calo step

Debottam Bakshi Gupta requested to merge dbakshig/athena:enablingCachedView_02_02_22 into master

This MR enables use of Cached Views in precisionCalo step on top of changes applied to precisionTrackingHypo. The step counts for probe legs for two cases

chainComp INFO       HLT_e26_lhtight_ivarloose_e5_lhtight_probe_L1EM22VHI:
chainComp INFO         stepCounts:
chainComp INFO           7: 4 -> 3
chainComp INFO           8: 4 -> 3
chainComp INFO         stepFeatures:
chainComp INFO           8: 11 -> 10
chainComp INFO           9: 5 -> 4
chainComp INFO       HLT_e26_lhtight_ivarloose_e5_lhvloose_idperf_probe_L1EM22VHI:
chainComp INFO         eventCount: 5 -> 4
chainComp INFO         stepCounts:
chainComp INFO           7: 5 -> 4
chainComp INFO           8: 5 -> 4
chainComp INFO         stepFeatures:
chainComp INFO           8: 14 -> 13

but as discussed with @tamartin this may be that we are passing some events in error while this flag (precisionCaloViewsMaker.CacheDisabled) is true i.e. we have one candidate, and we reconstruct it twice (because we are not re-using the cached result), and then we go and pass a two-electron chain with just one electron (for example). As I locally checked with a signal sample to see how much reduction we get in probe leg by removing the flag

athena.py --imf --perfmon --threads 1 --concurrent-events 1 --filesInput '/eos/home-d/dbakshig/data17_13TeV.00339070.physics_Main/data17_13TeV.00339070.physics_Main.merge.DRAW_EGZ.f887_m1831._0152.1' --evtMax 50 --skipEvents 0 -c 'from AthenaConfiguration.AllConfigFlags import ConfigFlags;ConfigFlags.Trigger.disableCPS=True;doEmptyMenu=True;doEgammaSlice=True;isOnline=True;doWriteBS=False;doWriteRDOTrigger=True;forceEnableAllChains=True;selectChains=["HLT_e26_lhtight_ivarloose_e5_lhtight_probe_L1EM22VHI"];setGlobalTag="CONDBR2-HLTP-2018-01"' TriggerJobOpts/runHLT_standalone.py 2>&1 |tee electron_ManyEvents.log

I don't see any reduction. So propose its best to move this change for upcoming 19th round of validation and we will get to see how much its affecting in large no. of sample (adding @cjmeyer , @jodafons )

Edited by Debottam Bakshi Gupta

Merge request reports