search

cms-l1t-offline / cmssw

CMS Offline Software
cms-sw.github.io/cmssw
Apache License 2.0
17 stars 26 forks source link

L1T_ZeroBias not firing in MC ? #571

Open rekovic opened 7 years ago

rekovic commented 7 years ago

During the review of PR https://github.com/cms-sw/cmssw/pull/19235, noticed that L1T_ZeroBias algorithm not firing in the L1T menu when running runTheMatrix workflow 10024.0.

Reproduced the problem with one of the latest IBs + the PR 19235

cmsrel CMSSW_9_2_X_2017-06-23-2300
cd CMSSW_9_2_X_2017-06-23-2300/src
cmsenv
git cms-init; 
git cms-merge-topic -u 19235
scramv1 b -j 8
#run the matrix test workflow 10024.0
runTheMatrix.py -l 10024.0

and confirmed that L1T_ZeroBias fails to fire in workflow 10024.0, in the step2, which uses GT auto:phase1_2017_realistic

However, L1T_ZeroBias fires as expected in the L1T unit-tests. Here is an example running on 760 MC GEN-SIM sample, with the last 2 customisation used to print the L1T report of accepts:

 =================  L1 Trigger Report  =====================================================================

 L1T menu Name   : L1Menu_Collisions2016_v9
 L1T menu Version: 0.2
 L1T menu Comment: 2016 pp physics menu

    Bit                  Algorithm Name                  Init    PScd  Final   PS Factor     Num Bx Masked
============================================================================================================
       0                                L1_ZeroBias         5      5      5         4          0
franzoni commented 7 years ago

Hi @rekovic can you include your customize functions also in the relvals test, using the full suite of conditions for 2017 MC?

rekovic commented 7 years ago

@franzoni Good idea, yes.

Martin-Grunewald commented 7 years ago

Hmm, why do you think adding the customisations should fix the problem?

rekovic commented 7 years ago

@Martin-Grunewald I think this was meant as a additional information printed, not as a fix.

tmatsush commented 7 years ago

With a glance of a code, you should fix... https://github.com/cms-l1t-offline/cmssw/blob/l1t-integration-CMSSW_9_2_0/L1Trigger/L1TGlobal/plugins/L1TExtCondProducer.cc#L137-L146

The signal names have been changed due to switch of main BPTX signals from NIM to VME boards, as well as internal signal naming changes in menu. I guess you should better expose signal name to python configuration file.

fwyzard commented 7 years ago

On 27 June 2017 at 12:30, tmatsush notifications@github.com wrote:

I guess you should better expose signal name to python configuration file.

Or via a global tag payload, to make it easier to sync with the L1 menu itself.

.A

rekovic commented 7 years ago

@tmatsush Thanks! Yes, this needs to be changed from hard-coded to configuration or GT payload.

rekovic commented 7 years ago

assign @apana.

franzoni commented 7 years ago

I think this was meant as a additional information printed, not as a fix.

indeed.

rekovic commented 7 years ago

@apana provided a fix https://github.com/apana/cmssw/tree/apana-add_new_BPTX_ext_names.

rekovic commented 7 years ago

@apana I just tested your fix

git remote add apana git@github.com:apana/cmssw.git
git fetch apana
git cherry-pick bb1e171
scramv1 b -j 8

In my tests L1T_ZeroBias still does not fire.

call to customisation function L1TGlobalSimDigisSummary imported from L1Trigger.Configuration.customiseUtils

process = L1TGlobalSimDigisSummary(process)

call to customisation function L1TAddInfoOutput imported from L1Trigger.Configuration.customiseUtils

process = L1TAddInfoOutput(process)

-------------------------------------------------------------------------------

re-run the step2

cmsRun step2_DIGI_L1_DIGI2RAW_HLT.py >& output.log

I will put on hold the the PR based off the above branch, that I started testing in the meantime. https://github.com/cms-sw/cmssw/pull/19492

Martin-Grunewald commented 7 years ago

I see L1_ZeroBias firing, using the PR... see post there

rekovic commented 7 years ago

Will retest with the https://github.com/cms-sw/cmssw/pull/19492 directly merged in latest IB from last night.

rekovic commented 7 years ago

OK. Checked with CMSSW_9_3_X_2017-06-29-2300 + this PR, and now I see L1_ZeroBias fire.

 =================  L1 Trigger Report  =====================================================================

 L1T menu Name   : L1Menu_Collisions2017_dev_r5
 L1T menu Version: 0.4
 L1T menu Comment: 2017 pp developing menu : processed with UTM 20170314

    Bit                  Algorithm Name                  Init    PScd  Final   PS Factor     Num Bx Masked
============================================================================================================
     450                              L1_AlwaysTrue        10     10     10         0          0
     451                               L1_NotBptxOR         0      0      0         0          0
     452                                 L1_BptxXOR         0      0      0         0          0
     453                           L1_ZeroBias_copy        10     10     10         0          0
     454                   L1_BPTX_BeamGas_Ref1_VME         0      0      0         0          0
     455                   L1_BPTX_BeamGas_Ref2_VME         0      0      0         0          0
     456                     L1_BPTX_BeamGas_B1_VME         0      0      0         0          0
     457                     L1_BPTX_BeamGas_B2_VME         0      0      0         0          0
     458                                L1_ZeroBias        10     10     10         0          0
     459                                L1_BptxPlus        10     10     10         0          0
     460                               L1_BptxMinus        10     10     10         0          0
     461                                  L1_BptxOR        10     10     10         0          0