uGT fix for muon showers full emulation and for HMT mismatches in the DQM workflow #40773

elfontan · 2023-02-15T11:15:16Z

PR description:

The purpose of this PR is to address multiple issues at the GT level related to the muon shower emulation. Here two main items are discussed:

uGT emulation of HMT
DQM mismatches

Thanks @eyigitba and @aloeliger for the work and the discussion.

uGT emulation of HMT
The MuonShowerCondition class was broken and was not evaluating at all the condition: this PR contains a first iteration of updates to get reasonable results (with a proper check of the condition for all HMT cases, able to distinguish between Nominal and Tight showers).

Summary of the main issues:

The checkObjectParameter function responsible for the check of the muon shower bit for each condition was never called in the class.
The vector of candidates in MuonShowerCondition (int numberObjects = candVec->size(useBx)) was returning four objects (as supposed to be given the splitting in four separate muon shower objects in the Global Board), but we never compare them one by one to the MuonShowerTemplate. For each condition, we should compare each object to the template to make sure one of them passes. All four objects are exclusive in which bits are set to true.
The declaration of the four muon shower objects in the GlobalBoard was wrong: without having them declared as pointers, they were not passed correctly to the MuonShowerCondition (that was receiving random objects).

Open items:

At the moment, the class is checking only the central BX: ReceiveMuonShowerObjectData is only looking at BX0. However, we can have showers in other BXs, too (as we saw last year with muon showers before fixing the timing). This has to be included.
Github issue in the OSW IB: BX Checking for Muon Showers cms-l1t-offline/cmssw#1071.
In the L1Ntuple workflow running on 1000 events using run 362720 (find the cmsDriver command below), two additional events with a nominal shower were found, and five of those events fire for a tight shower (when we would expect a nominal shower only). No mismatch when running the DQM with the below fix.
Note that L1Ntuple workflow runs the full L1 emulation starting from muonCSCDigis, so the steps in the middle that might be responsible for the discrepancy are: CSC digi unpacker, CSC shower re-emulation, EMTF shower re-emulation.
After a follow up with @kakwok, we realised that the new CSC thresholds for the HMT shower thresholds had not been updated after the deployment online: fix in Update HMT shower thresholds for HMT emulation in CSC #40829. Including this fix, a single event with a mismatch is found (having both a Nominal and Tight shower): 362720:105:115847543.
Further checks with @eyigitba for EMTF and @kakwok for CSC needed.
Github issue in the OSW IB: Muon Shower Emulator Mismatch cms-l1t-offline/cmssw#1070.

Validation:

L1 emulation run using the L1Ntuples worflow and checking the number of events triggered by HMT. Expected number of events in data: 12 L1_SingleMuShower_Nominal and 5 L1_SingleMuShower_Tight; current emulator counts: 13 L1_SingleMuShower_Nominal and 6 L1_SingleMuShower_Tight.
The testing cmsDriver command is:
cmsDriver.py l1Ntuple -s RAW2DIGI --python_filename=data.py -n 1000 --era=Run3 --data --conditions=126X_dataRun3_Prompt_v2 --customise=L1Trigger/L1TNtuples/customiseL1Ntuple.L1NtupleRAWEMU --customise=L1Trigger/Configuration/customiseReEmul.L1TReEmulFromRAW --filein=/store/data/Run2022G/EphemeralHLTPhysics0/RAW/v1/000/362/720/00000/36f350d4-8e8a-4e38-b399-77ad9bf351dc.root --nThreads 8

DQM mismatches (countings in data, zero countings in the emulator)

Strategy: Running the emulator with the DQM workflow (see instructions) for a specific run containing showers (run 362720).
Missing input for the hadronic showers in valGtStage2Digis: to be added here.
At the beginning, providing the uGT input coming from the unpacker, MuonInputTag = "gtStage2Digis:MuonShower", we still observed zero countings; switching to the GMT output MuonInputTag = "gmtStage2Digis:MuonShower", the mismatches disappeared. A fix of the GT unpacker for showers was needed: follow up with @dinyar in Fix expected version number of uGT with muon shower support #40822. Everything was tested again with this update and worked successfully (no dataVSemu mismatches in the DQM).

NOTE: An update of the script to produce test vectors for the uGT firmware-emulator comparison is also provided to rely on the unpacked uGT information as input for the muon shower objects.

PR validation:

Basic tests performed successfully starting from CMSSW_13_0_0_pre4

cmsrel CMSSW_13_0_0_pre4
cd CMSSW_13_0_0_pre4/src
cmsenv
git cms-addpkg L1Trigger/L1TGlobal
git cms-addpkg DQM/L1TMonitor
scram b distclean
git cms-checkdeps -a -A
scram b -j 8
scram b runtests
scram build code-checks
scram build code-format

cmsbuild · 2023-02-15T11:24:51Z

-code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-40773/34199

This PR adds an extra 28KB to repository

Code check has found code style and quality issues which could be resolved by applying following patch(s)

code-format:
https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-40773/34199/code-format.patch
e.g. curl -k https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-40773/34199/code-format.patch | patch -p1
You can also run scram build code-format to apply code format directly

cmsbuild · 2023-02-15T11:34:16Z

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-40773/34200

This PR adds an extra 28KB to repository

cmsbuild · 2023-02-15T11:34:44Z

A new Pull Request was created by @elfontan (Elisa Fontanesi) for master.

It involves the following packages:

DQM/L1TMonitor (dqm)
L1Trigger/L1TGlobal (l1)

@epalencia, @emanueleusai, @cmsbuild, @rekovic, @syuvivida, @pmandrik, @micsucmed, @cecilecaillol, @rvenditti can you please review it and eventually sign? Thanks.
@Martin-Grunewald, @missirol this is something you requested to watch as well.
@perrotta, @dpiparo, @rappoccio you are the release manager for this.

cms-bot commands are listed here

aloeliger · 2023-02-15T14:03:57Z

L1Trigger/L1TGlobal/src/GlobalBoard.cc

+          l1t::MuonShower* musOneNominalInTime = new l1t::MuonShower(false, false, false, false, false, false);
+          l1t::MuonShower* musOneTightInTime = new l1t::MuonShower(false, false, false, false, false, false);
+          l1t::MuonShower* musOutOfTime0 = new l1t::MuonShower(false, false, false, false, false, false);
+          l1t::MuonShower* musOutOfTime1 = new l1t::MuonShower(false, false, false, false, false, false);
+
+          musOneNominalInTime->setOneNominalInTime(mu->isOneNominalInTime());
+          musOneTightInTime->setOneTightInTime(mu->isOneTightInTime());
+          musOutOfTime0->setMusOutOfTime0(mu->musOutOfTime0());
+          musOutOfTime1->setMusOutOfTime1(mu->musOutOfTime1());
+
+          (*m_candL1MuShower).push_back(0, &(*musOneNominalInTime));
+          (*m_candL1MuShower).push_back(0, &(*musOneTightInTime));
+          (*m_candL1MuShower).push_back(0, &(*musOutOfTime0));
+          (*m_candL1MuShower).push_back(0, &(*musOutOfTime1));
+


I assume that using a std::unique_ptr is not possible here because of the .h function interface here?

cmssw/L1Trigger/L1TGlobal/interface/GlobalBoard.h

Line 153 in 29ecd54

inline const BXVector<const l1t::MuonShower*>* getCandL1MuShower() const { return m_candL1MuShower; }

If it is, I think that might be better, because I am concerned that the allocated musOneNominalInTime, musOneTightInTime,musOutOfTime0, & musOutOfTime1 may be leaking memory. They are put into the BXVector m_candL1MuShower but I don't see anything that explicitly de-allocates this allocated memory, and I don't see anything inside the BXVector code that would do it either.

Later clear is called here:

cmssw/L1Trigger/L1TGlobal/src/GlobalBoard.cc

Line 1094 in 29ecd54

m_candL1MuShower->clear();

but this more or less boils down to calling a std::vector.clear() I think, which may call destructors, but is not equivalent to delete on the objects to the best of my knowledge. There is also BX range setting here:

cmssw/L1Trigger/L1TGlobal/src/GlobalBoard.cc

Line 1095 in 29ecd54

m_candL1MuShower->setBXRange(m_bxFirst_, m_bxLast_);

But I don't think that has an effect if the BX numbers are the same as before, (isn't -2,-1,0,1,2 pretty typical for a BX vector?)

cmssw/DataFormats/L1Trigger/interface/BXVector.icc

Lines 20 to 45 in 08111d0

void BXVector<T>::setBXRange(int bxFirst, int bxLast) {

int bxF = bxFirst_;

int bxL = bxLast_;

if (bxFirst < bxFirst_) {

for (int i = 0; i < bxF - bxFirst; i++) {

itrs_.insert(itrs_.begin(), itrs_[0]);

}

}

if (bxFirst > bxFirst_) {

for (int i = 0; i < bxFirst - bxF; i++) {

deleteBX(bxF + i);

}

}

if (bxLast > bxLast_) {

for (int i = 0; i < bxLast - bxL; i++) {

addBX();

}

}

if (bxLast < bxLast_) {

for (int i = 0; i < bxL - bxLast; i++) {

deleteBX(bxL - i);

}

}

bxFirst_ = bxFirst;

bxLast_ = bxLast;

}

Perhaps I am missing something though. Is there something I am not seeing that makes sure that the memory allocated for musOneNominalInTime, musOneTightInTime,musOutOfTime0, and musOutOfTime1 is later de-allocated?

Hi @aloeliger , these changes were mostly done by me rather quickly. I didn't try using std::unique_ptr. This is only one possible fix to this issue. Anybody who is more of an expert on uGT emulator should look into this to see if there are more efficient solutions. I'm also not a huge fan of splitting one object to four objects which creates this problem to begin with. Other uGT objects are just pushed back to the vector just by their iterator.

epalencia · 2023-02-16T16:54:50Z

Please test

cmsbuild · 2023-02-16T19:31:08Z

+1

Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-87b664/30674/summary.html
COMMIT: 29ecd54
CMSSW: CMSSW_13_1_X_2023-02-16-1100/el8_amd64_gcc11
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week0/cms-sw/cmssw/40773/30674/install.sh to create a dev area with all the needed externals and cmssw changes.

Comparison Summary

Summary:

You potentially added 16 lines to the logs
Reco comparison results: 13 differences found in the comparisons
DQMHistoTests: Total files compared: 49
DQMHistoTests: Total histograms compared: 3556272
DQMHistoTests: Total failures: 12
DQMHistoTests: Total nulls: 0
DQMHistoTests: Total successes: 3556238
DQMHistoTests: Total skipped: 22
DQMHistoTests: Total Missing objects: 0
DQMHistoSizes: Histogram memory added: 0.0 KiB( 48 files compared)
Checked 213 log files, 164 edm output root files, 49 DQM output files
TriggerResults: no differences found

dinyar · 2023-02-20T14:16:46Z

Hi @elfontan,

I've put the fix of the uGT unpacker in #40822. Can you check if it works when run together with your changes? Once it's ok I can also create backports. (I assume to 13_0_X and 13_1_X?)

Cheers,
Dinyar

dinyar · 2023-02-20T14:17:33Z

DQM/L1TMonitor/python/L1TStage2Emulator_cff.py

@@ -130,6 +130,7 @@
 valGtStage2Digis = simGtStage2Digis.clone(
    ExtInputTag = "gtStage2Digis",
    MuonInputTag = "gtStage2Digis:Muon",
+    MuonShowerInputTag = "gmtStage2Digis:MuonShower",


Once #40822 is confirmed working and merged I think one should use the gt digis here, right?

elfontan · 2023-02-20T14:20:16Z

Thanks so much @dinyar! Sure, I am going to finalize some updates today/tomorrow and I will test your fix together with this PR, and update the MuonShowerInputTag in DQM/L1TMonitor/python/L1TStage2Emulator_cff.py before moving this one to Ready for review.

…ndrew)

cmsbuild · 2023-02-23T13:52:27Z

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-40773/34326

This PR adds an extra 40KB to repository

cmsbuild · 2023-02-23T13:52:47Z

Pull request #40773 was updated. @aloeliger, @epalencia, @emanueleusai, @cmsbuild, @syuvivida, @pmandrik, @micsucmed, @cecilecaillol, @rvenditti can you please check and sign again.

aloeliger · 2023-02-23T13:59:33Z

please test

cmsbuild · 2023-02-23T16:27:00Z

+1

Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-87b664/30839/summary.html
COMMIT: 4ab1cdb
CMSSW: CMSSW_13_1_X_2023-02-23-1100/el8_amd64_gcc11
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week1/cms-sw/cmssw/40773/30839/install.sh to create a dev area with all the needed externals and cmssw changes.

Comparison Summary

There are some workflows for which there are errors in the baseline:
22034.911 step 1
The results for the comparisons for these workflows could be incomplete
This means most likely that the IB is having errors in the relvals.The error does NOT come from this pull request

Summary:

You potentially added 14 lines to the logs
Reco comparison results: 9 differences found in the comparisons
DQMHistoTests: Total files compared: 49
DQMHistoTests: Total histograms compared: 3528709
DQMHistoTests: Total failures: 6
DQMHistoTests: Total nulls: 0
DQMHistoTests: Total successes: 3528681
DQMHistoTests: Total skipped: 22
DQMHistoTests: Total Missing objects: 0
DQMHistoSizes: Histogram memory added: 0.0 KiB( 48 files compared)
Checked 213 log files, 164 edm output root files, 49 DQM output files
TriggerResults: no differences found

emanueleusai · 2023-02-24T08:20:08Z

+1

emanueleusai · 2023-02-24T08:21:29Z

type l1t

aloeliger · 2023-02-24T08:52:34Z

+l1

cmsbuild · 2023-02-24T08:52:55Z

This pull request is fully signed and it will be integrated in one of the next master IBs (tests are also fine). This pull request will now be reviewed by the release team before it's merged. @perrotta, @dpiparo, @rappoccio (and backports should be raised in the release meeting by the corresponding L2)

perrotta · 2023-02-24T09:37:53Z

+1

Muon shower fix in the GT and temporary fix for the DQM workflow

ca86a77

cmsbuild added this to the CMSSW_13_1_X milestone Feb 15, 2023

cmsbuild added dqm-pending l1-pending pending-signatures tests-pending orp-pending code-checks-pending labels Feb 15, 2023

cmsbuild added code-checks-rejected and removed code-checks-pending labels Feb 15, 2023

Muon shower fix in the GT: code-format checks

29ecd54

cmsbuild added code-checks-pending and removed code-checks-rejected labels Feb 15, 2023

cmsbuild added code-checks-approved and removed code-checks-pending labels Feb 15, 2023

aloeliger reviewed Feb 15, 2023

View reviewed changes

cmsbuild added tests-started and removed tests-pending labels Feb 16, 2023

cmsbuild added tests-approved and removed tests-started labels Feb 16, 2023

dinyar reviewed Feb 20, 2023

View reviewed changes

kakwok mentioned this pull request Feb 21, 2023

[13_0_X]Update HMT shower thresholds for HMT emulation in CSC #40839

Merged

Fix to create the four different muon shower objects as shared_ptr (A…

30146a2

…ndrew)

cmsbuild added code-checks-approved and removed code-checks-pending labels Feb 23, 2023

This was referenced Feb 23, 2023

Muon Shower Emulator Mismatch cms-l1t-offline/cmssw#1070

Open

BX Checking for Muon Showers cms-l1t-offline/cmssw#1071

Closed

elfontan changed the title ~~uGT fix for muon showers and temporary fix in the DQM workflow for HMT mismatches~~ uGT fix for muon showers full emulation and for HMT mismatches in the DQM workflow Feb 23, 2023

cmsbuild added tests-started and removed tests-pending labels Feb 23, 2023

elfontan mentioned this pull request Feb 23, 2023

[Backport] uGT fix for muon showers full emulation and for HMT mismatches in the DQM workflow #40861

Merged

cmsbuild added tests-approved and removed tests-started labels Feb 23, 2023

cmsbuild added dqm-approved and removed dqm-pending labels Feb 24, 2023

cmsbuild added the l1t label Feb 24, 2023

cmsbuild added l1-approved fully-signed and removed l1-pending pending-signatures labels Feb 24, 2023

cmsbuild added orp-approved and removed orp-pending labels Feb 24, 2023

cmsbuild merged commit 7a8c07a into cms-sw:master Feb 24, 2023

epalencia mentioned this pull request Mar 1, 2023

uGT fix for muon showers full emulation and for HMT mismatches in the DQM workflow (backport of 40773) cms-l1t-offline/cmssw#1072

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

uGT fix for muon showers full emulation and for HMT mismatches in the DQM workflow #40773

uGT fix for muon showers full emulation and for HMT mismatches in the DQM workflow #40773

elfontan commented Feb 15, 2023 •

edited

Loading

cmsbuild commented Feb 15, 2023

cmsbuild commented Feb 15, 2023

cmsbuild commented Feb 15, 2023

aloeliger Feb 15, 2023

eyigitba Feb 15, 2023

epalencia commented Feb 16, 2023

cmsbuild commented Feb 16, 2023

dinyar commented Feb 20, 2023

dinyar Feb 20, 2023

elfontan commented Feb 20, 2023

cmsbuild commented Feb 23, 2023

cmsbuild commented Feb 23, 2023

aloeliger commented Feb 23, 2023

cmsbuild commented Feb 23, 2023

emanueleusai commented Feb 24, 2023

emanueleusai commented Feb 24, 2023

aloeliger commented Feb 24, 2023

cmsbuild commented Feb 24, 2023

perrotta commented Feb 24, 2023

	void BXVector<T>::setBXRange(int bxFirst, int bxLast) {
	int bxF = bxFirst_;
	int bxL = bxLast_;
	if (bxFirst < bxFirst_) {
	for (int i = 0; i < bxF - bxFirst; i++) {
	itrs_.insert(itrs_.begin(), itrs_[0]);
	}
	}
	if (bxFirst > bxFirst_) {
	for (int i = 0; i < bxFirst - bxF; i++) {
	deleteBX(bxF + i);
	}
	}
	if (bxLast > bxLast_) {
	for (int i = 0; i < bxLast - bxL; i++) {
	addBX();
	}
	}
	if (bxLast < bxLast_) {
	for (int i = 0; i < bxL - bxLast; i++) {
	deleteBX(bxL - i);
	}
	}
	bxFirst_ = bxFirst;
	bxLast_ = bxLast;
	}

uGT fix for muon showers full emulation and for HMT mismatches in the DQM workflow #40773

uGT fix for muon showers full emulation and for HMT mismatches in the DQM workflow #40773

Conversation

elfontan commented Feb 15, 2023 • edited Loading

PR description:

PR validation:

cmsbuild commented Feb 15, 2023

cmsbuild commented Feb 15, 2023

cmsbuild commented Feb 15, 2023

aloeliger Feb 15, 2023

Choose a reason for hiding this comment

eyigitba Feb 15, 2023

Choose a reason for hiding this comment

epalencia commented Feb 16, 2023

cmsbuild commented Feb 16, 2023

Comparison Summary

dinyar commented Feb 20, 2023

dinyar Feb 20, 2023

Choose a reason for hiding this comment

elfontan commented Feb 20, 2023

cmsbuild commented Feb 23, 2023

cmsbuild commented Feb 23, 2023

aloeliger commented Feb 23, 2023

cmsbuild commented Feb 23, 2023

Comparison Summary

emanueleusai commented Feb 24, 2023

emanueleusai commented Feb 24, 2023

aloeliger commented Feb 24, 2023

cmsbuild commented Feb 24, 2023

perrotta commented Feb 24, 2023

elfontan commented Feb 15, 2023 •

edited

Loading