Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[HGCal] Shrinking Layer Clustering internal data structures #30753

Merged

Conversation

felicepantaleo
Copy link
Contributor

PR description:

The memory footprint of the HGCalLayerClusterProducer grows with increasing number of events. This is solved by shrink_to_fit() after clear() when resetting these data structures.

@cmsbuild
Copy link
Contributor

The code-checks are being triggered in jenkins.

@felicepantaleo
Copy link
Contributor Author

@cmsbuild please test

@cmsbuild
Copy link
Contributor

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-30753/17072

@cmsbuild
Copy link
Contributor

cmsbuild commented Jul 16, 2020

The tests are being triggered in jenkins.

@cmsbuild
Copy link
Contributor

A new Pull Request was created by @felicepantaleo (Felice Pantaleo) for master.

It involves the following packages:

RecoLocalCalo/HGCalRecProducers

@perrotta, @jpata, @cmsbuild, @kpedro88, @slava77 can you please review it and eventually sign? Thanks.
@edjtscott, @vandreev11, @deguio, @sethzenz, @riga, @rovere, @lgray, @cseez, @apsallid, @pfs, @lecriste, @hatakeyamak, @clelange this is something you requested to watch as well.
@silviodonato, @dpiparo you are the release manager for this.

cms-bot commands are listed here

@cmsbuild
Copy link
Contributor

+1
Tested at: 763d196
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-d3b849/8010/summary.html
CMSSW: CMSSW_11_2_X_2020-07-16-1100
SCRAM_ARCH: slc7_amd64_gcc820

@cmsbuild
Copy link
Contributor

Comparison job queued.

@cmsbuild
Copy link
Contributor

Comparison is ready
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-d3b849/8010/summary.html

Comparison Summary:

  • No significant changes to the logs found
  • Reco comparison results: 0 differences found in the comparisons
  • DQMHistoTests: Total files compared: 35
  • DQMHistoTests: Total histograms compared: 2612943
  • DQMHistoTests: Total failures: 1
  • DQMHistoTests: Total nulls: 0
  • DQMHistoTests: Total successes: 2612894
  • DQMHistoTests: Total skipped: 48
  • DQMHistoTests: Total Missing objects: 0
  • DQMHistoSizes: Histogram memory added: 0.0 KiB( 34 files compared)
  • Checked 146 log files, 17 edm output root files, 35 DQM output files

@kpedro88
Copy link
Contributor

+upgrade

@cmsbuild
Copy link
Contributor

cmsbuild commented Jul 17, 2020

The tests are being triggered in jenkins.

@cmsbuild
Copy link
Contributor

Pull request #30753 was updated. @perrotta, @jpata, @cmsbuild, @kpedro88, @slava77 can you please check and sign again.

@cmsbuild
Copy link
Contributor

+1
Tested at: 0845af3
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-d3b849/8052/summary.html
CMSSW: CMSSW_11_2_X_2020-07-16-2300
SCRAM_ARCH: slc7_amd64_gcc820

@cmsbuild
Copy link
Contributor

Comparison job queued.

@slava77
Copy link
Contributor

slava77 commented Jul 17, 2020

running just the local HGCAL sequence in 8 threads I now get the peak RSS dropping from 9.0 GiB to 6.3 GiB. This is a really convincing improvement now.

@cmsbuild
Copy link
Contributor

Comparison is ready
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-d3b849/8052/summary.html

Comparison Summary:

  • No significant changes to the logs found
  • Reco comparison results: 4 differences found in the comparisons
  • DQMHistoTests: Total files compared: 35
  • DQMHistoTests: Total histograms compared: 2612943
  • DQMHistoTests: Total failures: 5
  • DQMHistoTests: Total nulls: 1
  • DQMHistoTests: Total successes: 2612889
  • DQMHistoTests: Total skipped: 48
  • DQMHistoTests: Total Missing objects: 0
  • DQMHistoSizes: Histogram memory added: -0.004 KiB( 34 files compared)
  • DQMHistoSizes: changed ( 10224.0 ): -0.004 KiB MessageLogger/Warnings
  • Checked 146 log files, 17 edm output root files, 35 DQM output files

@slava77
Copy link
Contributor

slava77 commented Jul 17, 2020

+1

for #30753 0845af3

  • code changes are in line with the PR description and the follow up review
  • jenkins tests pass and comparisons with the baseline show no (relevant) differences
  • in a local test with PU200 ttbar wf 23434.21 (D49) running just the local HGCAL sequence significant improvements are observed:
    • runtime RSS measured memory peak (from monitoring the process externally at 0.1 s sampling) goes down by 2.7 GiB on an 8-thread job and by 0.64 GiB on 1-thread job (from 2.64 GiB to 2.00 GiB)
    • CPU time of hgcalLayerClusters is down by 23% in a one-thread test (about 0.1% of total full RECO time)
    • I'll update the mem profiler report later

@kpedro88
Copy link
Contributor

+upgrade

@cmsbuild
Copy link
Contributor

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. @silviodonato, @dpiparo (and backports should be raised in the release meeting by the corresponding L2)

@silviodonato
Copy link
Contributor

+1

@cmsbuild cmsbuild merged commit f9f205e into cms-sw:master Jul 17, 2020
@silviodonato
Copy link
Contributor

runtime RSS measured memory peak (from monitoring the process externally at 0.1 s sampling) goes down by 2.7 GiB on an 8-thread job and by 0.64 GiB on 1-thread job (from 2.64 GiB to 2.00 GiB)

@tommasoboccali @fwyzard @srimanob do you want a backport of this PR to 11_1?

@kpedro88
Copy link
Contributor

@silviodonato there is already a backport in #30757

@srimanob
Copy link
Contributor

@silviodonato
RECO validation relvals for HLT TDR are running with 11_1_1. I would let it goes as it is for now and start validation call asap.

For additional PRs that affect RECO, we may review at the end for the minimal changes if benefit and see if we need limited validation for specific groups.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

6 participants