Skip to content

Conversation

@fwyzard
Copy link
Contributor

@fwyzard fwyzard commented Jun 23, 2022

This release contains many speed and space optimizations, from micro optimizations on common paths to rework of internal data structures and locking schemes, and many more too detailed to list below. Multiple percent of system level metric improvements were measured in tested production workloads.
The release has gone through large-scale production testing.

See https://github.com/jemalloc/jemalloc/releases/tag/5.3.0 for more details.

Simplify and make more uniform the syntax of the jemalloc{,-debug,-prof}.spec files.

@fwyzard
Copy link
Contributor Author

fwyzard commented Jun 23, 2022

please test

@cmsbuild
Copy link
Contributor

A new Pull Request was created by @fwyzard (Andrea Bocci) for branch IB/CMSSW_12_5_X/master.

@smuzaffar, @aandvalenzuela, @iarspider can you please review it and eventually sign? Thanks.
@perrotta, @dpiparo, @qliphy you are the release manager for this.
cms-bot commands are listed here

@fwyzard
Copy link
Contributor Author

fwyzard commented Jun 23, 2022

please test for el8_amd64_gcc11

@fwyzard
Copy link
Contributor Author

fwyzard commented Jun 23, 2022

please test for el8_aarch64_gcc10

@fwyzard
Copy link
Contributor Author

fwyzard commented Jun 23, 2022

please test for el8_ppc64le_gcc10

@fwyzard
Copy link
Contributor Author

fwyzard commented Jun 23, 2022

please test for slc7_amd64_gcc10

@fwyzard
Copy link
Contributor Author

fwyzard commented Jun 23, 2022

@gartung FYI
The release notes mention some extra functionality for the flame graphs as well.

@cmsbuild
Copy link
Contributor

+1

Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-5b2cd5/25723/summary.html
COMMIT: 0cc974e
CMSSW: CMSSW_12_5_X_2022-06-22-2300/el8_aarch64_gcc10
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week0/cms-sw/cmsdist/7953/25723/install.sh to create a dev area with all the needed externals and cmssw changes.

@cmsbuild
Copy link
Contributor

+1

Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-5b2cd5/25726/summary.html
COMMIT: 0cc974e
CMSSW: CMSSW_12_5_X_2022-06-22-2300/el8_ppc64le_gcc10
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week0/cms-sw/cmsdist/7953/25726/install.sh to create a dev area with all the needed externals and cmssw changes.

@cmsbuild
Copy link
Contributor

+1

Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-5b2cd5/25717/summary.html
COMMIT: 0cc974e
CMSSW: CMSSW_12_5_X_2022-06-22-2300/el8_amd64_gcc10
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week0/cms-sw/cmsdist/7953/25717/install.sh to create a dev area with all the needed externals and cmssw changes.

Comparison Summary

Summary:

  • No significant changes to the logs found
  • Reco comparison results: 4 differences found in the comparisons
  • DQMHistoTests: Total files compared: 50
  • DQMHistoTests: Total histograms compared: 3659307
  • DQMHistoTests: Total failures: 8
  • DQMHistoTests: Total nulls: 0
  • DQMHistoTests: Total successes: 3659277
  • DQMHistoTests: Total skipped: 22
  • DQMHistoTests: Total Missing objects: 0
  • DQMHistoSizes: Histogram memory added: 0.0 KiB( 49 files compared)
  • Checked 208 log files, 45 edm output root files, 50 DQM output files
  • TriggerResults: no differences found

@cmsbuild
Copy link
Contributor

+1

Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-5b2cd5/25719/summary.html
COMMIT: 0cc974e
CMSSW: CMSSW_12_5_X_2022-06-22-2300/slc7_amd64_gcc10
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week0/cms-sw/cmsdist/7953/25719/install.sh to create a dev area with all the needed externals and cmssw changes.

Comparison Summary

Summary:

  • No significant changes to the logs found
  • Reco comparison results: 72981 differences found in the comparisons
  • DQMHistoTests: Total files compared: 50
  • DQMHistoTests: Total histograms compared: 3659307
  • DQMHistoTests: Total failures: 468162
  • DQMHistoTests: Total nulls: 373
  • DQMHistoTests: Total successes: 3190750
  • DQMHistoTests: Total skipped: 22
  • DQMHistoTests: Total Missing objects: 0
  • DQMHistoSizes: Histogram memory added: -0.31499999999999995 KiB( 49 files compared)
  • DQMHistoSizes: changed ( 10224.0 ): -0.063 KiB SiStrip/MechanicalView
  • DQMHistoSizes: changed ( 11834.0 ): -2.372 KiB SiStrip/MechanicalView
  • DQMHistoSizes: changed ( 250202.181 ): -0.006 KiB SiStrip/MechanicalView
  • DQMHistoSizes: changed ( 25202.0 ): 0.117 KiB SiStrip/MechanicalView
  • DQMHistoSizes: changed ( 7.3 ): 2.009 KiB SiStrip/MechanicalView
  • Checked 208 log files, 45 edm output root files, 50 DQM output files
  • TriggerResults: found differences in 14 / 49 workflows

@cmsbuild
Copy link
Contributor

+1

Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-5b2cd5/25718/summary.html
COMMIT: 0cc974e
CMSSW: CMSSW_12_5_X_2022-06-22-1100/el8_amd64_gcc11
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week0/cms-sw/cmsdist/7953/25718/install.sh to create a dev area with all the needed externals and cmssw changes.

Comparison Summary

Summary:

  • No significant changes to the logs found
  • Reco comparison results: 63130 differences found in the comparisons
  • DQMHistoTests: Total files compared: 50
  • DQMHistoTests: Total histograms compared: 3659099
  • DQMHistoTests: Total failures: 311068
  • DQMHistoTests: Total nulls: 151
  • DQMHistoTests: Total successes: 3347858
  • DQMHistoTests: Total skipped: 22
  • DQMHistoTests: Total Missing objects: 0
  • DQMHistoSizes: Histogram memory added: 61.06600000000001 KiB( 49 files compared)
  • DQMHistoSizes: changed ( 10224.0 ): -0.352 KiB SiStrip/MechanicalView
  • DQMHistoSizes: changed ( 11834.0 ): 0.527 KiB SiStrip/MechanicalView
  • DQMHistoSizes: changed ( 23234.0,... ): 6.885 KiB MTD/Tracks
  • DQMHistoSizes: changed ( 23234.0,... ): 0.742 KiB MTD/BTL
  • DQMHistoSizes: changed ( 250202.181 ): -0.117 KiB SiStrip/MechanicalView
  • DQMHistoSizes: changed ( 25202.0 ): -0.012 KiB SiStrip/MechanicalView
  • DQMHistoSizes: changed ( 312.0 ): 0.004 KiB MessageLogger/Warnings
  • Checked 208 log files, 45 edm output root files, 50 DQM output files
  • TriggerResults: found differences in 14 / 49 workflows

@smuzaffar
Copy link
Contributor

+externals

@cmsbuild
Copy link
Contributor

This pull request is fully signed and it will be integrated in one of the next IB/CMSSW_12_5_X/master IBs (tests are also fine). This pull request will now be reviewed by the release team before it's merged. @perrotta, @dpiparo, @qliphy (and backports should be raised in the release meeting by the corresponding L2)

@smuzaffar
Copy link
Contributor

@fwyzard , PR tests look good, let me know when you are done with your tests

@fwyzard
Copy link
Contributor Author

fwyzard commented Jun 26, 2022

Running a GPU-enabled HLT menu on one of the new nodes, jemalloc 5.3.0 is ~1% faster than 5.2.1.

jemalloc 5.2.1

Running 10 times over 10100 events with 8 jobs, each with 32 threads, 32 streams and 1 GPUs
   621.7 ±   3.5 ev/s

jemalloc 5.3.0

Running 10 times over 10100 events with 8 jobs, each with 32 threads, 32 streams and 1 GPUs
   629.7 ±   6.5 ev/s

@fwyzard
Copy link
Contributor Author

fwyzard commented Jun 26, 2022

I don't know if anyone is interested in measuring the impact on the simulation or the offline reconstruction.

From my point of view, we can merge the PR.

@fwyzard
Copy link
Contributor Author

fwyzard commented Jun 26, 2022

By the way, just a reminder of what is the performance without jemalloc, using cmsRunGlibC:

Running 10 times over 10100 events with 8 jobs, each with 32 threads, 32 streams and 1 GPUs
   577.8 ±   2.7 ev/s (based on 9 measurements)

Requires: libunwind
## INCLUDE jemalloc-common
### RPM external jemalloc-prof %{jemalloc_version}
Requires: libunwind
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@fwyzard , we need to keep this statement before ## INCLUDE jemalloc-common to make sure the it is defined before the %prep section defined in the jemalloc-common . Can you please revert this change ( and may be add a comment why it needs to be on top)

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

will do

This release contains many speed and space optimizations, from micro
optimizations on common paths to rework of internal data structures and
locking schemes, and many more too detailed to list below. Multiple percent
of system level metric improvements were measured in tested production
workloads. The release has gone through large-scale production testing.

See https://github.com/jemalloc/jemalloc/releases/tag/5.3.0 for more details.
@fwyzard fwyzard force-pushed the IB/CMSSW_12_5_X/master_jemalloc_530 branch from 0cc974e to 412a6f8 Compare June 26, 2022 20:41
@fwyzard
Copy link
Contributor Author

fwyzard commented Jun 26, 2022

please test

@cmsbuild
Copy link
Contributor

Pull request #7953 was updated.

@cmsbuild
Copy link
Contributor

-1

Failed Tests: UnitTests
Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-5b2cd5/25813/summary.html
COMMIT: 412a6f8
CMSSW: CMSSW_12_5_X_2022-06-26-0000/el8_amd64_gcc10
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week1/cms-sw/cmsdist/7953/25813/install.sh to create a dev area with all the needed externals and cmssw changes.

The following merge commits were also included on top of IB + this PR after doing git cms-merge-topic:

You can see more details here:
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-5b2cd5/25813/git-recent-commits.json
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-5b2cd5/25813/git-merge-result

Unit Tests

I found errors in the following unit tests:

---> test testhep_ml had ERRORS

Comparison Summary

There are some workflows for which there are errors in the baseline:
39634.9 step 4
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:

  • No significant changes to the logs found
  • Reco comparison results: 11205 differences found in the comparisons
  • DQMHistoTests: Total files compared: 50
  • DQMHistoTests: Total histograms compared: 3659995
  • DQMHistoTests: Total failures: 6454
  • DQMHistoTests: Total nulls: 5
  • DQMHistoTests: Total successes: 3653514
  • DQMHistoTests: Total skipped: 22
  • DQMHistoTests: Total Missing objects: 0
  • DQMHistoSizes: Histogram memory added: -0.039 KiB( 49 files compared)
  • DQMHistoSizes: changed ( 138.4 ): -0.008 KiB JetMET/SUSYDQM
  • DQMHistoSizes: changed ( 138.5 ): -0.004 KiB JetMET/SUSYDQM
  • DQMHistoSizes: changed ( 312.0 ): -0.004 KiB MessageLogger/Warnings
  • DQMHistoSizes: changed ( 4.53 ): -0.023 KiB JetMET/SUSYDQM
  • Checked 208 log files, 45 edm output root files, 50 DQM output files
  • TriggerResults: found differences in 3 / 49 workflows

@smuzaffar
Copy link
Contributor

please test

@cmsbuild
Copy link
Contributor

+1

Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-5b2cd5/25830/summary.html
COMMIT: 412a6f8
CMSSW: CMSSW_12_5_X_2022-06-26-2300/el8_amd64_gcc10
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week1/cms-sw/cmsdist/7953/25830/install.sh to create a dev area with all the needed externals and cmssw changes.

The following merge commits were also included on top of IB + this PR after doing git cms-merge-topic:

You can see more details here:
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-5b2cd5/25830/git-recent-commits.json
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-5b2cd5/25830/git-merge-result

Comparison Summary

Summary:

  • No significant changes to the logs found
  • Reco comparison results: 0 differences found in the comparisons
  • DQMHistoTests: Total files compared: 50
  • DQMHistoTests: Total histograms compared: 3659995
  • DQMHistoTests: Total failures: 2
  • DQMHistoTests: Total nulls: 0
  • DQMHistoTests: Total successes: 3659971
  • DQMHistoTests: Total skipped: 22
  • DQMHistoTests: Total Missing objects: 0
  • DQMHistoSizes: Histogram memory added: 0.0 KiB( 49 files compared)
  • Checked 208 log files, 45 edm output root files, 50 DQM output files
  • TriggerResults: no differences found

@smuzaffar
Copy link
Contributor

+externals

@cmsbuild
Copy link
Contributor

This pull request is fully signed and it will be integrated in one of the next IB/CMSSW_12_5_X/master IBs (tests are also fine). This pull request will now be reviewed by the release team before it's merged. @perrotta, @dpiparo, @qliphy (and backports should be raised in the release meeting by the corresponding L2)

@smuzaffar smuzaffar merged commit 644f773 into cms-sw:IB/CMSSW_12_5_X/master Jun 27, 2022
@fwyzard fwyzard deleted the IB/CMSSW_12_5_X/master_jemalloc_530 branch July 6, 2022 22:44
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.

3 participants