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

XGBoost producer bugfix and unit test fix for non-x86_64 (14_1_X) #44531

Merged
merged 2 commits into from
Mar 31, 2024

Conversation

smorovic
Copy link
Contributor

@smorovic smorovic commented Mar 25, 2024

It has been reported that inference in the unit test does not pass on ARM and PPC architectures, so, in agreement with HLT, we leavel the unit test check only for x86_64 (which is sufficient for HLT workloads).

Below is the log of tests which were failing on ARM64 and ppc 64 LE. This always with same numerical values on these two architecture. Cause of that will be investigated.

src/RecoEgamma/PhotonIdentification/test/test_PhotonMvaXgb.cc:44: FAILED:
  CHECK_THAT( xgbScore, Catch::Matchers::WithinAbs(mva_score_v1[i], 0.0001) )
with expansion:
  0.91074f is within 0.0001 of 0.9863399863

src/RecoEgamma/PhotonIdentification/test/test_PhotonMvaXgb.cc:44: FAILED:
  CHECK_THAT( xgbScore, Catch::Matchers::WithinAbs(mva_score_v1[i], 0.0001) )
with expansion:
  0.82656f is within 0.0001 of 0.9750099778

src/RecoEgamma/PhotonIdentification/test/test_PhotonMvaXgb.cc:44: FAILED:
  CHECK_THAT( xgbScore, Catch::Matchers::WithinAbs(mva_score_v1[i], 0.0001) )
with expansion:
  0.0f is within 0.0001 of 0.00179

src/RecoEgamma/PhotonIdentification/test/test_PhotonMvaXgb.cc:44: FAILED:
  CHECK_THAT( xgbScore, Catch::Matchers::WithinAbs(mva_score_v1[i], 0.0001) )
with expansion:
  0.93808f is within 0.0001 of 0.9837399721

UPDATE: another commit was pushed. This fixes another problem with the XGBoost photon producer. It was not scaling e2x2 variable by SC energy (s4). It should not impact the unit test.

PR validation:

Unit test unchanged for x86_64 and should pass on all architectures now.

Update: found to be caused by using abs which behaves differently on ARM and PPC.
Fix which changes to std::abs has been pushed.

@cmsbuild
Copy link
Contributor

cmsbuild commented Mar 25, 2024

cms-bot internal usage

@cmsbuild
Copy link
Contributor

-code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-44531/39663

  • This PR adds an extra 12KB to repository

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

@cmsbuild
Copy link
Contributor

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-44531/39664

  • This PR adds an extra 12KB to repository

@cmsbuild
Copy link
Contributor

A new Pull Request was created by @smorovic for master.

It involves the following packages:

  • RecoEgamma/PhotonIdentification (reconstruction)

@jfernan2, @cmsbuild, @mandrenguyen can you please review it and eventually sign? Thanks.
@valsdav, @afiqaize, @lgray, @jainshilpi, @sameasy, @sobhatta, @ram1123, @a-kapoor, @varuns23, @Prasant1993, @Sam-Harper, @missirol this is something you requested to watch as well.
@sextonkennedy, @antoniovilela, @rappoccio you are the release manager for this.

cms-bot commands are listed here

@mmusich
Copy link
Contributor

mmusich commented Mar 25, 2024

@cmsbuild, please test

@cmsbuild
Copy link
Contributor

+1

Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-a3e1d4/38384/summary.html
COMMIT: 8455fbc
CMSSW: CMSSW_14_1_X_2024-03-25-1100/el8_amd64_gcc12
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week0/cms-sw/cmssw/44531/38384/install.sh to create a dev area with all the needed externals and cmssw changes.

Comparison Summary

Summary:

… ARM and PPC architectures.

Apparently "abs" is a floating point variant round only on x86_64, so converting it to std::abs
@cmsbuild
Copy link
Contributor

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-44531/39698

  • This PR adds an extra 16KB to repository

@cmsbuild
Copy link
Contributor

Pull request #44531 was updated. @jfernan2, @mandrenguyen, @cmsbuild can you please check and sign again.

@smorovic
Copy link
Contributor Author

@cmsbuild please test

@smorovic smorovic changed the title Unit test for XGBoost photon inference fix for non-x86_64 (14_1_X) XGBoost producer bugfix and unit test fix for non-x86_64 (14_1_X) Mar 27, 2024
@cmsbuild
Copy link
Contributor

+1

Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-a3e1d4/38460/summary.html
COMMIT: b8e5783
CMSSW: CMSSW_14_1_X_2024-03-27-1100/el8_amd64_gcc12
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week0/cms-sw/cmssw/44531/38460/install.sh to create a dev area with all the needed externals and cmssw changes.

Comparison Summary

Summary:

@mandrenguyen
Copy link
Contributor

type bug-fix

@mandrenguyen
Copy link
Contributor

+1

@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. @rappoccio, @sextonkennedy, @antoniovilela (and backports should be raised in the release meeting by the corresponding L2)

@mandrenguyen
Copy link
Contributor

type egamma

@antoniovilela
Copy link
Contributor

+1

@cmsbuild cmsbuild merged commit 3f1d5e4 into cms-sw:master Mar 31, 2024
11 checks passed
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.

5 participants