Add NNPuppiTauModel external spec and toolfile

[artlbv]

Summary

Adds the SCRAM build/install plumbing for the new cms-hls4ml/NNPuppiTauModel external package, following the same precedent already used for L1TSC4NGJetModel/AXOL1TL. This package replaces the in-tree, un-namespaced hls4ml weight arrays (w2, b2, ..., w20, b20) previously compiled directly into L1Trigger/Phase2L1ParticleFlow's TauNNIdHW, which were vulnerable to ELF symbol interposition when multiple hls4ml models were loaded into the same process (the bug class behind cms-sw/cmssw#49632).

This is part of a 3-PR migration:

1. New external repo: cms-hls4ml/NNPuppiTauModel — the externally-built package itself.
2. This PR — cmsdist spec + toolfile so SCRAM builds/installs the new external.
3. [Phase-2 L1T] Move TauNNIdHW to the externally-built NNPuppiTauModel package cmssw#51263 — cmssw changes to consume it via hls4mlEmulator::ModelLoader/hls4mlEmulator::Model.

Changes

• NNPuppiTauModel.spec (new) — RPM spec for the external, modeled on L1TSC4NGJetModel.spec/AXOL1TL.spec: fetches the release tarball from cms-hls4ml/NNPuppiTauModel, builds via make against hls4mlEmulatorExtras/hls, installs into $PREFIX.
• scram-tools.file/tools/NNPuppiTauModel/nnpuppitaumodel.xml (new) — SCRAM toolfile exposing NNPUPPITAUMODEL_BASE/LIBDIR environment variables for the new external, modeled on l1tsc4ngjetmodel.xml.
• cmssw-tools.spec — adds Requires: NNPuppiTauModel alongside the existing L1-trigger model externals (L1TSC4NGJetModel, L1TSC82ProngJetModel, etc.), so it's pulled into every CMSSW release.

Dependencies

Requires a tagged release of cms-hls4ml/NNPuppiTauModel to exist before this can build successfully.

PR validation

cms-bot code-checks will perform the first real build test of the spec/toolfile/package Makefile chain. Once this external is available in an IB, cms-sw/cmssw#51263 can build against it.

[cmsbuild]

cms-bot internal usage

[cmsbuild]

A new Pull Request was created by @artlbv for branch IB/CMSSW_20_1_X/master.

@akritkbehera, @cmsbuild, @iarspider, @raoatifshad, @smuzaffar can you please review it and eventually sign? Thanks.
@ftenchini, @mandrenguyen, @sextonkennedy you are the release manager for this.
cms-bot commands are listed here

[smuzaffar]

please test

[cmsbuild]

Pull request #10665 was updated.

[artlbv]

Ah the CMS-hls4ml repo does not yet exist! (Only my personal does exist.. @quinnanm @BenjaminRS @thesps could you create one?
The tests are useless otherwise

[smuzaffar]

@artlbv , https://github.com/cms-hls4ml is not managed by us.

[quinnanm]

The problem with NNPuppiTauModel is that it committed the hls4ml directly into cmssw (before that was unsupported), so cms-hls4ml/NNPuppiTauModel didn't exist and needed to be created, and the hls deleted from cms-sw needed to be migrated there. I created cms-hls4ml/NNPuppiTauModel but it is currently empty. Once that is filled and a release is made it will work as an external here and the cms-sw PR will work.

[artlbv]

I copied over the HLS code from my private repo into https://github.com/cms-hls4ml/NNPuppiTauModel

[smuzaffar]

thanks @artlbv , you also need to create the tag v1.0.0

[artlbv]

thanks @artlbv , you also need to create the tag v1.0.0

Just did it !

[artlbv]

@smuzaffar could you check/test please? thanks!

[quinnanm]

please test with cms-sw/cmssw#51263

[cmsbuild]

Pull request #10665 was updated.

[artlbv]

I accidentally made a spurious commit that I had then reset so the tests are luckily still valid!

now we need to wait for the trigger results to reappear for us to know if the PR worked fine

[smuzaffar]

assign l1

[smuzaffar]

+externals

[cmsbuild]

New categories assigned: l1

@BenjaminRS,@quinnanm you have been requested to review this Pull request/Issue and eventually sign? Thanks

[smuzaffar]

@BenjaminRS,@quinnanm any objections on getting this in 20.1.X IBs/releases ?

[smuzaffar]

please test for el9_amd64_gcc16

[quinnanm]

@BenjaminRS,@quinnanm any objections on getting this in 20.1.X IBs/releases ?

no, that's fine

[quinnanm]

+l1

[cmsbuild]

This pull request is fully signed and it will be integrated in one of the next IB/CMSSW_20_1_X/master IBs (tests are also fine). This pull request will now be reviewed by the release team before it's merged. @ftenchini, @sextonkennedy, @mandrenguyen (and backports should be raised in the release meeting by the corresponding L2)

[cmsbuild]

-1

Failed Tests: UnitTests
Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-b7aabe/54446/summary.html
COMMIT: bbc14f0
CMSSW: CMSSW_20_1_X_2026-06-30-2300/el9_amd64_gcc16
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week0/cms-sw/cmsdist/10665/54446/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:

• @fwyzard Update the NVIDIA compatibility and stub libraries for CUDA 13.3 #10678
• @fwyzard Update UCX to version 1.21.0 #10682
• @cms-sw Update tag for L1Trigger-L1TMuon to V01-13-00 #10690

You can see more details here:
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-b7aabe/54446/git-recent-commits.json
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-b7aabe/54446/git-merge-result

Failed Unit Tests

I found 1 errors in the following unit tests:

---> test RecoTrackerLSTCore-standalone-compilation had ERRORS

Comparison Summary

Summary:

• You potentially added 372 lines to the logs
• Reco comparison results: 80758 differences found in the comparisons
• DQMHistoTests: Total files compared: 45
• DQMHistoTests: Total histograms compared: 3414477
• DQMHistoTests: Total failures: 450667
• DQMHistoTests: Total nulls: 358
• DQMHistoTests: Total successes: 2963434
• DQMHistoTests: Total skipped: 18
• DQMHistoTests: Total Missing objects: 0
• DQMHistoSizes: Histogram memory added: 5.865 KiB( 44 files compared)
• DQMHistoSizes: changed ( 1000.0,... ): 0.004 KiB JetMET/SUSYDQM
• DQMHistoSizes: changed ( 2022.0010001 ): -0.008 KiB JetMET/SUSYDQM
• DQMHistoSizes: changed ( 2023.0020001,... ): -0.012 KiB JetMET/SUSYDQM
• DQMHistoSizes: changed ( 2024.0000001 ): -0.020 KiB JetMET/SUSYDQM
• DQMHistoSizes: changed ( 2024.0030001 ): -0.016 KiB JetMET/SUSYDQM
• DQMHistoSizes: changed ( 2024.0040001,... ): -0.004 KiB JetMET/SUSYDQM
• DQMHistoSizes: changed ( 2024.0050001 ): -0.027 KiB JetMET/SUSYDQM
• DQMHistoSizes: changed ( 7.3 ): 5.960 KiB SiStrip/MechanicalView
• Checked 195 log files, 163 edm output root files, 45 DQM output files
• TriggerResults: found differences in 12 / 44 workflows

[artlbv]

does this error have anything to do with this PR? @smuzaffar it seems some other PRs (one with a failed test) entered this test round

[smuzaffar]

no @artlbv , this error already exists in GCC16 IBs ( https://cmssdt.cern.ch/SDT/cgi-bin/logreader/el9_amd64_gcc16/CMSSW_20_1_X_2026-06-30-2300/unitTestLogs/RecoTracker/LSTCore#/ )