Migrate LST inputs to SoA collections #47793

ariostas · 2025-04-04T14:02:30Z

This PR refactors the LST inputs so that Portable SoA Collections are used from the very beginning. It also deduplicates some code since there were separate functions for standalone and CMSSW that played the same role.

It's mostly done, but we have a couple of questions (probably for @fwyzard or @makortel).

Running on CPU works fine, but I'm still having issues with GPUs. The issue is with LCG dictionaries, as I described in my Mattermost message. Initially, I thought that the ROOT name normalization issue that Matti mentioned could be the cause, but I think it's something else.
I had to split an SoA collection into two, since part of it is created before LST and some needs to be modified during the algorithm (to add some extra data). Do you know of a good way to combine the two collections into some sort of combined view, or do we just need to keep using the two separately. (Obviously, we could just define some new struct or class, but that doesn't sound great to me)

c.c. @slava77 @VourMa

cmsbuild · 2025-04-04T14:03:02Z

cms-bot internal usage

cmsbuild · 2025-04-04T14:04:38Z

-code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-47793/44377

Found files with invalid states:
- RecoTracker/LSTCore/interface/LSTInput.h:
  - Added: c202fe6
  - Modified: b39f14d, 38eca30, 0bb4f24
  - Deleted: 5cc9872
- RecoTracker/LSTCore/interface/alpaka/HitsRangesDeviceCollection.h:
  - Added: c202fe6
  - Deleted: b39f14d
- RecoTracker/LSTCore/interface/HitsRangesHostCollection.h:
  - Added: c202fe6
  - Deleted: b39f14d
- RecoTracker/LSTCore/interface/alpaka/LSTInputDeviceCollection.h:
  - Added: b39f14d
  - Deleted: 0ba4c29

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

code-checks:
https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-47793/44377/code-checks.patch
e.g. curl -k https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-47793/44377/code-checks.patch | patch -p1
You can also run scram build code-checks to apply code checks directly
code-format:
https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-47793/44377/code-format.patch
e.g. curl -k https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-47793/44377/code-format.patch | patch -p1
You can also run scram build code-format to apply code format directly

cmsbuild · 2025-04-04T14:14:37Z

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-47793/44378

Found files with invalid states:
- RecoTracker/LSTCore/interface/LSTInput.h:
  - Added: c202fe6
  - Modified: b39f14d, 38eca30, 0bb4f24
  - Deleted: 5cc9872
- RecoTracker/LSTCore/interface/alpaka/HitsRangesDeviceCollection.h:
  - Added: c202fe6
  - Deleted: b39f14d
- RecoTracker/LSTCore/interface/HitsRangesHostCollection.h:
  - Added: c202fe6
  - Deleted: b39f14d
- RecoTracker/LSTCore/interface/alpaka/LSTInputDeviceCollection.h:
  - Added: b39f14d
  - Deleted: 0ba4c29

cmsbuild · 2025-04-04T14:15:01Z

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

It involves the following packages:

DQM/TrackingMonitorSource (dqm)
HLTrigger/Configuration (hlt)
RecoTracker/IterativeTracking (reconstruction)
RecoTracker/LST (reconstruction)
RecoTracker/LSTCore (reconstruction)

@Martin-Grunewald, @antoniovagnerini, @cmsbuild, @jfernan2, @mandrenguyen, @mmusich, @rseidita can you please review it and eventually sign? Thanks.
@GiacomoSguazzoni, @Martin-Grunewald, @SohamBhattacharya, @VinInn, @VourMa, @arossi83, @dgulhan, @felicepantaleo, @fioriNTU, @gpetruc, @idebruyn, @jandrea, @missirol, @mmusich, @mtosi, @richa2710, @rovere, @sroychow, @threus this is something you requested to watch as well.
@antoniovilela, @mandrenguyen, @rappoccio, @sextonkennedy you are the release manager for this.

cms-bot commands are listed here

ariostas · 2025-04-04T14:27:50Z

Hmm I see that now there's #47697 and #47774, so these are all likely related issues

slava77 · 2025-04-07T14:46:44Z

test parameters:

enable_tests = gpu
workflows_gpu = 29634.704,29834.704
workflows = 29634.703,29834.703,29834.755
relvals_opt = -w upgrade,standard
relvals_opt_gpu = -w upgrade,standard

slava77 · 2025-04-07T15:01:28Z

@cmsbuild please test

VourMa · 2025-04-07T16:58:46Z

@slava77 I would propose we start testing also the .755 workflow to make sure that LST in Phase 2 HLT works fine.
In this particular case, I expect that workflow to crash in more than one ways, which I can follow up on and resolve once the rest of the issues are fixed.

slava77 · 2025-04-07T17:43:04Z

@slava77 I would propose we start testing also the .755 workflow to make sure that LST in Phase 2 HLT works fine. In this particular case, I expect that workflow to crash in more than one ways, which I can follow up on and resolve once the rest of the issues are fixed.

I don't see an accelerator specifier in this workflow.
Is it expected to be added to the CPU or GPU list?

VourMa · 2025-04-07T18:12:08Z

Is it expected to be added to the CPU or GPU list?

It should be in the "workflows" (not the "workflows_gpu") list.

cmsbuild · 2025-04-07T19:35:55Z

-1

Failed Tests: RelVals-CUDA RelVals-ROCM
Size: This PR adds an extra 20KB to repository
Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-a5bdf1/45418/summary.html
COMMIT: 650b643
CMSSW: CMSSW_15_1_X_2025-04-07-1100/el8_amd64_gcc12
Additional Tests: CUDA,ROCM
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week0/cms-sw/cmssw/47793/45418/install.sh to create a dev area with all the needed externals and cmssw changes.

Comparison Summary

Summary:

You potentially added 6 lines to the logs
Reco comparison results: 323 differences found in the comparisons
DQMHistoTests: Total files compared: 54
DQMHistoTests: Total histograms compared: 4206802
DQMHistoTests: Total failures: 7529
DQMHistoTests: Total nulls: 0
DQMHistoTests: Total successes: 4199253
DQMHistoTests: Total skipped: 20
DQMHistoTests: Total Missing objects: 0
DQMHistoSizes: Histogram memory added: -0.316 KiB( 53 files compared)
DQMHistoSizes: changed ( 29634.703,... ): -0.158 KiB Tracking/TrackParameters
Checked 231 log files, 200 edm output root files, 54 DQM output files
TriggerResults: no differences found

VourMa · 2025-05-07T09:39:05Z

Actually, could you squash the last commits into the previous one ?

In the last force push, I simplified the commit history by squashing together a few of the commits.

cmsbuild · 2025-05-07T09:40:14Z

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-47793/44732

There are other open Pull requests which might conflict with changes you have proposed:
- File DataFormats/Common/src/classes_def.xml modified in PR(s): SiStripClusterizer, an Alpaka port of the CUDA ClustersFromRawProducerGPU #47629
- File RecoTracker/LSTCore/src/alpaka/PixelQuintuplet.h modified in PR(s): Add pT3 DNN to LST for Improved Fake Rejection #47995
- File RecoTracker/LSTCore/src/alpaka/PixelTriplet.h modified in PR(s): Add pT3 DNN to LST for Improved Fake Rejection #47995
- File RecoTracker/LSTCore/standalone/code/core/write_lst_ntuple.cc modified in PR(s): Add pT3 DNN to LST for Improved Fake Rejection #47995

cmsbuild · 2025-05-07T09:40:34Z

Pull request #47793 was updated. @antoniovagnerini, @cmsbuild, @rseidita can you please check and sign again.

fwyzard · 2025-05-07T09:43:54Z

please test

cmsbuild · 2025-05-07T14:54:07Z

+1

Size: This PR adds an extra 44KB to repository
Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-a5bdf1/45907/summary.html
COMMIT: ecd1aed
CMSSW: CMSSW_15_1_X_2025-05-06-2300/el8_amd64_gcc12
Additional Tests: CUDA,ROCM
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week0/cms-sw/cmssw/47793/45907/install.sh to create a dev area with all the needed externals and cmssw changes.

Comparison Summary

Summary:

You potentially removed 198 lines from the logs
Reco comparison results: 4 differences found in the comparisons
DQMHistoTests: Total files compared: 57
DQMHistoTests: Total histograms compared: 4405457
DQMHistoTests: Total failures: 41
DQMHistoTests: Total nulls: 0
DQMHistoTests: Total successes: 4405396
DQMHistoTests: Total skipped: 20
DQMHistoTests: Total Missing objects: 0
DQMHistoSizes: Histogram memory added: -0.316 KiB( 56 files compared)
DQMHistoSizes: changed ( 29634.703,... ): -0.158 KiB Tracking/TrackParameters
Checked 240 log files, 206 edm output root files, 57 DQM output files
TriggerResults: found differences in 3 / 55 workflows

CUDA Comparison Summary

Summary:

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

ROCM Comparison Summary

Summary:

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

antoniovagnerini · 2025-05-08T08:53:04Z

+dqm

spurious differences seen from unstable WF 29634.0 (open issue Non-reproducibility in 29634.911 #45505 )

cmsbuild · 2025-05-08T08:53:30Z

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

mandrenguyen · 2025-05-08T13:53:02Z

+1

cmsbuild added this to the CMSSW_15_1_X milestone Apr 4, 2025

cmsbuild added reconstruction-pending dqm-pending hlt-pending pending-signatures tests-pending orp-pending code-checks-pending tracking changes-dataformats labels Apr 4, 2025

cmsbuild added code-checks-rejected code-checks-pending and removed code-checks-pending code-checks-rejected labels Apr 4, 2025

cmsbuild added code-checks-approved and removed code-checks-pending labels Apr 4, 2025

cmsbuild added tests-started and removed tests-pending labels Apr 7, 2025

cmsbuild removed the tests-started label Apr 7, 2025

cmsbuild added the code-checks-pending label May 7, 2025

cmsbuild added code-checks-approved and removed code-checks-pending labels May 7, 2025

cmsbuild added tests-started and removed tests-pending labels May 7, 2025

cmsbuild added tests-approved and removed tests-started labels May 7, 2025

cmsbuild added dqm-approved fully-signed and removed dqm-pending pending-signatures labels May 8, 2025

cmsbuild added orp-approved and removed orp-pending labels May 8, 2025

cmsbuild merged commit 8b5e975 into cms-sw:master May 8, 2025
18 checks passed

This was referenced May 8, 2025

Make static analyzer errors more visible cms-sw/cms-bot#2495

Merged

[GCC14] build gcc with frame-pointers; sync changes from el10/gcc14 cms-sw/cmsdist#9836

Merged

GNiendorf mentioned this pull request May 9, 2025

LST pixelSeeds Eta Name Correction #48048

Merged

This was referenced May 9, 2025

Silence Wdangling-reference in RecoLocalCalo/EcalRecProducers alpaka plugin #48041

Merged

[CUDART] Improved cuda-runtime package cms-sw/cmsdist#9828

Merged

VourMa mentioned this pull request May 11, 2025

Follow-up tasks for the LST algorithm #46746

Open

ariostas deleted the ariostas/lst_inputs branch May 19, 2025 17:05

ariostas mentioned this pull request Jun 25, 2025

Migrate LST outputs to Portable SoAs #48409

Open

Migrate LST inputs to SoA collections #47793

Migrate LST inputs to SoA collections #47793

Uh oh!

Conversation

ariostas commented Apr 4, 2025

Uh oh!

cmsbuild commented Apr 4, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cmsbuild commented Apr 4, 2025

Uh oh!

cmsbuild commented Apr 4, 2025

Uh oh!

cmsbuild commented Apr 4, 2025

Uh oh!

ariostas commented Apr 4, 2025

Uh oh!

slava77 commented Apr 7, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

slava77 commented Apr 7, 2025

Uh oh!

VourMa commented Apr 7, 2025

Uh oh!

slava77 commented Apr 7, 2025

Uh oh!

VourMa commented Apr 7, 2025

Uh oh!

cmsbuild commented Apr 7, 2025

Comparison Summary

Uh oh!

VourMa commented May 7, 2025

Uh oh!

cmsbuild commented May 7, 2025

Uh oh!

cmsbuild commented May 7, 2025

Uh oh!

fwyzard commented May 7, 2025

Uh oh!

cmsbuild commented May 7, 2025

Comparison Summary

CUDA Comparison Summary

ROCM Comparison Summary

Uh oh!

antoniovagnerini commented May 8, 2025

Uh oh!

cmsbuild commented May 8, 2025

Uh oh!

mandrenguyen commented May 8, 2025

Uh oh!

Uh oh!

Uh oh!

cmsbuild commented Apr 4, 2025 •

edited

Loading

slava77 commented Apr 7, 2025 •

edited

Loading