Allow Runtime Number of Hits for Alpaka Pixel Reconstruction

[AdrianoDee]

This PR proposes a solution for #44769. The culprit there is

cmssw/RecoLocalTracker/SiPixelClusterizer/plugins/alpaka/SiPixelRawToClusterKernel.dev.cc

Lines 509 to 513 in 6e82cc8

	constexpr auto MAX_HITS = TrackerTraits::maxNumberOfHits;
	for (uint32_t i : cms::alpakatools::independent_group_elements(acc, numberOfModules + 1)) {
	if (clus_view[i].clusModuleStart() > MAX_HITS)
	clus_view[i].clusModuleStart() = MAX_HITS;
	}

where we cut the number of hits to "only":

cmssw/Geometry/CommonTopologies/interface/SimplePixelTopology.h

Line 420 in 6e82cc8

static constexpr uint32_t maxNumberOfHits = 48 * 1024;

And the event of the crash has:

SiPixelClusterizerAlpaka results:
 > no. of digis: 301020
 > no. of active modules: 1794
 > no. of clusters: 50684
 > bpix2 offset: 19595

The poor man solution would be to rise the max to something safer such asstatic constexpr uint32_t maxNumberOfHits = 96 * 1024;. But seems to me a waste also because this procedure is anyway a remnant of that intermediate phase of the Alpaka port when we had no runtime sized histograms (before #43064).

So this PR involves a slightly bigger code refactoring in order to drop the fixed number of hits and use the runtime sized histograms where needed (namely for the hit to tuple map).

PR validation:

The setup in #44769 with this PR (on top of master or 14_0_X) doesn't crash anymore.

Backported to 14_0_X in #44774.

solves #44769.

[cmsbuild]

cms-bot internal usage

[mmusich]

tagging @malbouis (ORM)

[malbouis]

urgent

[malbouis]

type hlt-int

[AdrianoDee]

enable gpu

[rappoccio]

I think as soon as the checks finish we can merge in the interest of time.

[AdrianoDee]

test parameters:

• enable = gpu
• workflows = 12434.402,12434.403,12434.404
• workflows_gpu = 12434.402,12434.403,12434.404
• workflow_opts = -w upgrade
• workflow_opts_gpu = -w upgrade

[cmsbuild]

-code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-44773/40004

• This PR adds an extra 40KB to repository

• There are other open Pull requests which might conflict with changes you have proposed:

  • File RecoLocalTracker/SiPixelRecHits/plugins/alpaka/PixelRecHitKernels.dev.cc modified in PR(s): Alpaka vs CUDA DQM compare modules for pixel tracks objects #43964
  • File RecoTracker/PixelSeeding/plugins/alpaka/CAHitNtupletGeneratorKernels.h modified in PR(s): Alpaka vs CUDA DQM compare modules for pixel tracks objects #43964

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

• code-format:
  https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-44773/40004/code-format.patch
  e.g. curl -k https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-44773/40004/code-format.patch | patch -p1
  You can also run scram build code-format to apply code format directly

[cmsbuild]

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-44773/40005

• This PR adds an extra 36KB to repository

• There are other open Pull requests which might conflict with changes you have proposed:

  • File RecoLocalTracker/SiPixelRecHits/plugins/alpaka/PixelRecHitKernels.dev.cc modified in PR(s): Alpaka vs CUDA DQM compare modules for pixel tracks objects #43964
  • File RecoTracker/PixelSeeding/plugins/alpaka/CAHitNtupletGeneratorKernels.h modified in PR(s): Alpaka vs CUDA DQM compare modules for pixel tracks objects #43964

[cmsbuild]

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

It involves the following packages:

• Geometry/CommonTopologies (geometry)
• RecoLocalTracker/SiPixelClusterizer (reconstruction)
• RecoLocalTracker/SiPixelRecHits (reconstruction)
• RecoTracker/PixelSeeding (reconstruction)

@civanch, @Dr15Jones, @jfernan2, @makortel, @mdhildreth, @mandrenguyen, @cmsbuild, @bsunanda can you please review it and eventually sign? Thanks.
@threus, @mtosi, @dkotlins, @VourMa, @gpetruc, @ferencek, @felicepantaleo, @mmusich, @rovere, @dgulhan, @missirol, @JanFSchulte, @tsusa, @mroguljic, @tvami, @fabiocos, @VinInn, @GiacomoSguazzoni, @bsunanda this is something you requested to watch as well.
@rappoccio, @sextonkennedy, @antoniovilela you are the release manager for this.

cms-bot commands are listed here

[AdrianoDee]

please test

[AdrianoDee]

assign heterogeneous

[cmsbuild]

New categories assigned: heterogeneous

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

[cmsbuild]

+1

Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-bfffbf/38935/summary.html
COMMIT: 29d1688
CMSSW: CMSSW_14_1_X_2024-04-18-1100/el8_amd64_gcc12
Additional Tests: GPU
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week1/cms-sw/cmssw/44773/38935/install.sh to create a dev area with all the needed externals and cmssw changes.

Comparison Summary

Summary:

• You potentially removed 99 lines from the logs
• Reco comparison results: 49 differences found in the comparisons
• DQMHistoTests: Total files compared: 50
• DQMHistoTests: Total histograms compared: 3351672
• DQMHistoTests: Total failures: 9
• DQMHistoTests: Total nulls: 0
• DQMHistoTests: Total successes: 3351643
• DQMHistoTests: Total skipped: 20
• DQMHistoTests: Total Missing objects: 0
• DQMHistoSizes: Histogram memory added: 0.0 KiB( 49 files compared)
• Checked 213 log files, 177 edm output root files, 50 DQM output files
• TriggerResults: no differences found

GPU Comparison Summary

Summary:

• No significant changes to the logs found
• Reco comparison results: 24 differences found in the comparisons
• DQMHistoTests: Total files compared: 5
• DQMHistoTests: Total histograms compared: 71813
• DQMHistoTests: Total failures: 1281
• DQMHistoTests: Total nulls: 0
• DQMHistoTests: Total successes: 70532
• DQMHistoTests: Total skipped: 0
• DQMHistoTests: Total Missing objects: 0
• DQMHistoSizes: Histogram memory added: 0.0 KiB( 4 files compared)
• Checked 19 log files, 22 edm output root files, 5 DQM output files
• TriggerResults: no differences found

[antoniovilela]

+1
- Requested by ORM as urgent for data taking.

[antoniovilela]

merge
- Requested by ORM as urgent for data taking.

[mandrenguyen]

+1

[fwyzard]

+heterogeneous