Enable Alpaka backends based on the architecture

[fwyzard]

Enable the Alpaka backends for CUDA and ROCm only on the architectures that support them:

• CUDA is enabled on all architectures, as long as gcc is < 12
• ROCm is enabled only on x86

[cmsbuild]

A new Pull Request was created by @fwyzard (Andrea Bocci) for branch scramv3.

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

[fwyzard]

What we should do is something like

  <flags ALPAKA_BACKENDS="serial"/>
  <iftool name="cuda-gcc-support">
    <flags ALPAKA_BACKENDS="cuda"/>
  </iftool>
  <iftool name="rocm">
    <flags ALPAKA_BACKENDS="rocm"/>
  </iftool>

But that doesn't seem to work, so the changes in this PR currently implement the same effect, based on the architecture.

[fwyzard]

please test with cms-sw/cmssw#40832

[fwyzard]

please test with cms-sw/cmssw#40832 for el8_ppc64le_gcc11

[cmsbuild]

+1

Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-9ab3d7/30906/summary.html
COMMIT: a803314
CMSSW: CMSSW_13_1_X_2023-02-24-2300/el8_ppc64le_gcc11
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week1/cms-sw/cmssw-config/94/30906/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:

• @suchandradutta Fix to a customize python script in Phase2 Tracker Digizer  cmssw#40880
• @bsunanda Run3-gex155 Make the additional Run3 files as the step1 to make the payload for Run3 cmssw#40865

You can see more details here:
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-9ab3d7/30906/git-recent-commits.json
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-9ab3d7/30906/git-merge-result

[cmsbuild]

-1

Failed Tests: RelVals-INPUT
Summary: https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-9ab3d7/30909/summary.html
COMMIT: a803314
CMSSW: CMSSW_13_1_X_2023-02-26-0000/el8_amd64_gcc11
User test area: For local testing, you can use /cvmfs/cms-ci.cern.ch/week0/cms-sw/cmssw-config/94/30909/install.sh to create a dev area with all the needed externals and cmssw changes.

RelVals-INPUT

• 20834.020834.0_TTbar_14TeV+2026D88/step2_TTbar_14TeV+2026D88.log
• 20834.10320834.103_TTbar_14TeV+2026D88Aging3000/step2_TTbar_14TeV+2026D88Aging3000.log
• 20834.2120834.21_TTbar_14TeV+2026D88_ProdLike/step2_TTbar_14TeV+2026D88_ProdLike.log

Expand to see more relval errors ...

• 20834.5
• 20834.501
• 20834.75
• 20834.502
• 20834.76
• 20834.9
• 21034.21
• 21034.114

Comparison Summary

Summary:

• You potentially added 4 lines to the logs
• Reco comparison results: 8 differences found in the comparisons
• DQMHistoTests: Total files compared: 49
• DQMHistoTests: Total histograms compared: 3528955
• DQMHistoTests: Total failures: 6
• DQMHistoTests: Total nulls: 0
• DQMHistoTests: Total successes: 3528927
• DQMHistoTests: Total skipped: 22
• DQMHistoTests: Total Missing objects: 0
• DQMHistoSizes: Histogram memory added: 0.0 KiB( 48 files compared)
• Checked 213 log files, 164 edm output root files, 49 DQM output files
• TriggerResults: no differences found

[smuzaffar]

@fwyzard should not https://github.com/cms-sw/cmssw-config/blob/scramv3/SCRAM/Plugins/BuildRules.py#L1017-L1020 enough ? I think Self.xml can declare all backend <flags ALPAKA_BACKENDS="serial rocm cuda"/> and then scram should drop cuda if cuda-gcc-support isnot available and drop rocm backend of rocm is not available

[fwyzard]

Ah, I see... sure, that's enough.