[RFC] Make BeamSpotToCUDA constructible, and avoid building CUDADataFormats/BeamSpot when CUDA is not available according to cuda-gcc-support #31261

makortel · 2020-08-27T14:09:27Z

PR description:

Attempt to fix a build failure in gcc10, and after that to make workflows to run.

PR validation:

Building CUDADataFormats/BeamSpot succeeds (i.e. gets skipped) on gcc10.

cmsbuild · 2020-08-27T14:09:53Z

The code-checks are being triggered in jenkins.

cmsbuild · 2020-08-27T14:16:42Z

+code-checks

Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-31261/17933

This PR adds an extra 8KB to repository
There are other open Pull requests which might conflict with changes you have proposed:
- File CUDADataFormats/BeamSpot/BuildFile.xml modified in PR(s): Squash all Patatrack developments on top of CMSSW_11_3_0_pre5 #27983

cmsbuild · 2020-08-27T14:17:06Z

A new Pull Request was created by @makortel (Matti Kortelainen) for master.

It involves the following packages:

CUDADataFormats/BeamSpot

@perrotta, @makortel, @slava77, @cmsbuild, @fwyzard, @jpata can you please review it and eventually sign? Thanks.
@rovere this is something you requested to watch as well.
@silviodonato, @dpiparo, @qliphy you are the release manager for this.

cms-bot commands are listed here

makortel · 2020-08-27T14:19:17Z

test parameters:

release slc7_amd64_gcc10

makortel · 2020-08-27T14:19:24Z

@cmsbuild, please test

cmsbuild · 2020-08-27T14:19:41Z

The tests are being triggered in jenkins.

CMSSW_11_2_X_2020-08-27-1100/slc7_amd64_gcc820: https://cmssdt.cern.ch/jenkins/job/ib-run-pr-tests/8954/console Started: 2020/08/27 16:30

cmsbuild · 2020-08-27T15:38:47Z

+1
Tested at: 8d4c9eb
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-6c398b/8954/summary.html
CMSSW: CMSSW_11_2_X_2020-08-27-1100
SCRAM_ARCH: slc7_amd64_gcc820

cmsbuild · 2020-08-27T15:38:50Z

Comparison job queued.

cmsbuild · 2020-08-27T17:07:45Z

Comparison is ready
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-6c398b/8954/summary.html

Comparison Summary:

No significant changes to the logs found
Reco comparison results: 4 differences found in the comparisons
DQMHistoTests: Total files compared: 35
DQMHistoTests: Total histograms compared: 2609667
DQMHistoTests: Total failures: 5
DQMHistoTests: Total nulls: 1
DQMHistoTests: Total successes: 2609639
DQMHistoTests: Total skipped: 22
DQMHistoTests: Total Missing objects: 0
DQMHistoSizes: Histogram memory added: -0.004 KiB( 34 files compared)
DQMHistoSizes: changed ( 10224.0 ): -0.004 KiB MessageLogger/Warnings
Checked 149 log files, 22 edm output root files, 35 DQM output files

makortel · 2020-08-27T17:44:11Z

test parameters:

release = slc7_amd64_gcc10

makortel · 2020-08-27T17:44:17Z

@cmsbuild, please test

cmsbuild · 2020-08-31T16:02:14Z

Comparison job queued.

fwyzard · 2020-08-31T17:51:01Z

Ok, apparently in slc7_amd64_gcc10 clang-format prefers > > over >> for some reason.

Nope, it is not about the SCRAM_ARCH. Somehow clang-format gets confused by the #else...

Ah... OK, less worrying then.

fwyzard · 2020-08-31T17:52:06Z

On the other hand, is this really the approach we want to pursue for all CUDA plugins ?

I'd really like to find a different solution (despite of suggesting #ifdef in #28576 (comment)) for longer term. For now it might be the best we can do. I propose we discuss on this tomorrow in the core software meeting.

I remember... and unfortunately I don't have any other ideas at the moment.

cmsbuild · 2020-08-31T19:41:12Z

Comparison is ready
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-6c398b/9019/summary.html

Comparison Summary:

No significant changes to the logs found
Reco comparison results: 42111 differences found in the comparisons
DQMHistoTests: Total files compared: 35
DQMHistoTests: Total histograms compared: 2609667
DQMHistoTests: Total failures: 347643
DQMHistoTests: Total nulls: 3
DQMHistoTests: Total successes: 2261999
DQMHistoTests: Total skipped: 22
DQMHistoTests: Total Missing objects: 0
DQMHistoSizes: Histogram memory added: -0.805 KiB( 34 files compared)
DQMHistoSizes: changed ( 10224.0 ): -0.127 KiB SiStrip/MechanicalView
DQMHistoSizes: changed ( 250202.181 ): -0.986 KiB SiStrip/MechanicalView
DQMHistoSizes: changed ( 25202.0 ): 0.308 KiB SiStrip/MechanicalView
Checked 149 log files, 22 edm output root files, 35 DQM output files

perrotta · 2020-09-02T14:00:56Z

@cmsbuild, please test
(Last time it was tested with CMSSW_11_2_X_2020-08-28-2300, which did not include yet #31283: let check if the HI tests succeed now, first included in CMSSW_11_2_X_2020-08-31-2300)

cmsbuild · 2020-09-02T14:01:24Z

The tests are being triggered in jenkins.
Test Parameters:

ARCHITECTURE_FILTER = slc7_amd64_gcc10
CMSSW_11_2_X_2020-09-01-2300/slc7_amd64_gcc10: https://cmssdt.cern.ch/jenkins/job/ib-run-pr-tests/9066/console Started: 2020/09/02 16:03

cmsbuild · 2020-09-02T15:15:16Z

+1
Tested at: cd6d44f
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-6c398b/9066/summary.html
CMSSW: CMSSW_11_2_X_2020-09-01-2300
SCRAM_ARCH: slc7_amd64_gcc10

cmsbuild · 2020-09-02T15:15:18Z

Comparison job queued.

makortel · 2020-09-02T15:49:30Z

#31334 proposes an alternative by relaxing the pieces in CUDA utilities that can be built with just the host compiler (gcc) to be built with <iftool name="cuda"> instead of <iftool name="cuda-gcc-support">. We can make BeamSpotToCUDA to build and construct without any #ifdefs because no device code is called. Producers calling device still need some kind of protection (details to be figured out).

cmsbuild · 2020-09-02T18:56:13Z

Comparison is ready
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-6c398b/9066/summary.html

Comparison Summary:

No significant changes to the logs found
Reco comparison results: 42068 differences found in the comparisons
DQMHistoTests: Total files compared: 35
DQMHistoTests: Total histograms compared: 2609667
DQMHistoTests: Total failures: 347645
DQMHistoTests: Total nulls: 4
DQMHistoTests: Total successes: 2261996
DQMHistoTests: Total skipped: 22
DQMHistoTests: Total Missing objects: 0
DQMHistoSizes: Histogram memory added: -0.801 KiB( 34 files compared)
DQMHistoSizes: changed ( 10224.0 ): -0.127 KiB SiStrip/MechanicalView
DQMHistoSizes: changed ( 10224.0 ): 0.004 KiB MessageLogger/Warnings
DQMHistoSizes: changed ( 250202.181 ): -0.986 KiB SiStrip/MechanicalView
DQMHistoSizes: changed ( 25202.0 ): 0.308 KiB SiStrip/MechanicalView
Checked 149 log files, 22 edm output root files, 35 DQM output files

makortel · 2020-09-04T01:38:40Z

Closing in favor of #31334

cmsbuild added this to the CMSSW_11_2_X milestone Aug 27, 2020

cmsbuild added code-checks-pending comparison-pending heterogeneous-pending orp-pending pending-signatures reconstruction-pending tests-pending labels Aug 27, 2020

cmsbuild added code-checks-approved and removed code-checks-pending labels Aug 27, 2020

cmsbuild added tests-started and removed tests-pending labels Aug 27, 2020

cmsbuild added tests-approved and removed tests-started labels Aug 27, 2020

cmsbuild added comparison-available and removed comparison-pending labels Aug 27, 2020

cmsbuild added comparison-pending and removed comparison-available tests-approved labels Aug 27, 2020

cmsbuild added tests-rejected and removed tests-started labels Aug 31, 2020

silviodonato mentioned this pull request Aug 31, 2020

Patatrack integration - GPU beamspot data format and transfer (4/N) #31130

Merged

cmsbuild added comparison-available and removed comparison-pending labels Aug 31, 2020

makortel mentioned this pull request Sep 2, 2020

Fix crashing PixelOnlyGPU workflows (add missing CUDAServices) #31333

Merged

cmsbuild added comparison-pending tests-pending and removed comparison-available tests-rejected labels Sep 2, 2020

cmsbuild added tests-started and removed tests-pending labels Sep 2, 2020

cmsbuild added tests-approved and removed tests-started labels Sep 2, 2020

cmsbuild mentioned this pull request Sep 2, 2020

Build CUDA host code if CUDA libraries are available #31334

Merged

cmsbuild added comparison-available and removed comparison-pending labels Sep 2, 2020

makortel closed this Sep 4, 2020

makortel mentioned this pull request May 25, 2021

How to deal with plugins not being built for some architectures? #28576

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[RFC] Make BeamSpotToCUDA constructible, and avoid building CUDADataFormats/BeamSpot when CUDA is not available according to cuda-gcc-support #31261

[RFC] Make BeamSpotToCUDA constructible, and avoid building CUDADataFormats/BeamSpot when CUDA is not available according to cuda-gcc-support #31261

makortel commented Aug 27, 2020 •

edited

Loading

cmsbuild commented Aug 27, 2020

cmsbuild commented Aug 27, 2020

cmsbuild commented Aug 27, 2020

makortel commented Aug 27, 2020

makortel commented Aug 27, 2020

cmsbuild commented Aug 27, 2020 •

edited

Loading

cmsbuild commented Aug 27, 2020

cmsbuild commented Aug 27, 2020

cmsbuild commented Aug 27, 2020

makortel commented Aug 27, 2020

makortel commented Aug 27, 2020

cmsbuild commented Aug 31, 2020

fwyzard commented Aug 31, 2020

fwyzard commented Aug 31, 2020

cmsbuild commented Aug 31, 2020

perrotta commented Sep 2, 2020

cmsbuild commented Sep 2, 2020 •

edited

Loading

cmsbuild commented Sep 2, 2020

cmsbuild commented Sep 2, 2020

makortel commented Sep 2, 2020

cmsbuild commented Sep 2, 2020

makortel commented Sep 4, 2020

[RFC] Make BeamSpotToCUDA constructible, and avoid building CUDADataFormats/BeamSpot when CUDA is not available according to cuda-gcc-support #31261

[RFC] Make BeamSpotToCUDA constructible, and avoid building CUDADataFormats/BeamSpot when CUDA is not available according to cuda-gcc-support #31261

Conversation

makortel commented Aug 27, 2020 • edited Loading

PR description:

PR validation:

cmsbuild commented Aug 27, 2020

cmsbuild commented Aug 27, 2020

cmsbuild commented Aug 27, 2020

makortel commented Aug 27, 2020

makortel commented Aug 27, 2020

cmsbuild commented Aug 27, 2020 • edited Loading

cmsbuild commented Aug 27, 2020

cmsbuild commented Aug 27, 2020

cmsbuild commented Aug 27, 2020

makortel commented Aug 27, 2020

makortel commented Aug 27, 2020

cmsbuild commented Aug 31, 2020

fwyzard commented Aug 31, 2020

fwyzard commented Aug 31, 2020

cmsbuild commented Aug 31, 2020

perrotta commented Sep 2, 2020

cmsbuild commented Sep 2, 2020 • edited Loading

cmsbuild commented Sep 2, 2020

cmsbuild commented Sep 2, 2020

makortel commented Sep 2, 2020

cmsbuild commented Sep 2, 2020

makortel commented Sep 4, 2020

makortel commented Aug 27, 2020 •

edited

Loading

cmsbuild commented Aug 27, 2020 •

edited

Loading

cmsbuild commented Sep 2, 2020 •

edited

Loading