backport of #13831: Speedup Gsf component merging, return components by const reference

[fwyzard]

includes #15259 (backport of #13594)

[fwyzard]

please test

[fwyzard]

does not seem to give any improvements to the HLT cpu usage:

Performance for reference
average from 42 tests:
        throughput:   60.9 +/-  0.6 ev/s
        event time:  348.2 +/-  2.8 ms

Performance for backport_13594+13831
average from 42 tests:
        throughput:   60.7 +/-  0.6 ev/s
        event time:  348.0 +/-  2.8 ms

[fwyzard]

tracked at #15151

[cmsbuild]

The tests are being triggered in jenkins.
https://cmssdt.cern.ch/jenkins/job/ib-any-integration/14224/console

[cmsbuild]

A new Pull Request was created by @fwyzard (Andrea Bocci) for CMSSW_8_0_X.

It involves the following packages:

CommonTools/Utils
DataFormats/Math
RecoParticleFlow/PFTracking
RecoTracker/TrackProducer
RecoVertex/GaussianSumVertexFit
TrackingTools/GeomPropagators
TrackingTools/GsfTools
TrackingTools/GsfTracking
TrackingTools/KalmanUpdators
TrackingTools/TrajectoryState

@cmsbuild, @cvuosalo, @slava77, @davidlange6 can you please review it and eventually sign? Thanks.
@ghellwig, @mmarionncern, @battibass, @makortel, @abbiendi, @GiacomoSguazzoni, @rafaellopesdesa, @jhgoh, @lgray, @bellan, @HuguesBrun, @mschrode, @rovere, @cbernet, @gpetruc, @VinInn, @trocino, @dgulhan, @bachtis this is something you requested to watch as well.
@slava77, @smuzaffar you are the release manager for this.

cms-bot commands are list here #13028

[cmsbuild]

+1
Tested at: ba71754
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-15279/14224/summary.html

[cmsbuild]

Comparison is ready
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-15279/14224/summary.html

[makortel]

Just a note that 47cf66d is already part of #15240 (as commit 7cf0c3b). Probably doesn't matter in practice, but I thought to write it down anyway.

[slava77]

merge conflicts

[slava77]

maybe remove this, hoping to remove changes in RECO products and simplify integration?
... I didn't look if anything else could change the RECO products in this PR

[VinInn]

this is a major component of the speedup...

[fwyzard]

I actually tried removing -Ofast, but it did not affect the trigger results, i.e. the same discrepancy is there with and without it. .A

[slava77]

On 7/26/16 10:42 AM, Vincenzo Innocente wrote:

In TrackingTools/GsfTools/plugins/BuildFile.xml
#15279 (comment):

@@ -1,3 +1,4 @@
+

this is a major component of the speedup...

There is essentially no speedup based on tests in HLT with PU40 done by
Andrea.

(also, recall another thread we had on -Ofast in April: there is no
significant speedup from this flag
in CMSSW RECO application
at least on the ~ 3yo Intel and AMD hardware I have tested)

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
https://github.com/cms-sw/cmssw/pull/15279/files/ba717540b0163453d6beeec0f7d2cf0fd87765b8#r72300077,
or mute the thread
https://github.com/notifications/unsubscribe-auth/AEdcbmcKCPhLRfTExfF15PI8TYBW-m_Zks5qZkcYgaJpZM4JU68_.

[VinInn]

This is in contradiction with my test of the original version of this PR for 810

[fwyzard]

rebased on top of CMSSW_8_0_16

[fwyzard]

please test

[cmsbuild]

The tests are being triggered in jenkins.
https://cmssdt.cern.ch/jenkins/job/ib-any-integration/14300/console

[cmsbuild]

Pull request #15279 was updated. @cmsbuild, @cvuosalo, @slava77, @davidlange6 can you please check and sign again.

[cmsbuild]

+1
Tested at: 24dfc4e
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-15279/14300/summary.html

[cmsbuild]

Comparison is ready
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-15279/14300/summary.html

[slava77]

-1
as discussed in #15151 (comment)