CA Extension to strips #47090
CA Extension to strips #47090
Conversation
|
bdanzi, CMSSW_15_0_X branch is closed for direct updates. cms-bot is going to move this PR to master branch. |
|
cms-bot internal usage |
|
-code-checks Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-47090/43269
Code check has found code style and quality issues which could be resolved by applying following patch(s)
|
|
+code-checks Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-47090/43271
|
|
Pull request #47090 was updated. |
|
please test |
|
-code-checks Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-47090/43280
Code check has found code style and quality issues which could be resolved by applying following patch(s)
|
|
+code-checks Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-47090/43281
|
|
Pull request #47090 was updated. |
|
please test |
|
@smuzaffar, looks like #47090 (comment) didn't have the desired effect. Is it a syntax issue, or does the PR need to be "open for review" to allow it? |
|
allow @bdanzi test rights |
thanks @smuzaffar ! |
|
please test |
|
-1 Failed Tests: UnitTests Unit TestsI found 1 errors in the following unit tests: ---> test test-das-selected-lumis had ERRORS Comparison SummarySummary:
|
|
enable gpu |
|
|
||
| # CMSSW version specific customizations | ||
| def customizeHLTforCMSSW(process, menuType="GRun"): | ||
|
|
||
| process = customiseForOffline(process) | ||
|
|
||
| process = configureFrameSoAESProducers(process) |
There was a problem hiding this comment.
What is the purpose of this customisation which is always executed? Is it purely technical (no changes in physics expected), or making CA extension changes?
And what is the relation to the above customisation file customizeHLTforAlpakaStripNoDoubletRecovery.py which isn't called explicitly anywhere?
There was a problem hiding this comment.
What is the purpose of this customisation which is always executed? Is it purely technical (no changes in physics expected),
without it, the HLT processing fails:
----- Begin Fatal Exception 15-Jan-2025 10:53:31 CET-----------------------
An exception of category 'NoRecord' occurred while
[0] Processing Event run: 386593 lumi: 235 event: 535875922 stream: 0
[1] Running path 'AlCa_PFJet40_CPUOnly_v11'
[2] Calling method for module alpaka_serial_sync::CAHitNtupletAlpakaPhase1/'hltPixelTracksSoASerialSync'
Exception Message:
No "FrameSoARecord" record found in the EventSetup.
Please add an ESSource or ESProducer that delivers such a record.
----- End Fatal Exception -------------------------------------------------
I suppose it's for compatibility reasons (if the customization is not present then the seeding module expects an ES product which is not available).
On the other hand looking at the trigger results there are trigger decision changes, so the PR as it is it's not transparent.
And what is the relation to the above customisation file customizeHLTforAlpakaStripNoDoubletRecovery.py which isn't called explicitly anywhere?
I am assuming customizeHLTforAlpakaStripNoDoubletRecovery is the "meaty" part of this PR (the one actually making the physics changes) -- though I agree it's not called anywhere and that's not good because we need a way to test it.
@bdanzi @cms-sw/tracking-pog-l2 please elaborate / clarify.
There was a problem hiding this comment.
Indeed, enabling customizeHLTforAlpakaStripNoDoubletRecovery would be the ultimate goal following this PR.
As is, this customization is introduced to enable the possibility for other downstream users to test it, with the idea to possibly enable it afterwards. Are you suggesting that this is called in a special workflow?
There was a problem hiding this comment.
Are you suggesting that this is called in a special workflow?
IMHO that would be useful for reviewing purposes.
There was a problem hiding this comment.
For physics changes outside the customizeHLTforAlpakaStripNoDoubletRecovery, I guess those are introduced due to the lowering of cuts on minYsizeB1 and minYsizeB2 in Geometry/CommonTopologies/interface/SimplePixelTopology.h that affects also the CA applied only on pixels since I was observing better performance with minor increase in fake+duplicates, especially in EOR3 scenario + generalization of Kernel_simpleTripletCleaner
|
enable profiling |
PR description:
This PR should introduce the usage of strips layers to the offline reco and HLT Cellular Automaton (CA) usage (work done in 15_0_0_pre1):
customizeHLTforAlpakaStripfor HLT andstripNTupletFitfor offline reco, final geometry (Geometry/CommonTopologies/interface/SimplePixelStripTopology.py) pairs is still work in progressminYsizeB1andminYsizeB2cuts + generalization ofKernel_simpleTripletCleanerTest on TTbar PU EOR3 TRK DPG v7:
http://uaf-3.t2.ucsd.edu/~bdanzi/plots_pixelsOnlyValidation_AllCases/plots_hlt_hltPixel/effandfakePtEtaPhi.png where results in blue and red are produced with a cleaned 14_2_0 CMSSW release while the black one comes along with this PR changes
PR validation:
Before submitting I was checking
cc @mmasciov @slava77
Work co-authored-by: @AdrianoDee Adriano Di Florio [email protected]