-
Notifications
You must be signed in to change notification settings - Fork 4.6k
LST: add trackToTrack comparisons for GPU vs CPU tracking #48508
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
LST: add trackToTrack comparisons for GPU vs CPU tracking #48508
Conversation
|
cms-bot internal usage |
|
test parameters:
|
|
+code-checks Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-48508/45442
|
|
A new Pull Request was created by @slava77 for master. It involves the following packages:
@AdrianoDee, @Moanwar, @antoniovagnerini, @cmsbuild, @ctarricone, @DickyChant, @jfernan2, @mandrenguyen, @miquork, @rseidita, @srimanob, @subirsarkar can you please review it and eventually sign? Thanks. cms-bot commands are listed here |
|
@cmsbuild please test |
|
-1 Failed Tests: RelVals rocmUnitTests RelValsROCm Unit TestsI found 0 errors in the following unit tests: CUDA Comparison SummarySummary:
ROCM Comparison SummarySummary:
|
failures are unrelated see #47859 (comment) |
|
@cmsbuild please test |
|
-1 Failed Tests: rocmUnitTests ROCm Unit TestsI found 0 errors in the following unit tests: Comparison SummarySummary:
CUDA Comparison SummarySummary:
ROCM Comparison SummarySummary:
|
|
+1 |
|
+Upgrade |
@cms-sw/dqm-l2 @cms-sw/pdmv-l2 signatures are missing I'm not sure still if @cms-sw/heterogeneous-l2 is going to assign this PR |
|
From a cursory look I don't think we need to assign to and sign for @cms-sw/heterogeneous-l2 . |
@cms-sw/orp-l2 this was open for 4 weeks |
|
+pdmv |
|
@cms-sw/dqm-l2 (now also directly @antoniovagnerini @rseidita @ctarricone) |
|
+dqm |
|
This pull request is fully signed and it will be integrated in one of the next master IBs (tests are also fine). This pull request will now be reviewed by the release team before it's merged. @antoniovilela, @mandrenguyen, @rappoccio, @sextonkennedy (and backports should be raised in the release meeting by the corresponding L2) |
|
+1 |
the monitoring sequence is added in combination with
trackingLSTproc modifier for the offline setup.In presence of the DQM (
TrackingOfflineDQMClient) sequence this would trigger running a clone of the regular LST tracking inHighPtTripletStepTaskSerialSync(so that a no-DQM variant, e.g. used for timing, would not run the extra variant of this iteration).DQM plots should show up in the
/DQMData/Run 1/Tracking/Run summary/TrackBuilding/ValidationWRTSerialSync/highPtTripletStepfolderI made an attempt for a somewhat generic setup by introducing a
trackToTrackCPUSequence, which can eventually be populated with other comparisons.The implementation and population of the
HighPtTripletStepTaskSerialSynctask is a bit tedious,Tested with 29834.704
The efficiency and deltaeta plots are shown below.
this PR is supposed to address the
Implement a "GPU vs CPU" workflowpart of #46746 (a followup to the LST integration PR #45117 (comment)) and was somewhat high on the wish list related to the presentation in the GPU meeting May 19A specific implementation was not particularly well specified; so, this may become an RFC.
@VourMa @fwyzard