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
recover multi-thread reproducibility in CTPPS diamond local track y position #18779
recover multi-thread reproducibility in CTPPS diamond local track y position #18779
Conversation
…sitionInitial and yWidthInitial instead of hardcoded 0,0
A new Pull Request was created by @slava77 (Slava Krutelyov) for master. It involves the following packages: RecoCTPPS/TotemRPLocal @perrotta, @cmsbuild, @slava77, @davidlange6 can you please review it and eventually sign? Thanks. cms-bot commands are listed here |
@cmsbuild please test |
The tests are being triggered in jenkins. |
Hi @slava77, Anyway, thanks again @slava77! |
Comparison job queued. |
Comparison is ready Comparison Summary:
|
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 requires discussion in the ORP meeting before it's merged. @davidlange6, @smuzaffar |
+1 |
clear yPosition and width after the fit; test for first hit using yPositionInitial and yWidthInitial instead of hardcoded 0,0
The issue was discovered in 80X.
I'm actually not fully capturing the logic of
yPosition_
assignment based on the first hit y position. It looks partially as a way to cache the detector alignment.In my test job in 80X (running on run 283877) I used 8 threads. The original CTPPSDiamondTrackRecognition.cc can possibly make 2 (arms) * 8 (threads) different values of track y depending which event and which hit ends up on the thread first. The fixed version corresponds to position setting in each event.
On 200 events: the original getY0 values (with repeat counts)
The fixed:
If these positions are in cm, they look like random roundoffs of zeros.
Still, the assignment should be made without dependence on event order or the number of threads.
@forthommel please check and either confirm that this fix has a meaning or provide a replacement PR.
Thank you.