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
Start conversion to using user-defined literals for Geometry #24509
Start conversion to using user-defined literals for Geometry #24509
Conversation
…or improved performance
The code-checks are being triggered in jenkins. |
+code-checks Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-24509/6371 |
A new Pull Request was created by @cvuosalo (Carl Vuosalo) for master. It involves the following packages: DetectorDescription/Core @civanch, @Dr15Jones, @cvuosalo, @ianna, @mdhildreth, @cmsbuild can you please review it and eventually sign? Thanks. cms-bot commands are listed here |
Note that two small changes in program behavior have been implemented. First, the old code was truncating fractional values in degrees of rotations, so values of n.0 to n.899999 were truncated to n.0 degrees (where n could be positive or negative). On the advice of @ianna, this truncation has been eliminated, but it might cause a very small change in the value of rotations. |
@cmsbuild please test |
The tests are being triggered in jenkins. |
Comparison job queued. |
Comparison is ready Comparison Summary:
|
hold |
Pull request has been put on hold by @ianna |
@ianna: I will restore the truncation of rotation angle to the nearest degree for the GEM code and test to see if the discrepancies disappear. |
Comparison job queued. |
Comparison is ready Comparison Summary:
|
@cvuosalo - geometry validation seems to be fine for the RPC geometry. DTs do show superficial differences in a norm vector, for example:
vs
To reproduce the results, please:
Please, see: |
The DT reconstruction geometry has not been updated since CMSSW 5.0. It differs from the currently built from 10.3.x pre-release: |
unhold |
Testing the DT reco geometry against 103X_postLS2_design_Candidate_2018_09_27_08_17_34 GT candidate queue shows no differences. |
+1 There are no differences in either 2019 or 2023D17. The other 2023D* scenario differences could be due to low statistics. |
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. @davidlange6, @slava77, @smuzaffar, @fabiocos (and backports should be raised in the release meeting by the corresponding L2) |
@ianna 2018 and older setups use the DB snapshot, so you rely on upgrade scenarios for direct DD validation, am I correct? |
@fabiocos - no, the DD validation is done by comparison XML vs DB vs local DB with a script:
There are no differences in the 2019 GEM reco geometry: |
+1 |
It is hoped that replacing constants from CLHEP/Units/GlobalSystemOfUnits.h with user-defined literals from DetectorDescription/Core/interface/DDUnits.h will improve performance by eliminating unnecessary floating point operations and conversions. This PR is a preliminary step in a campaign to convert the Geometry code to employ user-defined literals.
The changes in this PR do not appreciably change performance since the affected code is run only once at the start of simulation, but it does show the sort of changes entailed in the conversion to user-defined literals.