-
Notifications
You must be signed in to change notification settings - Fork 4.3k
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
fix HGCAL Layer Cluster times in heterogeneous workflows at HLT #45838
base: master
Are you sure you want to change the base?
Conversation
cms-bot internal usage |
+code-checks Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-45838/41582 |
A new Pull Request was created by @AuroraPerego for master. It involves the following packages:
@cmsbuild, @jfernan2, @mandrenguyen, @srimanob, @subirsarkar can you please review it and eventually sign? Thanks. cms-bot commands are listed here |
thanks @AuroraPerego for the investigation and the fix. |
@cmsbuild please test |
+1 Size: This PR adds an extra 20KB to repository Comparison SummarySummary:
|
for (unsigned int i = 0; i < clusters->size(); ++i) { | ||
times[i].reserve(16); | ||
timeErrors[i].reserve(16); | ||
} |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Would it make sense to embed this loop into the previous one at lines 69 to 77? In this way, the heuristic is not even necessary and the size can be reserved correctly (i.e. the maximum possible, eventually some rechits will be discarded from the timing computation).
95e7d74
to
c733b85
Compare
@cmsbuild please test |
The bot seems to be a little stuck... |
+code-checks Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-45838/41597 |
Pull request #45838 was updated. @jfernan2, @mandrenguyen, @srimanob, @subirsarkar can you please check and sign again. |
+1 Size: This PR adds an extra 20KB to repository Comparison SummarySummary:
|
is there a way to make this change appear in the bot tests? E.g. by running wf |
I'm unsure because this workflow runs the producer I've changed, but the collection of clusters times is not saved in the event. |
and does it have any bearing on trigger decisions? |
I doubt :) |
I am confused. So this PR is of no consequence? what's the purpose? |
when running CLUE on GPU at HLT for Phase2, the timing assigned to the clusters is wrong. |
Thinking about it the effects should be visible downstream, e.g. in tracksters and candidates. They are saved in the event and their time is computed using the clusters times. |
test parameters:
|
please test |
-1 Failed Tests: RelVals-GPU RelVals-GPUValueError: Undefined workflows: 31834.492 Comparison SummarySummary:
|
test parameters:
|
please test |
+1 Size: This PR adds an extra 12KB to repository Comparison SummarySummary:
GPU Comparison SummarySummary:
|
tbh, I am more confused than ever. |
The comparisons show tiny difference in pixel tracks offline alpaka GPU vs alpaka GPU if I understand correctly. |
PR description:
The
times
andtimeErrors
vectors used to store the rechits time were initialized with a size of 16, but thenpush_back
was used to fill them leaving the first 16 entries as 0.This PR removes the allocation of 16 elements in the constructor, using
reserve()
instead.The change affects only heterogeneous workflows at the HLT.
PR validation:
Tested on wf 31834.492, the HGCAL Layer Clusters times are expected to change:
old -> pre fix
new -> post fix
FYI @rovere