-
Notifications
You must be signed in to change notification settings - Fork 4.4k
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
Make DetSetNew to support concurrent filler #12411
Conversation
The thread_local in RandomNumberGeneratorService was being used to support a backwards compatible interface. The old interface is also supported for certain transitions via another thread_local from the framework. It appears that the old interface is no longer being called by non-framework supported transitions.
We no longer support calling RandomNumberGenerate::mySeed at specific framework transitions. Test now enforces the change.
@cmsbuild, please test |
The tests are being triggered in jenkins. |
+1 The following merge commits were also included on top of IB + this PR after doing git cms-merge-topic: |
#endif | ||
|
||
#include <atomic> | ||
#include <thread> |
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.
What symbol do you need from this include?
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.
none left, indeed.
I think we should have a later pull request which gets rid of the |
+1 |
+1 The following merge commits were also included on top of IB + this PR after doing git cms-merge-topic: |
static DetSetVector<T>::Item & dummy() { | ||
assert(false); | ||
static DetSetVector<T>::Item d; return d; | ||
} | ||
FastFiller(DetSetVector<T> & iv, id_type id, bool isaveEmpty=false) : |
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.
Were all uses of FastFiller
replaced with TSFastFiller
? If so, then in a subsequent pull request we should remove FastFiller
.
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.
No, this is not foreseen at least at the moment (at least on my personal side).
RecoJets /JetAnalyzers /src /myFilter.cc and RecoJets /JetAnalyzers /src /myJetAna.cc generate a lot of static analyzer warnings. Could you please fix them? https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-12411/9851/llvm-analysis/ |
@cvuosalo , I think we should survive with those (or ask jet people to take care... |
@cvuosalo , those warnings are not related to this PR... |
+1 Changing DetSetNew to support concurrent fillers.. There should be no change in monitored reco quantities. The code changes are satisfactory. There are static analyzer warnings about RecoJets/JetAnalyzers/src/myFilter.cc and RecoJets /JetAnalyzers /src /myJetAna.cc that need to be fixed in a later PR. Jenkins tests against baseline CMSSW_8_0_X_2015-11-18-2200 show no significant differences, as expected. |
@VinInn: We are just noticing that the history of this PR includes a lot of irrelevant merges. Could you please clean it up? |
@cvusalo, please give verbatim recipe on how to cleanup. |
Make DetSetNew to support concurrent filler
With this PR DetSetNew support concurrent fillers.
The main application is Strip Unpacking and Clusterization on Demand at HLT.
Passed all tests and show no regression in sequential mode.
How to test the concurrency (besides the unit test)?
Waiting for the in-the-event module concurrency you can try concurrent PatternReco
take 134.91
add
to step2_L1REPACK_HLT.py
and
you can also switch VI_DEBUG in ClustersFromRawProducer.cc and compare
the dump (better just the event by event VI VI clusters lines)
At the moment it will show some differences due to the fact that concurrent PatternReco is not reproducible due to cleaning of seeds and candidate (or a bug???)
I will see if I can had a flag to make it fully reproducible at the cost of slowness and memory-blowup...
btw the current version (contrary to 7_6) will stop with