Convert all remaining MiniAOD modules to threaded FW (76X) #10620
Conversation
|
A new Pull Request was created by @lgray (Lindsey Gray) for CMSSW_7_6_X. Convert all remaining MiniAOD modules to threaded FW (76X) It involves the following packages: CommonTools/ParticleFlow @cmsbuild, @cvuosalo, @vadler, @monttj, @slava77 can you please review it and eventually sign? Thanks. |
|
Ah, it seems I need to play a little whack-a-mole. |
|
@cmsbuild please test |
|
The tests are being triggered in jenkins. |
|
-1 ^
1 warning generated.
>> Compiling edm plugin /tmp/cmsbuild/workspace/ib-any-integration/CMSSW_7_6_X_2015-08-06-2300/src/PhysicsTools/PatAlgos/plugins/PATHemisphereProducer.cc
>> Compiling edm plugin /tmp/cmsbuild/workspace/ib-any-integration/CMSSW_7_6_X_2015-08-06-2300/src/PhysicsTools/PatAlgos/plugins/PATJetProducer.cc
In file included from /tmp/cmsbuild/workspace/ib-any-integration/CMSSW_7_6_X_2015-08-06-2300/src/PhysicsTools/PatAlgos/plugins/PATElectronProducer.cc:2:
/tmp/cmsbuild/workspace/ib-any-integration/CMSSW_7_6_X_2015-08-06-2300/src/PhysicsTools/PatAlgos/plugins/PATElectronProducer.h:231:86: error: no matching member function for call to 'consumes'
tokens = edm::vector_transform(labels, [this](IsolationLabel const & label){return consumesedm::ValueMap(label.second);});
^~~~~~~~~~~~~~~~~~~~~~~~~~~
/tmp/cmsbuild/workspace/ib-any-integration/CMSSW_7_6_X_2015-08-06-2300/src/PhysicsTools/PatAlgos/plugins/PATElectronProducer.cc:159:3: note: in instantiation of function template specialization 'pat::PATElectronProducer::readIsolationLabelsreco::IsoDeposit' requested here
readIsolationLabels(iConfig, "isoDeposits", isoDepositLabels_, isoDepositTokens_);
^
you can see the results of the tests here: |
lgray
force-pushed
the
threaded_miniaod_76X
branch
from
1bf3925 to
a1961e4
Compare
|
@cmsbuild please test Fixed const correctness for clang. |
|
Coool. Thanks @lgray |
|
The tests are being triggered in jenkins. |
cmsbuild
added
comparison-pending
tests-pending
and removed
comparison-available
tests-rejected
labels
|
-1 Tested at: 27eea18 ---> test testJetMETCorrectionsType1MET had ERRORS you can see the results of the tests here: |
|
Unit test failures were not because of this change (DAS failure + data format problem in PhysicsTools/PatAlgo unit test). |
|
@monttj could you please sign here? Thanks! |
|
@slava77 FYI, after the Puppi and MET PRs for MiniAOD 74X are in (... if they go in), then the 74X version of this PR just becomes a straight backport. |
|
@lgray ok, once those get in, I can backport this one |
|
.. actually, I'd rather have the full thing from #10806 |
|
As you wish. On Mon, Aug 24, 2015 at 10:47 PM, Slava Krutelyov [email protected]
|
|
@cmsbuild please test |
|
The tests are being triggered in jenkins. |
cmsbuild
added
tests-pending
tests-started
and removed
tests-rejected
tests-pending
labels
|
+1 |
This PR converts all modules used in MiniAOD to the threaded framework.
Short matrix passed for both single and multithreaded jobs. Thread efficiency looked good, MiniAOD job was pegged at ~200% for a 2 core test.
Once we are done examining the various changes I will squash the commits down to something less sprawling.
I have now tested this incrementing the number of threads from 2 up to 8. Thread efficiency is good up until 6 threads and then it stabilizes, and the initialization time explodes rather quickly beyond 6 threads.
Converting a new legacy module that appeared in 76X did increase the thread efficiency.
Backports to 75X and 74X after we have some more thorough testing (validation and such) and the dust settles from the MiniAOD reprocessing campaign.
@Dr15Jones