[Insight-developers] Dashboard: Looking at Valgrind

Fri Dec 17 10:27:21 EST 2010

brian.jouy.inra.fr is still running the full set of test with valgrind  
every nights.
At some points, I thought this would be problematic, but to date, this  
is still not a problem for us to continue to do that.

Gaëtan

Le 17 déc. 10 à 15:27, Bradley Lowekamp a écrit :

> Ugh,
>
> So that means that we are running on the same 10% of the toolkit  
> each night. I suppose the idea here was to modulate the START  
> parameter on the day of the month or something to rotate through the  
> tests.
>
>
> Luis,
>
> Your fix for the membership functions removed like 50 errors on each  
> of the valgrinds last night. That is some nice impact!
>
> Brad
>
>
>
> On Dec 16, 2010, at 5:10 PM, Nicolas Rannou wrote:
>
>> Hi Brad, Luis and Gaetan,
>>
>> So far on clownfish, we are running the memory analysis with the  
>> following args:
>> set(CTEST_TEST_ARGS START 0 STRIDE 10 PARALLEL_LEVEL 8)
>>
>> It is totally arbitrary.
>> We should define a testing strategy between the different test  
>> machines, since we can't cover all the tests on the same computer.
>> I guess the original plan was to divide the tests between  
>> brian.jouy and clownfish, based on the parity of the test number.
>>
>>     Nicolas
>>
>>
>> On 12/16/2010 03:44 PM, Bradley Lowekamp wrote:
>>>
>>>
>>> On Dec 16, 2010, at 3:10 PM, Luis Ibanez wrote:
>>>
>>>> Hi Brad,
>>>>
>>>> Thanks for pointing this out.
>>>>
>>>> Here is the summary of valgrind builds from clownfish
>>>> since December first.:
>>>>
>>>> http://www.cdash.org/CDash/index.php?project=Insight&date=2010-12-14&filtercount=3&showfilters=1&filtercombine=and&field1=site/string&compare1=63&value1=clownfish&field2=hasdynamicanalysis/bool&compare2=0&value2=&field3=buildstarttime/date&compare3=83&value3=december%201%202010
>>>>
>>>> I believe that this machine is doing different sections
>>>> of the toolkit on different days, so comparing one day
>>>> to the next may not be representing a real increase or
>>>> decrease in the number of issues.
>>>
>>> Well looking at the valgrind errors from night to night, they  
>>> appear to be the same tests each day... so either they are all  
>>> getting executed each night or the same ones run each night.
>>>
>>> As clownfish and brian.jouy report mostly all of the same test  
>>> failures with valgrind each night, I'd hope that they are running  
>>> complete tests, or else we are missing a significant about of  
>>> valgrind checking!
>>>
>>> brian.jouy appears to be running most of the tests each night:
>>> http://www.cdash.org/CDash/index.php?project=Insight&date=2010-12-14&filtercount=3&showfilters=1&filtercombine=and&field1=site/string&compare1=63&value1=clownfish&field2=hasdynamicanalysis/bool&compare2=0&value2=&field3=buildstarttime/date&compare3=83&value3=december%201%202010
>>>
>>> I thought that when parallel ctest with the "-j" flag was added  
>>> then valgrind builds were able to complete the testing of the  
>>> entire toolkit  in a reasonable time.
>>>
>>> I am not sure what clownfish is doing with such few tests being  
>>> run each night, on the main part of the dashboard.
>>>
>>> Brad
>>>
>>>>
>>>> That said,
>>>> you have a point in that the number of
>>>> issues should be zero anyways               :-)
>>>>
>>>> ---
>>>>
>>>> The problem with the itkGaussianMembershipFunction.txx
>>>> is the lack of initialization for m_Mean in the constructor.
>>>>
>>>> I just pushed a patch to Gerrit for this issue:
>>>> http://review.source.kitware.com/#change,576
>>>>
>>>> This takes care of the following:
>>>>
>>>> itkGaussianMixtureModelComponentTest
>>>> http://www.cdash.org/CDash/viewDynamicAnalysisFile.php?id=996928
>>>>
>>>> itkGaussianMembershipFunctionTest
>>>> http://www.cdash.org/CDash/viewDynamicAnalysisFile.php?id=996927
>>>>
>>>> itkSampleClassifierFilterTest7
>>>> http://www.cdash.org/CDash/viewDynamicAnalysisFile.php?id=996926
>>>>
>>>>
>>>> Thanks
>>>>
>>>>
>>>>    Luis
>>>>
>>>>
>>>>
>>>>
>>>> ------------------------------------------------
>>>> On Thu, Dec 16, 2010 at 10:00 AM, Bradley Lowekamp <blowekamp at mail.nih.gov 
>>>> > wrote:
>>>> Hello,
>>>>
>>>> I noticed that there a bit of an increase in the number of  
>>>> Valgrind defects today. I believe this is likely related to the  
>>>> increased number of statistics example that were added.
>>>>
>>>> http://www.cdash.org/CDash/viewDynamicAnalysis.php?buildid=802880
>>>>
>>>> Particularly there is likely a real issue with the  
>>>> GaussianMembershipFunction:
>>>>
>>>> http://www.cdash.org/CDash/viewDynamicAnalysisFile.php?id=996287
>>>>
>>>> The number of Valgrind error has been increasing, for this to be  
>>>> a useful tool, it need to be looked at and address.
>>>>
>>>> Brad
>>>>
>>>> ========================================================
>>>> Bradley Lowekamp
>>>> Lockheed Martin Contractor for
>>>> Office of High Performance Computing and Communications
>>>> National Library of Medicine
>>>> blowekamp at mail.nih.gov
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> Powered by www.kitware.com
>>>>
>>>> Visit other Kitware open-source projects at
>>>> http://www.kitware.com/opensource/opensource.html
>>>>
>>>> Kitware offers ITK Training Courses, for more information visit:
>>>> http://kitware.com/products/protraining.html
>>>>
>>>> Please keep messages on-topic and check the ITK FAQ at:
>>>> http://www.itk.org/Wiki/ITK_FAQ
>>>>
>>>> Follow this link to subscribe/unsubscribe:
>>>> http://www.itk.org/mailman/listinfo/insight-developers
>>>>
>>>>
>>>
>>> ========================================================
>>> Bradley Lowekamp
>>> Lockheed Martin Contractor for
>>> Office of High Performance Computing and Communications
>>> National Library of Medicine
>>> blowekamp at mail.nih.gov
>>>
>>>
>>
>
> ========================================================
> Bradley Lowekamp
> Lockheed Martin Contractor for
> Office of High Performance Computing and Communications
> National Library of Medicine
> blowekamp at mail.nih.gov
>
>

-- 
Gaëtan Lehmann
Biologie du Développement et de la Reproduction
INRA de Jouy-en-Josas (France)
tel: +33 1 34 65 29 66    fax: 01 34 65 29 09
http://voxel.jouy.inra.fr  http://www.itk.org
http://www.mandriva.org  http://www.bepo.fr

-------------- next part --------------
A non-text attachment was scrubbed...
Name: PGP.sig
Type: application/pgp-signature
Size: 203 bytes
Desc: Ceci est une signature ?lectronique PGP
URL: <http://www.itk.org/mailman/private/insight-developers/attachments/20101217/3580e6d8/attachment.pgp>