[Rtk-users] rtkfdkcudatest bug

[Simon Rit]

I have difficulties following you but if this is a bug in your opinion
and you can design a small test case, you can fill a bug report and
we'll work on it...
Simon

On Wed, Dec 5, 2012 at 10:05 AM, Sébastien Brousmiche
<sebastien.brousmiche at uclouvain.be> wrote:
> Ok, thanks. I fixed it.
>
> Using TRY_AND_EXIT_ON_ITK_EXCEPTION around the
> m_feldkamp->GetOutput()->UpdateOutputInformation() line, I got an exception
> from the ExtractImageFilter in FDKConeBeamReconstructionFilter about
> input/output regions with inconsistent sizes.
> It is a rather general and obscure exception for almost anything in ITK ;-)
>
> Actually, the piece of code doing the cuda initialization was in a method
> different from the one updating the pipeline  with a newly received image.
> I guess ITK doesn't like that!
>
>
>
>
>
>
>
> -----Original Message-----
> From: simon.rit at gmail.com [mailto:simon.rit at gmail.com] On Behalf Of Simon
> Rit
> Sent: mercredi 5 décembre 2012 09:35
> To: Sébastien Brousmiche
> Cc: Marc Vila Oliva; rtk-users at openrtk.org
> Subject: Re: rtkfdkcudatest bug
>
> I can already tell you that we will never have it running on any
> workstation! But we are doing our best...
> It would help if you would tell us the output of the test fail and if you
> would try to catch the exceptions in your inline test and tell us what is
> the error you get, for example using the macro
> TRY_AND_EXIT_ON_ITK_EXCEPTION. It is a lot more informative than the line
> where there is a crash...
> Thanks in advance,
> Simon
>
> On Wed, Dec 5, 2012 at 9:00 AM, Sébastien Brousmiche
> <sebastien.brousmiche at uclouvain.be> wrote:
>> Hi,
>>
>> You're right, as it doesn't crash (or segfault) it's not really a
>> bug... but the test doesn't pass so there is a bug.
>> In conclusion: "The bug has its reasons that the reason ignore!"
>>
>> It does depend on the hardware. How could we be sure that it will run
>> correctly on any workstation?
>>
>> Actually I run the tests because I changed a bit the way I use the
>> inline CUDA FDK and it doesn't work anymore.
>> It crashes at the second line:
>>
>> FDKCUDAType* fdkcuda = static_cast<FDKCUDAType*>(
>> m_feldkamp.GetPointer() );
>> m_feldkamp->GetOutput()->UpdateOutputInformation();
>>
>> Any idea?
>>
>> Sébastien
>>
>>
>>
>> -----Original Message-----
>> From: simon.rit at gmail.com [mailto:simon.rit at gmail.com] On Behalf Of
>> Simon Rit
>> Sent: mardi 4 décembre 2012 21:47
>> To: Sébastien Brousmiche
>> Cc: Marc Vila Oliva; rtk-users at openrtk.org
>> Subject: Re: rtkfdkcudatest bug
>>
>> Hi,
>> Is this a bug or does give you an error message? I can't remember why
>> I changed it but I fear this is not a bug, it may rather depend on the
>> hardware and the cuda version.
>> Simon
>>
>> On Tue, Dec 4, 2012 at 6:59 PM, Sébastien Brousmiche
>> <sebastien.brousmiche at uclouvain.be> wrote:
>>> Fix compatibility when CUDA is not presen
>>
>