[Insight-users] benchmark-computation time

[John Drescher]

On Tue, Jul 24, 2012 at 10:37 AM, John Drescher <drescherjm at gmail.com> wrote:
> On Tue, Jul 24, 2012 at 10:35 AM, amir_k <kacakici at hotmail.com> wrote:
>> I have just wrote the same question in the VTK forum, but the subject
>> deserves an adequate place in this forum as well.
>>
>> I've been running the watershed filter since few days now. At first, I
>> thought my filter was not working (i.e pipeline lost connection -
>> uninitialized data passing at the filter) but it does finally. It just takes
>> soo long time to execute the program. Here are my two cases:
>>
>> 1)  DICOM images ( 512x512x10) with not so-detailed data
>> - it took >30 mins to execute the watershed filter ->Update(). my two input
>> parameters for lowerThreshold and outputScalelevel are 0.002 and 0.2
>> respectively.
>>
>> 2) I tried this time to run the exemple WatershedSegmentation2.exe
>> - this program needs a 3D input file , so I use as proposed in the forum the
>> brain data file (brainweb1e5a10f17.mha). this file dimension is 181x217x181.
>> - the output is output.vtk and the other two input parameters are 0.01 and
>> 0.15.
>> - it s been running for >30 mins now while here I am writing in this forum
>> (hahaha -I still can laugh but my supervisor seems pissed off with my work
>> already)
>>
>
> Run your program in Release mode. My same comment about 100 times as
> long in debug mode applies to itk as well as vtk.
>

Also remember when you do that 100% of your program must be in Release
mode. You can not safely mix and match Release and Debug in the same
application with Visual Studio.

John