[vtk-developers] Rendering hangs in viewport

[Taras Shchehelskyi]

Hello Will,

Yes, you are right. Problem starts occurring only after long run time.
Thank you for your suggestion about memory leak. Will check this.

Thanks,
Taras
5 лип. 2016 14:17 "Will Schroeder" <will.schroeder at kitware.com> пише:

> Taras-
>
> My initial reaction is that there is a memory leak. If I understand you
> correctly, you are running 12-15 hours and then problems start occurring.
> Even a small leak can add up over that length of time and corrupt memory. I
> would first instrument your classes and see if they are leaking; if not it
> could be a VTK problem although we try and test for this kind of behavior.
>
> Best,
> W
>
> On Tue, Jul 5, 2016 at 7:00 AM, Taras Shchehelskyi <shchegelskij at gmail.com
> > wrote:
>
>>
>>
>>
>>
>> Hello,
>>
>>
>>
>> I am part of the team that is working on project that uses VTK 7.0 for
>> rendering mesh grid (not big, about 400k triangles) couple point clouds
>> (about 8000 – 30000 points per cloud, up to 4 clouds) + couple very simple
>> meshes. Our app can work fine for 12-15h for then main viewport hangs. When
>> I do some mouse interaction (click + move or scroll) some meshes became
>> invisible for couple seconds and that became visible again. Also when I
>> resize viewport window scene aspect ratio became wrong. It works like
>> aspect ratio does not depends of viewport size.
>>
>>
>>
>> Also we noticed other strange thing. Our point clouds update dynamically
>> during app work. We create 2 classes:
>>
>> class CameraPointCloudSource : public vtkPolyDataAlgorithm
>>
>> and
>>
>> class PointCloudColoring : public vtkPolyDataAlgorithm.
>>
>>
>>
>> I pass data (vtkPolyData with initialized points and vertexes) from
>> CameraPointCloudSource to PointCloudColoring. I setup pipeline that output
>> of the CameraPointCloudSource is input of the PointCloudColoring.
>>
>>
>>
>> Sometimes (can’t always reproduce this) *CameraPointCloudSource::
>> RequestData* stops to be called and *CameraPointCloudSource::RequestData*
>> always get the same input data (the same point cloud the same with the same
>> points). As result I always see in viewport the same point cloud
>> visualization. I checked, code that responsible for point cloud generation
>> works fine and produce correct point clouds.
>>
>>
>>
>> For test we also tried to use new VTK PointCloud filters. There we also
>> noticed similar behavior. We set filters chain.  Alter some long time (>
>> 12h of work) sometimes some filters star producing the same output that not
>> depends of input point cloud. Interesting that some filters work fine and
>> some start producing the same output. This issue can be in combination with
>> viewport rendering hangs. Sometimes can happen before hang. Sometimes can
>> happen after rendering hang.
>>
>>
>>
>> Our app uses many threads for generating data. We have many mutexes to
>> protect data. But we are pretty sure that there are no deadlocks. Also we
>> checked couple times that we not share Actors and Mappers between threads.
>> We have couple viewports. But I was able to reproduce bug even when only
>> one viewport is active. Interesting that when one viewport can hang, other
>> viewports that uses the same data can work fine.
>>
>>
>>
>> We are using Windows 7/10 x64 and MS VS 2013.
>>
>>
>>
>>
>>
>> Any ideas what can lead to this problem?
>>
>>
>>
>> P.S. Sorry for my English, I am not a native speaker.
>>
>>
>>
>>
>>
>> Thanks,
>>
>> Taras
>>
>> _______________________________________________
>> Powered by www.kitware.com
>>
>> Visit other Kitware open-source projects at
>> http://www.kitware.com/opensource/opensource.html
>>
>> Search the list archives at: http://markmail.org/search/?q=vtk-developers
>>
>> Follow this link to subscribe/unsubscribe:
>> http://public.kitware.com/mailman/listinfo/vtk-developers
>>
>>
>>
>
>
> --
> William J. Schroeder, PhD
> Kitware, Inc. - Building the World's Technical Computing Software
> 28 Corporate Drive
> Clifton Park, NY 12065
> will.schroeder at kitware.com
> http://www.kitware.com
> (518) 881-4902
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/vtk-developers/attachments/20160705/af841663/attachment.html>