[Paraview] Paraview and tiled display

Thu Sep 3 15:47:39 EDT 2009

Greg,

I've been able to reproduce it. Let me try to track it down.

Utkarsh

On Thu, Sep 3, 2009 at 3:26 PM, Utkarsh
Ayachit<utkarsh.ayachit at kitware.com> wrote:
> Forwarding to the list, to see if anyone else has seen this.
>
> Utkarsh
>
>
> ---------- Forwarded message ----------
> From: Greg Abram <gda at tacc.utexas.edu>
> Date: Thu, Sep 3, 2009 at 3:24 PM
> Subject: RE: [Paraview] Paraview and tiled display
> To: Utkarsh Ayachit <utkarsh.ayachit at kitware.com>
>
>
> Utkarsh,
>
> Thanks for the quick fix - applied to 3.6, and it works.  However, I'm
> using a Sphere source for testing, and when I set the two resolutions
> to 200, I get an immediate crash.  I'm using a 3x3 display, and each
> is 2560x1600.  Didn't happen when I used 1x1, 1x2 and 2x2 tiles - but
> did at 3x3.  Also doesn't happen if I set the resolutions to only 100.
>
> Greg
>
> c6:00808] *** Process received signal ***
> [c6:00808] Signal: Segmentation fault (11)
> [c6:00808] Signal code: Invalid permissions (2)
> [c6:00808] Failing at address: 0x7f7ef3715000
> [c6:00808] [ 0] /lib/libpthread.so.0 [0x7f7eff2b17d0]
> [c6:00808] [ 1]
> /home/gda/local/lib/paraview-3.6/libicet.so(icetDecompressImage+0x245)
> [0x7f7f0d8128c5]
> [c6:00808] [ 2] /home/gda/local/lib/paraview-3.6/libicet_strategies.so
> [0x7f7f0da18c56]
> [c6:00808] [ 3]
> /home/gda/local/lib/paraview-3.6/libicet_strategies.so(icetSendRecvLargeMessages+0x397)
> [0x7f7f0da190e7]
> [c6:00808] [ 4]
> /home/gda/local/lib/paraview-3.6/libicet_strategies.so(icetRenderTransferFullImages+0xee)
> [0x7f7f0da193fe]
> [c6:00808] [ 5] /home/gda/local/lib/paraview-3.6/libicet_strategies.so
> [0x7f7f0da1a90f]
> [c6:00808] [ 6]
> /home/gda/local/lib/paraview-3.6/libicet.so(icetDrawFrame+0xe31)
> [0x7f7f0d810961]
> [c6:00808] [ 7]
> /home/gda/local/lib/paraview-3.6/libvtkPVFilters.so(_ZN15vtkIceTRenderer12DeviceRenderEv+0x20a)
> [0x7f7f0f9361fa]
> [c6:00808] [ 8]
> /home/gda/local/lib/paraview-3.6/libvtkRendering.so.pv3.6(_ZN11vtkRenderer6RenderEv+0x4ff)
> [0x7f7f0730488f]
> [c6:00808] [ 9]
> /home/gda/local/lib/paraview-3.6/libvtkRendering.so.pv3.6(_ZN21vtkRendererCollection6RenderEv+0x81)
> [0x7f7f07300f11]
> [c6:00808] [10]
> /home/gda/local/lib/paraview-3.6/libvtkRendering.so.pv3.6(_ZN15vtkRenderWindow14DoStereoRenderEv+0xa5)
> [0x7f7f07312295]
> [c6:00808] [11]
> /home/gda/local/lib/paraview-3.6/libvtkRendering.so.pv3.6(_ZN15vtkRenderWindow10DoFDRenderEv+0x34e)
> [0x7f7f073120fe]
> [c6:00808] [12]
> /home/gda/local/lib/paraview-3.6/libvtkRendering.so.pv3.6(_ZN15vtkRenderWindow10DoAARenderEv+0x464)
> [0x7f7f073127b4]
> [c6:00808] [13]
> /home/gda/local/lib/paraview-3.6/libvtkRendering.so.pv3.6(_ZN15vtkRenderWindow6RenderEv+0x10c)
> [0x7f7f073114dc]
> [c6:00808] [14]
> /home/gda/local/lib/paraview-3.6/libvtkRendering.so.pv3.6(_ZN22vtkXOpenGLRenderWindow6RenderEv+0x40)
> [0x7f7f073ea340]
> [c6:00808] [15]
> /home/gda/local/lib/paraview-3.6/libvtkParallel.so.pv3.6(_ZN25vtkMultiProcessController10ProcessRMIEiPvii+0x180)
> [0x7f7f08a047a0]
> [c6:00808] [16]
> /home/gda/local/lib/paraview-3.6/libvtkParallel.so.pv3.6(_ZN25vtkMultiProcessController11ProcessRMIsEii+0x9e)
> [0x7f7f08a0487e]
> [c6:00808] [17]
> /home/gda/local/lib/paraview-3.6/libvtkPVServerCommon.so(_ZN16vtkProcessModule5StartEiPPc+0x18a)
> [0x7f7f0fc2cbea]
> [c6:00808] [18]
> /home/gda/local/lib/paraview-3.6/libvtkPVFilters.so(_ZN9vtkPVMain3RunEP12vtkPVOptions+0x59)
> [0x7f7f0f8a96f9]
> [c6:00808] [19]
> /home/gda/local/lib/paraview-3.6/pvserver-real(main+0x8b) [0x40150b]
> [c6:00808] [20] /lib/libc.so.6(__libc_start_main+0xf4) [0x7f7efe8221c4]
> [c6:00808] [21]
> /home/gda/local/lib/paraview-3.6/pvserver-real(__gxx_personality_v0+0xb9)
> [0x401389]
> [c6:00808] *** End of error message ***
> ICET,3:ERROR: Sizes of src and dest do not agree.
> mpirun noticed that job rank 0 with PID 665 on node c7 exited on
> signal 15 (Terminated).
> 8 additional processes aborted (not shown)
>
> Gregory D. Abram
> Texas Advanced Computing Center
> JJ Pickle Research Campus - ROC 1.101 (R8700)
> 10100 Burnet Road
> Austin, Texas 78758-4497
>
> gda at tacc.utexas.edu
> (512) 471-8196
>
>
> ________________________________________
> From: Utkarsh Ayachit [utkarsh.ayachit at kitware.com]
> Sent: Wednesday, September 02, 2009 7:32 PM
> To: Greg Abram
> Cc: Berk Geveci; paraview at paraview.org
> Subject: Re: [Paraview] Paraview and tiled display
>
> Greg,
>
> I've committed a fix for this issue on to the ParaView CVS head. Once
> you confirm that the issue is resolved, I'll move the patch to 3.6
> branch and it will be included in 3.6.2 patch release.
>
> Attached is a patch for the changed files against the 3.6 branch to
> make it easier for you to test against your current checkout.
>
> Utkarsh
>
> On Wed, Sep 2, 2009 at 5:31 PM, Greg Abram<gda at tacc.utexas.edu> wrote:
>> Hi, Berk -
>>
>> Thanks for the prompt reply.  I'm using 3.6.
>>
>> Greg
>>
>> Gregory D. Abram
>> Texas Advanced Computing Center
>> JJ Pickle Research Campus - ROC 1.101 (R8700)
>> 10100 Burnet Road
>> Austin, Texas 78758-4497
>>
>> gda at tacc.utexas.edu
>> (512) 471-8196
>>
>>
>> ________________________________________
>> From: Berk Geveci [berk.geveci at kitware.com]
>> Sent: Wednesday, September 02, 2009 3:44 PM
>> To: Greg Abram
>> Cc: paraview at paraview.org
>> Subject: Re: [Paraview] Paraview and tiled display
>>
>> Hi Greg,
>>
>> I guess we need to do better testing of tiled display rendering. This
>> is a bug that I can reproduce in 3.7 (which version are you using?).
>> We will fix this asap.
>>
>> -berk
>>
>>
>> On Wed, Sep 2, 2009 at 4:00 PM, Greg Abram<gda at tacc.utexas.edu> wrote:
>>>
>>> Can't believe I never tried this before - I'm trying to use PV to render to a tiled display with 2 nodes driving 2 screens each in TwinView mode, and oriented one above the other.  Initially, all is well - the target screens show the grey background with the axes centered and the screens are filled.  Just what you'd expect.  Then, as soon as I put some geometry into the display (such as the Sphere source) smaller windows appear on the tile displays, one showing the top half of the correct image and the other the bottom, but clipped to the bounding box of the data.  Like the results of the first pass of a binary swap compositing -  data from both processes shows up in each.   When these windows appear, the rest of the tile displays revert to the original desktop background.  If I add --use-offscreen-rendering to the pvservers the two smaller windows still appear, but are black.
>>>
>>> So - how do I get rid of those windows?   And is there a way to suppress rendering to the PV client?
>>>
>>> Thanks,
>>> Greg
>>>
>>> Gregory D. Abram
>>> Texas Advanced Computing Center
>>> JJ Pickle Research Campus - ROC 1.101 (R8700)
>>> 10100 Burnet Road
>>> Austin, Texas 78758-4497
>>>
>>> gda at tacc.utexas.edu
>>> (512) 471-8196
>>>
>>> _______________________________________________
>>> Powered by www.kitware.com
>>>
>>> Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.html
>>>
>>> Please keep messages on-topic and check the ParaView Wiki at: http://paraview.org/Wiki/ParaView
>>>
>>> Follow this link to subscribe/unsubscribe:
>>> http://www.paraview.org/mailman/listinfo/paraview
>>>
>> _______________________________________________
>> Powered by www.kitware.com
>>
>> Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.html
>>
>> Please keep messages on-topic and check the ParaView Wiki at: http://paraview.org/Wiki/ParaView
>>
>> Follow this link to subscribe/unsubscribe:
>> http://www.paraview.org/mailman/listinfo/paraview
>>
>