[Paraview] Using CUDA accelerated vtkm filter

Robert Maynard robert.maynard at kitware.com
Fri Jul 6 11:03:01 EDT 2018


Hi Sumin,

So it looks like the fix will require you to either build ParaView from
gitlab ( https://gitlab.kitware.com/paraview/paraview ) as the official
releases have a bug which has been fixed. If you have been building from
the release sources, building from git is very similar (
https://gitlab.kitware.com/paraview/paraview/blob/master/Documentation/dev/build.md
).

On Fri, Jul 6, 2018 at 9:57 AM Sumin Hong <sumin246 at gmail.com> wrote:

> Thanks for your reply.
>
> I tried v5.4.0 first and then v5.5.2 again. However, the same problem
> occurred in both versions.
>
> Both codes are the distribution version of the home page instead of
> superbuild.
>
> For the compile option, only PARAVIEW_USE_MPI, PARAVIEW_USE_VTKM,
> VTKm_ENABLE_CUDA, and VTKm_ENABLE_MPI are activated without touching
> anything else.
>
> Sumin
>
>
> 2018년 7월 6일 (금) 오후 9:46, Robert Maynard <robert.maynard at kitware.com>님이 작성:
>
>> Hi Sumin,
>>
>> What exact version of ParaView are you building? I have been unable to
>> reproduce this issue with git master.
>>
>> On Tue, Jul 3, 2018 at 10:41 AM Robert Maynard <
>> robert.maynard at kitware.com> wrote:
>>
>>> Hi,
>>>
>>> The "caught ErrorBadDevice: Device with id 2 is either not in the list
>>> or is invalid."  is thrown from VTK-m when somebody has asked to
>>> execute on the CUDA device adapter ( with id == 2 ) and components of
>>> VTK-m have been built without CUDA enabled. The device id has nothing
>>> to do with which specific GPU VTK-m will use.
>>>
>>> As for where this mismatch in compiling with CUDA support is
>>> occurring, somebody from the VTK-m team will look into it.
>>> On Tue, Jul 3, 2018 at 7:12 AM Sumin Hong <sumin246 at gmail.com> wrote:
>>> >
>>> > Dears users,
>>> >
>>> >
>>> > I would like to use the CUDA accelerated vtkm filter in ParaView.
>>> However, according to the release notes, binary versions do not seem to
>>> support CUDA acceleration properly. (
>>> https://www.paraview.org/Wiki/ParaView_Release_Notes).
>>> >
>>> >
>>> >  So I compiled  ParaView-v5.5.2 code on Ubuntu with VTKm_ENABLE_CUDA
>>> option.  There was no problem in the build, but when I actually execute
>>> vtkm filter (contour filter) on paraview, I got this error message  "caught
>>> ErrorBadDevice: Device with id 2 is either not in the list or is invalid."
>>> >
>>> >
>>> > This seems like the vtkm filter does not set up the proper GPU device.
>>> > The used system has installed only one GPU device (TITAN X Pascal)
>>> with CUDA 8.0.
>>> >
>>> > So I think the device ID should be 0.
>>> > Also, there is no problem when using a default contour filter without
>>> vtkm.
>>> >
>>> >
>>> >
>>> > Any help on how to solve this will be appreciated.
>>> >
>>> >
>>> > Best regards,
>>> >
>>> > Sumin.
>>> >
>>> >
>>> > _______________________________________________
>>> > Powered by www.kitware.com
>>> >
>>> > ParaView discussion is moving! Please visit
>>> https://discourse.paraview.org/ for future posts.
>>> >
>>> > 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
>>> >
>>> > Search the list archives at: http://markmail.org/search/?q=ParaView
>>> >
>>> > Follow this link to subscribe/unsubscribe:
>>> > https://public.kitware.com/mailman/listinfo/paraview
>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://public.kitware.com/pipermail/paraview/attachments/20180706/178cae5f/attachment.html>


More information about the ParaView mailing list