[vtk-developers] vtkRectilinearGrid speed-up in FindPoint + ComputeStructuredCoordinates

Bryn Lloyd lloyd at itis.ethz.ch
Mon Apr 30 02:16:18 EDT 2012


Cory,

Thanks for the link.

I followed the explanation and posted the changes in the topic:
http://review.source.kitware.com/#/c/5412/

Bryn



On 4/29/2012 2:17 AM, Cory Quammen wrote:
> Bryn,
>
> If you follow the steps at this page, your changes should get
> incorporated fairly quickly:
>
> http://www.vtk.org/Wiki/VTK/Git/Develop
>
> Patches I've sent through have been merged to master in a day or two
> at most with the help of great and active reviewers.
>
> Cory
>
> On Sat, Apr 28, 2012 at 8:07 PM, Bryn Lloyd<lloyd at itis.ethz.ch>  wrote:
>> Hi
>>
>> I've been trying to make our stream-line visualization faster. As a result I
>> found that the vtkRectilinearGrid is using linear algorithms to locate both
>> cells and points.
>> A binary search would be more optimal (logarithmic) for medium to large
>> sized grids.
>>
>> I implemented a new version of vtkRectilinearGrid::FindPoint and
>> vtkRectilinearGrid::ComputeStructuredCoordinates, which use a binary search.
>> The speed up for a grid of dimensions around 100*100*200 is close to a
>> factor of 10 (this is a small-medium sized grid for our application).
>> Clearly, it would be great if these changes could enter the next release of
>> VTK, since they will make interpolation (used in stream-lines) and other
>> algorithms quite a bit faster for rectilinear grids.
>>
>> How can I contribute these changes such that they have a chance to enter in
>> the next release.
>> I have written tests, which make sure the result is still the same as with
>> old implementation and am prepared to put some effort into
>> explaining/documenting/testing the changes as needed.
>>
>> Thanks&  regards
>> Bryn
>>
>> _______________________________________________
>> Powered by www.kitware.com
>>
>> Visit other Kitware open-source projects at
>> http://www.kitware.com/opensource/opensource.html
>>
>> Follow this link to subscribe/unsubscribe:
>> http://www.vtk.org/mailman/listinfo/vtk-developers
>>
>
>



More information about the vtk-developers mailing list