Hi Jan,

On 9 May 2015 at 13:20, Jan Ciger <[email protected]> wrote:
> If we were to replace the current osg::Vec/osg::Matrix math with a
> template based one in the future then it would be better to just go
> with Eigen (http://eigen.tuxfamily.org/index.php?title=Main_Page )
> instead of reinventing the wheel again.
>
> Eigen is heavily optimized
> (http://eigen.tuxfamily.org/index.php?title=Main_Page#Overview )  and
> supports a lot more than what the current osg::Vec/osg::Math does. It
> is also "lightweight" as a dependency, as there are no libraries to
> link (header-only library).
>
> However, I feel that replacing the current math would be a large
> project with a lots of potential for breakage (e.g. alignment issues,
> different in-memory representation, etc.), so tread carefully here.

For the most part the OSG does occasional vector math computations on
small blocks of data here SIMD optimization are of little use.  The
main use of such optimizations would be animation of large number of
vertices, but these days this is something best deferred to the GPU.
For the core scene graph I don't see SIMD optimizations being of
importance to scene graph, and going forward I don't expect it to
become any more important as we move more and more operations to the
GPU.

For specific places I can see value in SIMD optimization, but it's not
a core scene graph issue.

Robert.
_______________________________________________
osg-submissions mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org

Reply via email to