Hi Kim, Chris,
If you're interested in the timing differences between FFTSS and FFTW I found that FFTSS took approximately twice as long to compute 1000 64x64 complex to complex FFTs. My tests were fairly rudimentary though. I believe J-S found less of a difference when he tested the timings on his machine.
Yes, I found negligible difference between FFTW single, FFTW double and FFTSS. Not sure what it depends on, but I have a quad-core with 4GB RAM. And since we're talking about something that takes just a few hundredths of a second anyways, the difference is even less significant... Even if it took twice the time, it wouldn't change much in practice.
Creating the mesh from the FFT results is what takes most time. You'll see if you change some FFT settings that dirty the ocean mesh (in the oceanExample) it takes a few seconds before the next frame is drawn. This is done in the update traversal when the dirty flag is set on the OceanTechnique.
If you're testing out parameters to customize the visual look of the ocean, you can disable automatic dirtying of the geometry, so that it doesn't regenerate each time you make a change. Press capital 'D' to disable this, and then change all your parameters and press small 'd' to dirty the geometry manually when you want to see the result.
I'd like to eventually refactor the OceanTechnique a bit. Right now it combines the generation function (FFT) and the meshing (geomipmapping). Separating the two would allow us to change one independently of the other. That's one of the things I'll be doing in the near future.
J-S -- ______________________________________________________ Jean-Sebastien Guay [email protected] http://www.cm-labs.com/ http://whitestar02.webhop.org/ _______________________________________________ osg-users mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
