The basic problem is this needs the __device__ (etc.) constructs in the math directory files and inlining everything in those files. And they'll need some cleanup at the same time.
Added support, and changed eavlSurfaceNormalMutator to use vector math as a test. Confirmed to work on CUDA. The newly added transform mutator also uses them.
The basic problem is this needs the
__device__
(etc.) constructs in the math directory files and inlining everything in those files. And they'll need some cleanup at the same time.