MapBufferObject/UnmapBufferObject timings problem

I want to use VBO with CUDA but I have a problem, the cudaGLMapBufferObject and cudaGLUnmapBufferObject functions takes too long to execute (16ms). With that time I can only have 14ms in CUDA if I want 30ms execution time per frame.

I would like to contact CUDA programmers, but I only have this place where I can get some support.

I want to handle big meshes, and keep it in GPU main memory, that´s why I need VBO.

Somebody help me please!!!