Working from one Cuda context with many OpenGL/DirectX windows?

I need to process data from one Cuda context and present it in several DirectX or OpenGL windows for different views (two implementations). The results are drawn using textures (Cuda data is copied into textures)

What is the best way to achieve this? (as when using cuD3D9CtxCreate and cuGLCtxCreate and the id of a specific window is passed)