OpenCL+ OpenGL

I’ve cross-posted these questions “Graphics Technology > DirectX / OpenGL & other APIs” and " CUDA GPU Computing > General CUDA GPU Computing Discussion", in hopes of an answer… but I should have posted it here first:

(1) Can on call OpenCL calls from multiple threads, in particularly not the main thread? I saw that this was not possible with CUDA, but what about OpenCL… the 1.1 spec says it should be ok, but that does not mean it actually is.

(2) Given that OpenCL and OpenGL are using the same device to do work, outside of setting the work group size (and work item size) is there a way to control “how much GPU horsepower OpenCL” will use? The use case is to use OpenCL to calculate the data for some static meshes (but the parameters and mesh generation functions determined at run time) while rendering and once the meshes are ready to use GL to render. The simplest would be something of this form: “Tell OpenCL how much percentage of GPU to use as a hint/suggestion”… it does not have to be perfect precise, but for example lets say a frame takes 16ms to render, I run a heavy OpenCL thingy and want to keep the fram render time under 40 or so ms, so I set “let OpenCL use 50%”…

Best Regards