There is no fixed relationship; render time is dependent upon the complexity of the math involved in resolving light interaction in the scene (when Maxwell is referred to as a light simulator, that is an entirely accurate characterization). As such, more purely reflective materials can cause longer render times, since they cause light to continue bouncing around in the scene; more diffuse materials, on the other hand, may scatter the light paths they receive in directions where they more likely to terminate. That is not a general rule, though, and no such rule can actually be stated, since it depends upon the nature of each individual scene.
Assuming a Sketchup woodgrain texture, or imported jpeg of a carpet image, how might one optomize the roughness and SL level for best rendering results?
Just to clarify, there is no such thing as optimization of SL; Sampling Level simply means the level of refinement at which the calculation is stopped. As I say, Maxwell is happy to continue rendering forever if you let it, progressively reaching higher and higher sampling levels.
On the question of woodgrain, carpet, or otherwise, then, the answer goes to what I state above: it depends on the nature of the scene in question. A mirror surface reflecting into nowhere is highly efficient, since the rays it reflects bounce off into space, thereby contributing nothing more to the lighting of the scene. If it reflects into another mirror though, and vice versa, the rays will only stop bouncing between the two once the inefficiency of each has reduced their energy to a sufficiently low level. A purely diffuse material reflects light rays in a perfectly random manner; here, consider a diffuse sphere lying on a diffuse plane. When light from a source strikes the sphere, a certain percentage of rays are reflected to the plane, and are in turn divided again with some finally reaching the camera. That is to say, with such a material, a good deal of the light energy is used up in the immediate region of the sphere.
Consider though, the same sphere, with an only slightly-rough material, placed inside of a box with a similar material, where the light of a single spotlight has been directed to fall upon the sphere. Consider how the light is scattered from the sphere's surface, to the inner surfaces of the box, with some of the resulting rays being directed again back toward the sphere. Eventually, some of these rays reach the camera, producing an image. Other rays make their way from the spotlight, bounce around in the box many times, being divided and reflected according to the roughness of the materials used at each bounce, and never do end up contributing very much to the image.
Keep in mind as well, that colors should be seen in terms of their affecting both visual color, and the efficiency of light interaction. Meaning, a black object reflects no light; it has zero efficiency. A pure white object, on the other hand, is perfectly efficient, and probably does not actually exist in reality; this is why it is generally recommended to keep RGB color components (meaning the individual R, G, and B values) in Maxwell in the 220-230 range.
So to sum up, an answer to your question cannot be flatly stated. There exist general situations to avoid, many of which can be inferred from what I write above, but specific advice can only be given for optimizing specific scenes when one has knowledge of how those scenes are constructed. If you have a particular problem, you can post about it (probably better to do that in the Maxwell Render forum
, rather than this one, which is for plugin-specific questions), including images (you can upload to imageshack.us or similar and link using the Img button) and a description of what you are trying to do, and what problems you are running into.