Each NUKE working file is unique. From a technical point, there can’t be a one-fits-all solution to generate the fastest output possible. It is crucial to understand that simply bumping up threads in smartRender will not always necessarily mean a faster result. Often times the key in getting the fastest possible result is in having an optimized working file. There are many good articles out there that discuss best practices for Nuke node trees. Discussing these here are out of the bounds of this documentation. Following best practices and creating clean working files is a good starting point to guarantee quick results.
Keep also in mind that experimenting with the amount of threads is sometimes key. While one working file might work well when setting up a high amount of render threads will probably not work the same for another working file.
Please have a look at the smartRender faq which provide dedicated sections that show you how you can trouble shoot if a working file is not rendering as expected.