I will be investigating use of the Math Kernal Library to speed up signal processing code and was wondering if anyone has achieved vast improvement from its use. Also, are there any gotchas that I should be aware of.