emanager regression traverse problem

hi all,

today my teammates say :"hi,you should do some traverses".........

there are fourteen thousand legal input parameters more or less ,because before i found bugs which should be found at the very beginning, i usually use random,but the quantity is not large,and moreover,i pay too much focus on the parameter edges, at a result ,the more normal ones (bugs) should be found  but  they weren't,i was so disappointed.  finally i wanna do one traverse,i use emanager, you can change the parameter count in vsif files, then there is one question here:should i pass the parameter count in the vsif file in to my e -env or my compile scripts????? 

one disappointed man wait & thank you for the answers