If you are reading values, the search cursor with a where clause is faster. It also depends on what you are extracting and the size of your dataset, the gains may be minimal.
I have a script to run distance-based point clustering. it takes almost 4 mins to finish only 600 pts. do you mind to have a look? help me to facilitate the process?