performance is an issue

Really? So what where the results of your benchmark? How much faster does it need to be? Did you profile the code and determine that this area is the bottleneck?

Although those questions look somewhat snide, in fact they are important. Unless you know what you are aiming for any why when trying to "make code efficient" you are most likely wasting time. If you need a 10 thousand times speed up then the place to look for a solution is likely to be vastly different than if you only need a 10 times speed up. But chances are there is a better solution (like cashing results in a database) that don't need this specific code to be any faster at all.

Dramatic improvements in code execution time don't generally come from fiddling with implementation. They come from reconsidering the problem and developing or finding a better algorithm. Almost always that requires an understanding of the bigger picture.