Re: Custom Data Structure vs. Primitive Array objects

In last semester, I wrote a program for algorithm analysis class. It
was critical to make the program to run as fast as possible. My first
approach was making custom classes that can represent the required data
model for the given problem. But, the program ran extremely slow with
a large data set. I suffered from that performance problem and I
decided to rewrite the program with only primitive data classes that
provided in Java for default. obviously, there was no change in
programming logics or algorithms. It began to ran comparatively faster
than before. (It was amazingly reduced the running time.) This was a
weird observation for me because the custom data structure (class) that
I wrote before was just a data container with few methods for data
field accessors (getters/setters); there was no intensive computation
in the class. I'm just reasoning the observation by a guess. "Java
might handling custom objects slow due to the garbage collector
mechanism and maintaining the objects (retain counter, etc)." But, this
is just a guess. I have no clear reason to explain what I observed.
Would this happen in other programming language/environment, too?