Java EE (Java Enterprise Edition) General Discussion

I need to parse 1000+ xml streams continuously one after another.
Can any one have idea about SAXBuilder (org.jdom), SAXParser?
Among these two which holds better performance in terms of memory utilization as primary criteria?
As per JDOM documentation SAXBuilder constructs a tree in memory.
I've tried by parsing 100 files one after another each of size 3MB.
SAXBuilder consumed less memory comparatively SAXParser I thought SAXBuilder utilized more memory because it has to construct a tree in memory. I could not able to identify reason for this.

Finally my requirement is to parse 1000 + with less memory utilization.