This is a function of the way Linux reports threads - nothing to be alarmed about. Your memory footprint is no where near what you think it is, too.
Not to worry. As Crackers points out this is function of
how Linux displays threads as processes. It is a little
confusing, but do a ps-ef and look at the PPID for the
JVM. Notice that they are all the same number? PPID is
the Parent Process ID.
It might "appear" that your JVM is taking up heaps of
memory, what is actually happening is the threads
as "processes" are mapped into the same address space
as the PPID.
Hope this helps.
Thanks for the replies guys. That's what I thought but I feel better now that my suspicions have been confirmed. I've been away from *Unix* for a few years and my background previously was SunOS/Solaris. The man page seemed to indicate that threads weren't reported by default (ps -e) but I guess I was mistaken.