Based on what you wrote so far, it could be Linux or application caching at work. Has the node that's using 440MB of RAM been up longer than the 63MB one? Run free -m if you can. Daemons like spamd / mysqld / apache "build" their cache over time. Hence my asking about uptime.
As Unixy said, make sure you are comparing real memory usage compared to buffered/cached memory. Also go through your processes to isolate any differences between the two servers and/or what the culprit could possibly be.