Originally Posted by rangrej25
process actual physical RAM usage
That isn't a well defined quantity. For any useful definition, it isn't easily measured.
The RES column in top
(and available many other places) is probably the best easy to get approximation of the each process's physical ram use.
Some physical ram will be counted multiple times if you add up the RES columns. So RES is usually a slight over estimate of physical ram use.
Some pages may be frequently soft faulted between a process and the page cache (so part of a meaningful physical ram use of that process, but not included in RES). So RES could be an under estimate of physical ram use. Most of cache memory is effectively free. But some of it is actively soft faulted. I don't think there is a decent way to measure what fraction of cache should be interpreted as free.