regarding the sluggish behaviour of X during other activity, NetBSD used
to have a couple pretty intensely bad scheduling bugs that caused X (among
other things) in certain conditions to become basically unusable.  as i
recall, ross harvey fixed them a couple of years ago, and it made a *huge*
difference.  the `barely noticable' effect on NetBSD did not exist 3 years
ago... not that the disk-cache-consumes-all-available-memory issue isn't
real and should be addressed.