> I recall the original proposal was for compressed virtual memory, that to
> me translates to disk space - unless the box is diskless on a network.

Actually, the big potential win on "compressed VM" is not in disk
_space_, even on local disks; local disk is cheap.

It's in disk accesses, which are incredibly slow, even compared to the
time it takes to uncompress a page on a "reasonable" cpu.

On faster CPUs (oh, say, like the 233MHz 21064As sitting next to me),
then the difference becomes even greater.



chris