No subject

[utzoo!decvax!ucbvax!unix-wizards]

>From obrien at RAND-UNIX Tue Oct 27 21:46:34 1981
	The general feeling is that the 11 has too few paging registers to
make a paging scheme worthwhile on this architecture.  Also, except on
70's, you don't have I/O paging hardware available, so swapping scattered
pieces of a process is hard and raw I/O becomes impossible.  One or two
people actually built, or tried to build, a paging PDP-11 UNIX, but either
gave it up or it never caught on.

        Things like VAXen are a different story.  There, while the
text/data/bss trichotomy has been retained at the user level, in fact
things are done in discreet pages and swapped that way.  UNIX/32V does
"scatter/gather" swapping, where a process can be scattered throughout
memory but must be all in-core to run (i.e. no page faults allowed), so it
looks like the kind of UNIX you have in mind.  Given the notion of enough
page registers, though, and enough I/O paging hardware as well, the
continuation to the Berkeley VMUNIX fully paged system is natural enough.
Paging to a solid-state device would be wonderful and I wonder why no one's
tried it.

	DEC OS's, I believe, are worse than UNIX.  They've only recently
started supporting split I/D on 11's.