Path: utzoo!attcan!uunet!tut.cis.ohio-state.edu!ucbvax!agate!darkstar!wilbur.nas.nasa.gov
From: ciotti@wilbur.nas.nasa.gov (Robert B. Ciotti)
Newsgroups: comp.os.research
Subject: Re: swapping intelligently
Message-ID: <2531@darkstar.ucsc.edu>
Date: 11 Apr 90 18:53:53 GMT
Sender: usenet@darkstar.ucsc.edu
Organization: NASA Ames Research Center, Moffett Field, CA
Lines: 26
Approved: comp-os-research@jupiter.ucsc.edu


In article <2402@darkstar.ucsc.edu> skh@vi.ri.cmu.edu (Steve Handerson) writes:
>
>
>When I was working with lisp I'd have loved to limit the amount
>of memory that lisp could hog for paging.
...
>swapping resources.  I don't care if lisp grinds to a halt,
>if I can edit unhampered at the same time (if my editor's swap
>requests always take priority).
>
>Do any operating systems already do this?
>
>-- Steve

The 205/ETA10 (possibly the 203) machines from Control Data running the 
VSOS/EOS operating systems had a feature to limit/allocate the number of 
working set pages for a process. It would be interesting to know if this
made it into VE (Another CDC proprietary OS). This particularly necessary 
feature however never made it into UNIX for the ETA machines and was really 
a problem because a single process (say doing a wide stride on a very large 
vector), could bring the whole machine to a halt. I was going to go in and 
modify the page stealer to limit working set sizes by stealing pages from 
the process causing this type of behavior, but time was short...

 Bob