home *** CD-ROM | disk | FTP | other *** search
- Path: EU.net!sun4nl!xs4all!usenet
- From: jtv@xs4all.nl (Jeroen T. Vermeulen)
- Newsgroups: comp.sys.amiga.misc
- Subject: Re: New Press Release!
- Date: Wed, 27 Mar 96 22:06:21
- Organization: Leiden University, Mathematics & Computer Science, The Netherlands
- Distribution: world
- Message-ID: <19960327.7BACD20.13AE9@asd11-03.dial.xs4all.nl>
- References: <4iofm6$e4j@serpens.rhein.de> <DoLC38.2G4@info.uucp> <4ir19b$mml@serpens.rhein.de> <oj6wx4dne22.fsf@hpsrk.fc.hp.com> <4ivdqe$62a@serpens.rhein.de> <19960323.75240E0.D624@contessa.phone.net> <4j37re$k6q@serpens.rhein.de> <19960325.7665B58.91C6@contessa.phone.net>
- NNTP-Posting-Host: asd11-03.dial.xs4all.nl
- Mime-Version: 1.0
- Content-Type: text/plain; charset=iso-8859-1
- Content-Transfer-Encoding: 8bit
- X-NewsSoftware: GRn 2.1 Feb 19, 1994
-
- In article <19960325.7665B58.91C6@contessa.phone.net> mwm@contessa.phone.net
- (Mike Meyer) writes:
-
- > Thrashing is normally used to indicate the situation where something
- > is BROKEN in the demand paging system, such that you're spending more
- > time waiting on pages than performing user functionality.
-
- Thrashing is what occurs when your working set exceeds the size of your frame
- set (physical memory), so page replacement algorithms that are normally
- near-optimal such as LRU hit their worst case and every page in the working set
- is swapped out and back in again during every iteration of your outer loop.
-
- [ When thrashing occurs, you might actually be better of not using a stack
- algorithm and just taking "pot luck" on which pages are kept in core. This does
- not mean there is something broken in the demand paging system; that would seem
- more likely for pre-paging systems. ]
-
- Sometimes there is simply no way of reducing your working set; the program is
- going to thrash in that case and there's nothing you can do but sit and wait.
- Hence I think Michael is right in saying that thrashing is acceptable for batch
- programs (if I understand him correctly): Users don't hold their breath while
- waiting for batch jobs to finish.
-
-
- > Of course, none of paging, swapping or thrashing are acceptable in
- > systems that demand quick real-time response.
-
- Perhaps this should be: "none of paging, swapping or thrashing are acceptable
- for *processes* that demand quick real-time response". Swapping out non-realtime
- code and data could actually improve real-time response in some cases because
- you have some more memory to trade off for speed.
-
- --
- ============================================================================
- # Jeroen T. Vermeulen \"How are we doing kid?"/ Yes, we use Amigas. #
- #--- jtv@xs4all.nl ---\"Oh, same as always."/-- ... --#
- #jvermeul@wi.leidenuniv.nl \ "That bad, huh?" / Got a problem with that? #
- "I'm very superstitious about walking under falling masonry"
-