home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!TOROLAB6.VNET.IBM.COM
- From: ADUNSMUI@TOROLAB6.VNET.IBM.COM (Al Dunsmuir)
- Message-ID: <19920828.144500.49@almaden.ibm.com>
- Date: Fri, 28 Aug 92 17:29:18 EDT
- Newsgroups: comp.os.os2.misc
- Subject: Re: OS2 WINS!! I give up!
- Organization: IBM Canada Laboratory Ltd, Toronto, CANADA
- Disclaimer: This posting represents the poster's views, not those of IBM
- News-Software: UReply 3.0
- References: <1992Aug27.015416.6882@maccs.dcss.mcmaster.ca> <1992Aug27.030506.4609@midway.uchicago.edu> <liuyu.714945700@kramden>
- <1992Aug28.200235.7250@njitgw.njit.edu>
- Lines: 51
-
- In <1992Aug28.200235.7250@njitgw.njit.edu> David Charlap writes:
- >In article <liuyu.714945700@kramden> liuyu@kramden.nyu.edu (Liuyu) writes:
- >>>Due to some recent experience one should also suspect the RAM cache.
- >>>Some vendors, it seems, cut corners and supply slightly out of spec
- >>>cache memory, substituting cache that is too slow for the system.
- >>>This, too, can result in memory parity errors.
- >>
- >>Can you explain why this happens only under OS/2? Is it just another excuse,
- >>blaming hardware? Does it happen under other complex programs like Unix or
- >>Novell? Thanks.
- >
- >OS/2, due to it s high amount of context switching, has some strange
- >ability to push intermittently bad memory over the edge. It's
- >happened on many computers. Replacing the memory usually works to
- >cure it.
- >
- >My father's computer at work had its memory "pushed over the edge" by
- >Windows 3.1. So it's nothing inherent in OS/2. (And when Windows 3.1
- >died, it took most of his /Windows directory with it - a complete
- >re-install of Windows and all Windows apps had to be performed).
- >Again, replacing the memory cured everything.
- >
- >--
- > |) David Charlap "I don't even represent myself
- > /|_ dic5340@hertz.njit.edu sometimes so NJIT is right out!.
- > ((|,)
- > ~|~ Hi! I am a .signature virus, copy me into your .signature file.
- >
-
- Some of the extra stress is due to two differences between when one executes
- a program from memory, and when one simply reads/writes that memory using a
- program (such as a DOS VDISK device driver).
-
- A) A DOS device driver typically will access a smallish block of the
- memory (copying data in or out), then not touch the card for what
- appears to the card to be a relatively long time. If the refresh
- circuitry doesn't properly handle the cases when a particular
- range of addresses is used without a break (code in a long, tight
- loop), your memory won't get refreshed properly and you lose data.
-
- A) When executing a program the address lines change in a fairly random
- manner. Some memory cards (or memory chips) can keep up with relatively
- slow changes, but aren't fast enough to keep up with the instruction
- addresses. Sooner or later the card fetches storage from the wrong
- memory address, and your program dies a horrible death.
-
- Al Dunsmuir Internet: adunsmui@torolab6.vnet.ibm.com
- CODE Development II + TEAM OS/2 Voice: (416) 448-3315
- IBM Canada Laboratory Ltd, Toronto, CANADA
-
- OS/2 V2: Why settle for anything less?
-