Date: Wed, 30 Aug 1995 15:36:30 -0500 (CDT) From: Jim Bryant <jbryant@argus.iadfw.net> To: freebsd-hackers@freebsd.org Subject: Re: S.O.S -2.1Stable and ASUSP54TP4 (fwd) Message-ID: <199508302036.PAA03912@argus.iadfw.net>
next in thread | raw e-mail | index | archive | help
In reply:
>From jbryant Wed Aug 30 15:36:00 1995
Subject: Re: S.O.S -2.1Stable and ASUSP54TP4
To: rgrimes@gndrsh.aac.dev.com (Rodney W. Grimes)
Date: Wed, 30 Aug 1995 15:36:00 -0500 (CDT)
In-Reply-To: <199508291458.HAA03827@gndrsh.aac.dev.com> from "Rodney W. Grimes" at Aug 29, 95 07:58:01 am
X-Mailer: ELM [version 2.4 PL24]
Content-Type: text
Content-Length: 2770
In reply:
My point is that if he was locking up due to a memory error, at least parity might
trap and show WHERE the error might be...
> > > Are you using non-parity memory? As you know, memory errors will not trap
> > > unless you are using parity memory. If you want to use FreeBSD for anything
> > > other than home-use, I strongly suggest parity memory. (Get a warrantee!)
> >
> > Actually, no triton based board traps parity errors. If you need what itsy-
> > bitsy comfort parity checks give you then you are going to have to sacrifice
> > performance and use a neptune based board.
Not any more, apparently my ASUS manual shows that parity memory is acceptable,
although it does not say if it will or will not trap.
> And to state my reason for agreement that parity is a ``itsy-bitsy comfort'',
> think about the fact that 80% of your memory access are going to a L2
> cache that has never had parity on it, yet has a same FIT rate as the
> main memory system. Basically your more likely today to take a single
> bit error in your cache as you are in main memory :-(.
>
> > Personally I think parity memory is a poor investment; assuming I'm clear
> > on the techical details, once the P6 becomes a commodity item (real soon
> > now) the next motherboard you buy will want ECC memory, anyway. Its neat
> > the way this industry creates its own demand, eh?
>
> :-). Yes, but unfortanelty, ECC memory, though being supported by next
> generation chip sets, will be, IMHO, a rare thing in the PC market simply
> due to clone MB manufactures who will see it as an additional cost to
> them.
True, but at least parity will tell if it's bad most of the time.
As far as ECC, I really do not forsee many commodity CPU's shipping with
ECC. If DEC is any indicator of the cost of ECC memory, it's use will remain
relegated to high end systems... Getting a Gig or Two of ECC next week with our
AlphaServers [Either 512M/512M or 1024M/1024M, I forget, [2 Computers, 4 CPUs]...
> > The upside to this is that financial institutions and similiar nervous
> > nellies will migrate their mission-critical servers to cost-effective
> > intel platforms whereupon demand for FreeBSD will go through the roof.
> > Now is the time to buy WC stock.
> >
> > :-)
I thought FreeBSD was a seperate entity and not part of WC?
We still have a bit of work until I would call it Mission Critical
capable... Getting close though, but then again I'm picky :)
Jim
--
All opinions expressed are mine, if you | "I will not be pushed, stamped,
think otherwise, then go jump into turbid | briefed, debriefed, indexed, or
radioactive waters and yell WAHOO !!! | numbered!" - #1, "The Prisoner"
jbryant@argus.iadfw.net, Sr. System/Network Admin, Internet America
--
All opinions expressed are mine, if you | "I will not be pushed, stamped,
think otherwise, then go jump into turbid | briefed, debriefed, indexed, or
radioactive waters and yell WAHOO !!! | numbered!" - #1, "The Prisoner"
jbryant@argus.iadfw.net, Sr. System/Network Admin, Internet America
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199508302036.PAA03912>