Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 14 Jul 1997 15:44:19 +0300
From:      Nadav Eiron <nadav@barcode.co.il>
To:        dmaddox@scsn.net
Cc:        Wes Peters - Softweyr LLC <softweyr@xmission.com>, Nick Johnson <spatula@gulf.net>, questions@FreeBSD.ORG
Subject:   Re: A few solutions
Message-ID:  <33CA1F23.1876@barcode.co.il>
References:  <Pine.BSI.3.96.970713130431.16120A-100000@pompano.pcola.gulf.net> <199707140404.WAA07219@xmission.xmission.com> <19970714072626.64852@scsn.net> <33CA1547.AD3@barcode.co.il> <19970714082521.61598@scsn.net>

Next in thread | Previous in thread | Raw E-Mail | Index | Archive | Help
Donald J. Maddox wrote:
> 
> On Mon, Jul 14, 1997 at 03:02:15PM +0300, Nadav Eiron wrote:
   [snip]
> 
> Well, all I can say is that according to the User's Manual for my
> ASUS P/I-P55T2P4 mainboard, all I need for ECC is parity RAM, and
> ECC enabled.
> 
> I notice, however, that the ECC supported by this MB only supports
> 1-bit error correction...  Maybe the ECC RAM you two are speaking of
> allows for more sophisticated error correction?

Probably. Saying they correct 1 bit is not a complete specification of
the strength of the code they use. You need to know one bit out of how
many. Standard parity RAM uses one extra bit per byte, and can detect
one bad bit per byte. By aggregating these extra bits from more than one
byte you may be able to, say, correct one bit in 32bits or such (sorry,
didn't make the calculations, I took coding theory back when I was doing
my M.Sc. and I don't remember that much :-( ), but depending on the kind
of errors you have, this may be better *or* worse than just parity. If
you have 32bits with 4 bad bits, each in a different byte, standard
parity will detect it, but this type of ECC won't. There is no such
thing as a free lunch, at least not in coding.

Nadav



Want to link to this message? Use this URL: <http://docs.FreeBSD.org/cgi/mid.cgi?33CA1F23.1876>