[Beowulf] Re: Beowulf Digest, Vol 10, Issue 16

Bogdan Costescu Bogdan.Costescu at iwr.uni-heidelberg.de
Tue Dec 14 10:39:37 PST 2004


[ took out the cc-list, everybody's subscribed ]

On Tue, 14 Dec 2004, Mark Hahn wrote:

> > The consensus seems to be that registered ECC recycled from the
> > S2466N systems will not work in boards designed for unbuffered
> > ECC, which means that both the 754 and 939 are out.
> I believe the relation is the other way: you cannot use unbuffered
> dimms in a board which assumes buffered dimms.

Hmm, my experience was with an Intel 875P based mainboard that was
advertised as being able to use only unbuffered ECC: it did work with
unbuffered ECC, but not with buffered ECC - both memory modules were
from Kingston and the only difference in specs was the buffering
thing.

Furthermore, it was my understading that, for systems that did not
have support for ECC, using unbuffered ECC was still possible, as the
ECC-related signals were not connected or ignored, but the rest of the
signalling was the same. However, buffered ECC would mostly not work
in such systems - and this was later supported by practice: a buffered
ECC module (incidentally also from a retired Tyan S2460 system) did
not work in most non-ECC mainboards that I've had on hand to try; it
only worked in an AMD760-based single CPU board that was also not
advertised to support ECC, but I guess that the pins were not
connected while the signalling was just common in the chipset with the
MP version.

This data supports the original statement, while not really
contradicting Mark's ;-)

-- 
Bogdan Costescu

IWR - Interdisziplinaeres Zentrum fuer Wissenschaftliches Rechnen
Universitaet Heidelberg, INF 368, D-69120 Heidelberg, GERMANY
Telephone: +49 6221 54 8869, Telefax: +49 6221 54 8868
E-mail: Bogdan.Costescu at IWR.Uni-Heidelberg.De




More information about the Beowulf mailing list