🚀 go-pugleaf

In article <C17oF6.L61@inews.Intel.COM> thamilto@fnugget.intel.com (Tony Hamilton) writes:
>                                                   Anyway... the manual
>suggests 80ns RAM, or faster. Because it was available and _very_ cheap, I
>popped in 100ns 1x9's in all 8 extra sockets, giving me 12MB total. Everything
>works fine, and I can verify the RAM is working for my applications which
>can use it.
>
>My question is: every tech person I talk to says not to do this, but they
>can't explain the reasoning behind that caution. I've heard it may have
>something to do with the bus and addressing, but I don't see any problems with
>the integrity of the memory. Should I be able to get away with this, or is
>there still a need for concern?
>
>Tony Hamilton
>thamilto@pcocd2.intel.com
There's an excellant chance that a certain amount of the slower spec'ed
memory will work.  Just don't expect a guarantee.  The faster memory
has been certified to run at that speed.  The slower memory chips were
"fall-outs" from the faster tests.  As I understand it the same goes for
CPU chips.  You can expect a certain amount of success from running them
faster, but be warned that if it does you can feel lucky.  

I am also using 100ns 1x9s on a faster system.  It's been running just 
fine for the last year.

--
Larry Autry
Silicon Graphics, St. Louis
autry@sgi.com