Boston Linux & Unix (BLU) Home | Calendar | Mail Lists | List Archives | Desktop SIG | Hardware Hacking SIG
Wiki | Flickr | PicasaWeb | Video | Maps & Directions | Installfests | Keysignings
Linux Cafe | Meeting Notes | Blog | Linux Links | Bling | About BLU

BLU Discuss list archive


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Request for assistance



Mark Glassberg Fornarotto jun-30-96 wrote in a message to Mike Bilow:

 MGFj> I did not intend my message to give the impression that I
 MGFj> hacked the  solution myself.  I was instructed to disable
 MGFj> parity checking by a technician at the firm which made my
 MGFj> Composite SIMMs.  I appreciate your warning and have enabled
 MGFj> this function with no ill effects on either DOS or Linux.

This is something like taking your car back to the place where you just paid
them to rebuild the transmission in your car:

You: It makes this loud grinding noise.

Them: No problem, we'll put some sound-deadening insulation around it.

You: A little while after the noise starts, the car won't move.

Them: No problem, we'll give you this handy "Call for Help" flag.

Be aware that some SIMM makers will go so far as to generate synthetic parity,
making the bit come out right all the time to fake out the motherboard rather
than storing and checking it to maintain the validity of the data.

 MGFj> On the issue of mixing memory, I bought the six 70nS SIMMs
 MGFj> from a technician who has worked on my system in the past
 MGFj> and is familiar with it.  He stated that he couldn't get 80s
 MGFj> but that the 70s would work if I put my ("tested at 70nS")
 MGFj> 80s above them, which seems to be the case.

It is always absolutely safe to put faster memory in the place of slower
memory.  If the motherboard calls for 80 ns and you used 70 ns, that is safe.

I have seen somes cases where the company that prints the numbers on the chips
is lying.  In fact, I have seen SIMMs where the numbers put on the chips by the
maker of the chips were obviously replaced by new numbers put on by the maker
of the SIMMs, and I would certainly avoid things like this.

 MGFj> I would very much appreciate your explaining the risks in
 MGFj> this approach, so that I can evaluate the benefits of
 MGFj> replacing the 80s with 70s.

If your motherboard needs 70 ns chips, then you should not be using 80 ns
chips.  If your motherboard needs 80 ns chips, you can mix 70 and 80 ns chips
freely.  This "80 ns tested at 70 ns" business is nonsense: if the chip maker
could have certified the chips at 70 ns, they would have done so.

What is more disturbing is that you mentioned that you bought six SIMMs.  While
you can freely mix speeds of memory, you cannot mix types of memory within the
same bank.  If you have a 32-bit processor such as a 386DX or any 486, then a
bank is comprised of either one 72-pin SIMM or four 30-pin SIMMs.  If you made
up a bank of four SIMMs by using two that were retrofitted from 256 Kb SIMMs
and two that were designed properly, then this is a bad thing.
 
-- Mike




BLU is a member of BostonUserGroups
BLU is a member of BostonUserGroups
We also thank MIT for the use of their facilities.

Valid HTML 4.01! Valid CSS!



Boston Linux & Unix / webmaster@blu.org