Xenix 386 v2.3 bug?

[Paul De Bra]

In article <401 at impch.UUCP> egon at impch.UUCP (Lukas Knobloch) writes:
>
>Hiya
>
>From time to time we've got problems with our System it means:
>
>panic:Memory failure - parity error
>
>Hit any......
>
>does anybody know a prevention against this???????? :-(
>i'll think it's the EGA but i'm a bit unsure

You don't give much info on your system, but let me try to give some
general advice, cause I've had to tackle this problem several times before.

The most likely cause is that your memory chips are not quite fast enough.
If you have a "cheap" clone the memory is usually barely fast enough and
sometimes fails. It often is a matter of just that one nanosecond too slow
to perform reliably. (It need not be just the memory chips, could be the
supporting hardware as well.)

Other possibility is that some periferal board is not up to your bus speed.

I have been able to solve several of these problems by changing the positions
of the boards. On a generic AT the rightmost slots (towards the power supply)
give you a bit more flexibility than the leftmost slots. Usually a display
adapter is put in the leftmost slot. Moving it as far to the right as possible
can help. Also, if you have memory boards (you may not have any on the 16-bit
bus since you mention 80386) the EGA-card may not like to be far away from the
memory boards. I found one instance where a system with 2 memory boards would
only function if the EGA-card was installed right between the 2 memory boards.
(Putting the boards in adjacent slots and the EGA-card next to them didn't
help. The EGA had to sit between the memory cards.)

Maybe you may think all of this is nonsense. I can assure you it is not.
I recently installed a second memory board in an 8 Mhz AT (with 1 wait state).
Both memory boards had 120ns chips, so plenty fast. Well, the startup memory
test would not even pass unless I moved the memory boards as far to the
right as possible, and the EGA next to them.

Hope this helps. Start shuffling boards!

Paul.
not 
-- 
------------------------------------------------------
|debra at research.att.com   | uunet!research!debra     |
------------------------------------------------------