Re: TWGS questions

[supertimer@aol.com (Supertimer)]

spec@vax2.concordia.ca (Mitchell Spector) wrote:

>>If it were the cable, then 16 and 8 bit software will fail at the
>>same time.
>
>    Not neccesarily, as I stated above, it seems only logical that
>16-bit software is more prone to crashes and lock-ups than 8-bit
>software.

It might be true unless your accelerator only has an 8k
cache.  Then, naturally, the pattern of crashing reflects
the CPU because the older CPUs had problems with
certain 16-bit instructions.

I've had a Windows PC exhibit something similar.  After
a time, the CPU failed to run correctly at its rated speed.
I had to underclock it.  When clocked at its rated speed,
the degraded chip would run perfectly under DOS, but
would crash under Windows 95 while booting (Windows
protection fault error).

Selective failure like this, especially when the size of
the program is made less relevant by a small cache,
is indicative of CPU failure.