More on Cyrix 6x86

[mikebw at bilow.bilow.uu.ids.net (Mike Bilow)]

I thought the list might be interested in some follow-up on the Cyrix 6x86 and
NT problem.  Cyrix tells me that their position is this:

1. Microsoft believes that there is a compatibility problem between NT 4.0 and
the Cyrix 6x86 at revision 2.6 and earlier.  Cyrix is unable to reproduce the
problem, however, on any revision of their CPU.

2. Microsoft currently ships NT 4.0 with code to disable the on-chip cache when
any Cyrix 6x86 is detected.  Within the week, they will release a service pack
for NT which will allow the cache to run enabled on a 6x86r3.7 or later.

3. It appears that Microsoft's test which resulted in their determination that
a problem existed was based on a 6x86r2.6 installed on an Asus motherboard
which was certified only for use with a 6x86r3.7 or later CPU.

In order to determine the revision level of a 6x86, Cyrix instructs as follows:

   A lot code will read something like G8XX5624A. The first four
   chars are where the silicon was built and where it was
   packaged, the 5 is the lot code(5 is a rev 2.6, 9 or A is a
   3.7), the 6 means 1996, the 24 is the work week within the year
   and the A is the lot within the week. If you are supplying
   customers systems that are non NT then there wont be an issue.

My personal opinion is that this tends to confirm my earlier suspicion about
the problem being the motherboard rather than the CPU and not specific to NT.
 
-- Mike