*** Craig Hart wrote in a message to Scott McNay:
CH> Absolutely. I have at least 5 of them here to test any time.
CH> Yes, I have tried CPUID and all sorts of combinations of
CH> cyrix config registers literally hundreds of times on these
CH> chips with no success.
I finally had a chance to run my program on one earlier today, so I got an 
answer to that a few hours early.  
 > I haven't had an opportunity to check the 6x86MX and MediaGX yet, but if
 > it is on by default, then GOOD; they finally turned their brains on...
CH> Exactly. And about time too.
They may have finally pulled their heads outta their backsides, but I think 
it's kinda late.  On the other hand, the new owner apparently deals with 
highly integrated stuff... such as the MediaGX.
CH> The 6x86 does not use the flags in the same manner as
CH> earlier Cyrix CPU's. Most coders use CPUID to 'check that
CH> the CPU id pentium class' - catch 22 situation. Granted, one
CH> could test for a common pentium+ opcode such as rdmsr or
CH> something and work on that basis.
You just closed the circle yourself... If you can test for RDMSR, you can 
test for CPUID in the exact same manner.
Hmmm... does the 6x86 support RDMSR/WRMSR/RDTSC?  I thought it didn't.  The 
time stamp counter is a VERY nice feature, since it allows exact calculation 
of the CPU speed, without needing a table lookup for every CPU under the sun.
CH> Take it on my experience: there is no uttely safe way to
CH> reliably detect the cyrix 6x86 without accesing ports 22h
CH> and 23h.
I think I've found a way... I'll post code in a day or two, possibly later 
tonight.
--Scott.
--- timEd 1.01
---------------