🚀 go-pugleaf

zeke@fasttech.com (Bohdan Tashchuk) writes:

>>>NeXTStep is said to not run on the Cyrix 486 chips because "they are
>>>not real 486's".  No one has come up with a reliable reason _why_ NeXT
>>>doesn't think they are "real 486's" or why NeXTStep won't run on them.

>>I don't have any real data, but I expect the problem is that the Cyrix
>>chip doesn't have a floating point unit.  Next's own systems have all
>>had standard FPUs, and reviews of NeXTStep say that it won't run on a
>>486SX, only a DX.

>A 10/21/92 semi-official net posting on NextStep claims it runs on SXs:

>   General Requirements:
>   CPU - 486 based PC Compatible Computer. This includes 486SX, 486DX,
>   486DX/2. Future Intel Microprocessor designs in the x86 family will
>   also be supported. ISA or EISA expansion bus. 486DX and 486DX/2's are
>   recommended for better performance.

>   Will NeXTSTEP 486 run on the Cyrix 486SLC?
>   No. The Cyrix chip not a true 486.

>I asked for an explanation of "true 486" in comp.sys.intel. A Next person
>seemed to imply that Next didn't want to bother resolving the supervisor
>state differences between the chips. Nothing very convincing. I had a theory
>involving "arrogant propeller-heads", but the Next person didn't agree with it.

>My guess is that NextStep is sluggish enough that Next saw no reason to bother
>with the Cyrix parts, which are lower performance than Intel 486?Xs.

>I've also seen Intel apologists and employees repeat the "not a 486" claim in
>various groups. They've succeeded admirably in their primary goal, which is
>to sow Fear, Uncertainty, and Doubt as widely as they can. What I haven't seen
>yet is an explanation. I'll ask the question again here:

>	Please tell me exactly WHY the Cyrix isn't a 486? What is it that
>	Intel feels is the essence of a 486?

>	Let me say a priori that a response of "made or licensed by Intel"
>	isn't sufficient in and of itself.

The jist I got was that NeXTstep wants to maintain compatible File Systems
between 680X0 and X86 platforms.  To do this, they needed to do a lot of
byte swapping on the X86 platforms.  Since the 486 has a simple instruction
to do it while the 386 does not, NeXT just went with the 486.  I think their
justification was that you needed the horsepower of the 486 to run NeXTstep
anyway - atleast when you only look at X86 processors from Intel!  This
assumption (of only considering Intel's product line) may, indeed, be the
real culprit as I think the Cyrix ne-486 is quite capable of running NeXTstep
from a performance perspective.

Note:  The above impressions come from a Bay Area Next Users Group (BANG)
       meeting I attended several months ago where the manager in charge
       of NeXTstep "486" gave a rather detailed talk on the product.  I'm
       sorry I can't remember his name - but I do know he drives a 928 (or
       something similarly fast) with a licence plate reading "NEXT 486".
       He joked all night about choosing the 486 (and later the "586")
       because it is FFAASSTT - like his car, get it?  BTW: I took this
       "486" decision to be yet one more indicator that NeXT does not
       target the mass market.  They seem to have completely yielded the
       so-called "low end" to Microsoft, et. al. and are focusing on
       finding that niche between the Desktop PC and Technical Workstations.
       While they are certainly having some short-term success in this
       market, I think they are doomed!  I'm reminded of a line from "Other
       People's Money" in which DeVito is explaining the folly of a strategy
       of "gaining market share in a shrinking market".  The niche NeXT has
       staked out sure looks to me like one of those "shrinking markets".

Ed "doom and gloom" McClanahan