🚀 go-pugleaf

In <139@complex.complex.is> frisk@complex.is (Fridrik Skulason) writes:

>spagiola@frinext.stanford.edu (Stefano Pagiola) writes:

>>: Will NeXTSTEP 486 run on 386 machines?
>>: No. NeXTSTEP 486 uses several 486 specific features that enhance
>>: the performance of NeXTSTEP. NeXTSTEP 486 will support any true 
>>: 486.
>>: 
>>: Will NeXTSTEP 486 run on the Cyrix 486SLC?
>>: No. The Cyrix chip not a true 486.

>>IMHO, if there are in fact processor-specific advantages to requiring  
>>a 486, doing so makes sense.

>Yeah, but I don't see any sensible reason for not making 386+387 the
>minimum requirement - the differences between a 486 and a system with 386+387
>are not exactly major, from a programming point of view.

All this means is that you haven't looked hard enough.  For a system
requiring a lot of power and making heavy use of the math processor,
it makes perfect sense to restrict it to the i486.  This lets you
specially optimize for the i486, use the (few) extra i486
instructions, optimize pipelining for the known timings of the
instructions, etc.  If running on a 386+387 results in such poor
performance as to be virtually unusable, why not simply target the
hardware that the user is going to need anyway?

>No, I would guess that something else is going one here.  This actually
>reminded me of something - what seems a long time ago, some programs checked
>the PC ROM, and refused to run, unless they found an IBM copyright message
>there.  Of course, other BIOS manufacturers found a way around this - one
>BIOS even contains the text "Some software expects to find Copyright IBM here"
>with "Copyright IBM located in exactly the right position"

>Well, the only benefit I can see for making the software 486-specific is that
>it would help selling '486 processors.  And who would benefit .... right,
>Intel.

So Intel owns NeXT (which produces its own Motorola-based machines)?
I don't think so.  There are times when a conspiracy theory makes
sense and times when one doesn't.  This is one of the latter.

>Letting the program to run on only true 486's does not help the end-user much,
>but it is sure to hurt Cyrix and AMD....time to sell their stock :-) I guess.

If AMD produces a close enough clone, it might run NeXTStep.  It's
going to have to be REAL close, though, I would bet.

>Now, I am not suggesting that Intel may actually have paid for having them
>make the software '486 specific...it wouldn't be nice to say that, but........

It also probably wouldn't be legal (to say OR to do).

-- 
"Insisting on perfect safety is for people who don't have the balls to live
 in the real world."   -- Mary Shafer, NASA Ames Dryden
------------------------------------------------------------------------------
Fred.McCall@dseg.ti.com - I don't speak for others and they don't speak for me.