Having worked for Intel in the way back.  There was another reason for
putting out 486SX's, increased yield.   If there was an error encountered
during post production testing in the FP unit, disable it, voila 486SX and
sell it.

they're out to make money ya know,

On Tue, 13 Feb 2001, John Nemeth wrote:

> On May 25,  5:58pm, "Thomas Mueller" wrote:
> } 
> } Correct with one exception.  Original 486 CPU, subsequently renamed 486 DX, 
> } included floating point instructions, hence no need to emulate.  Later, for some
> } cockeyed reason, Intel put out a 486 SX, which was a 486 with floating-point 
> } part disabled.  Then a 487 processor could be added to remedy this deficiency.
> } So a 486 SX would also require MATH_EMULATE.  I can't imagine where one would
> 
>      Intel put out the 486SX for marketing reasons.  The early 486's
> ran at low clock speeds (16 - 20 MHZ) and were relatively expensive.
> Several companies, such as AMD and Cyrix were putting out 386 clones
> running at up to 40MHz for a lot lower price and whooping Intel's
> butt.  They responded by coming out with a low priced 486 without a
> math-co, namely the 486SX.  BTW, the 487 was really just a 486DX with
> an extra pin that when inserted into a 487 co-pro socket would tell the
> motherboard to disabled the original processor, which was most likely
> soldered on.  This strange situation didn't last long and was quickly
> replaced by ZIF sockets.