PC Magazine back in the early 90's had something
that mentioned this. If I
remember correctly it was one of the Dvorak columns. He mentioned that the
486SX was basically a marketing ploy by Intel to allow them to get rid of
486DX chips with defective Math CoPro units.
More or less there have been also 486SX with full working but
disabled FPUs, since the 486SX was also used to deploy a lower
price unit without touching the 'real' 486 price.
As for a Math CoPro for the 486, I'm not sure I
ever saw a 487 chip, but I
always figured that they took the chips that didn't cut it as a normal
processor but had a good Math CoPro, and sold them as 487's.
No, the 487 was just an 486DX with no modifications to the
processor it self - just (AFAIR) one of the former unused
pins now delivered a signal to disable the 486SX. So systems
with 486SX and 487 just had two complete CPUs (And double
power consumption :).
I've no idea if this is true, but it made sense to
me, since why through
out a chip with a good processor, with you can just package it as a 486SX,
and sell it at reduced cost. Sounds to me like everyone won. After all,
how many people really felt the need for a Math CoProcessor in the early
90's?
Gee - I had a 8087 in my first XT - I never used it beside some
tests, but i HAD one - super power computer ! Yes, I wrote some
programms to play with the ability to have two concurent processors
running, but no real world app. And I still don't know for what a
FPU is usefull - Anything can still be done in integer - you just
have to think (sometimes a bit harder) about what you are about to
do.
Gruss
Hans
--
Ich denke, also bin ich, also gut
HRK