[rescue] Some days ....

[Francisco Javier Mesa-Martinez]

> Amen, brother.  IBM really copped out by just daisychaining two
> 8-channel IRQ chips in the AT architecture.  They should have built in a
> single 32-channel (or better yet, 64-channel) chip, IMHO.

They were limited by the original 8088... for what they had to work with,
and the time constrains they were working under, it is not that bad of a
design. I was not intended to become the architecture standard for over 2
decades by any means, just like the 8088 was not intended to be a real
CPU!

Funny thing: PC was designed to be as a safety meassure to get something
out of the door ASAP before more personal computers would make it to the
market. They thought that a real IBM design would be released at a later
date and the PC would at least provide some sales in the mean time, we all
know what happened next.

In the same way the 8088 was a safety design inside intel, they just had a
dozen of designers with very limited resources to produce a "safety"
extention to the 8080. Intel was betting on the i432 those days, so they
believed that x80 was dead... but just in case have a chip ready (hence
the brain dead memory segmentation, reduced interrupts, plenty of bugs
-A20 line anyone!-). Again, the chip was not intended to become the
standard by any means.... and voila, here we are.

So the moral of the story is, design a pretty good product and also design
a crappy safety option system just in case. I assume your crappy design
will provide the bulk of your company's revenue for the next decades :).