🚀 go-pugleaf

In article <3805@ames.arpa>, mike@ames.arpa (Mike Smithwick) writes:
> In article <3279@umn-cs.cs.umn.edu> davidli@umn-cs.UUCP (Dave Meile) writes:
> > But I still insist that, for the non "power user" there's no need for it.
	[ several examples of short-sighted thinking ommited for brevity]
>           "Well, when we were designing this thing, Jay asked 
> 	   'So, how much memory do you think people will put
> 	   on this thing?' We all looked at each other, and said
> 	   that NO ONE would ever have any need for more than 512K.
> 	   So he gave the graphic chips 19 pin addressing."
> 				Dale Luck (I think) explaining the
> 				origins of Chip memory.

Pinouts on chips are not exactly free; they have a definite
manufacturing cost and per-connection reliability implications.
Engineers spend quite a bit of time juggling this scarce and valuable
resource (the pins) amongst all the possible things that they might
wish to ship in and out of the package.  Hence all those bl**dy
multiplexed address/data schemes on cpu's (almost anything from Intel
last time I looked), multiplexed addresses on DRAMS, kluges to
redefine pins according to system configuration (Intel again, at
least on the 808[68] parts), and ghods know how many other egregious
hacks over the years.  I mean, most engineers understand the KISS*
princple, and they don't design these horrible interfaces just for
the joy of complexity [well, most of them, anyway; I have dark
suspicions...].

So let's not be _too_ critical of Jay - your other examples were
much more convincing.  My pet peeve is that the Atari people didn't
put a base-and-limit register into the big glue chip that controls
the memory array(s) on the ST.  If I had base and limit registers, I
could get true, transparent multitasking up on the ST (which, of
course, Dave Meile could ignore; that's his privilege).

--
Ross Alexander,
Sr. Systems Programmer & bottlewasher @ Athabasca University,
alberta!auvax!rwa

* KISS == Keep It Simple, Stupid!