Droid issues - Legacy Bash IFS var clobbering VLIW offset?

[Lennart Sorensen]

On Fri, Nov 25, 2011 at 04:24:35PM -0500, Walter Dnes wrote:
>   What Intel *REALLY* did wrong...
> 
> * The original 8080 was a 16-bit processor that addressed 65,536 bytes
> 
> * The 8086 (and the similar 8088) was the next version.  To increase the
>   adressable space, they used a 16-bit base register and a 16-bit offset
>   register.  The real stupidity was that the address was calculated as...
>
>   (16 * base_register) + offset register

Wasn't the 8085 in between?

>   See http://en.wikipedia.org/wiki/Intel_8086#Segmentation  This meant
>   that 2 16-bit registers combined to only 20 bits of address space
>   (i.e. 1 megabyte) in real mode DOS.  Throwing away 12 bits meant that
>   there were 4096 ways to express the same address.  It's a crying shame
>   that Intel didn't use a base+offset scheme that went something like...
> 
>   (4096 * base_register) + offset register
> 
>   That would've given 28 bits of address space, (i.e. 256 megabytes) in
>   real mode DOS.  The history of DOS would've been totally different,
>   without EMS and HIMEM garbage.

You wanted intel to plan ahead?  Why would they do that? :)

Certainly expanding from 16 to 20 to 24 to 32 bits of address space with
each generation was pretty stupid.  How long until you start to see a
trend there and do it right.

It's not like they had to add all the external address pins right away
if they had gone to say 24 (or 28) bit right from 16.

Of course since they used -16 as the reset address, they would have had
to do something with the external address lines to allow the rom to be
mapped to the end of address space though.  That could have been done
without having to add all the address pins right away.

-- 
Len Sorensen
--
The Toronto Linux Users Group.      Meetings: http://gtalug.org/
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://gtalug.org/wiki/Mailing_lists