Date   : Wed, 17 Feb 2010 00:31:10 +0000
From   : afra@... (Phill Harvey-Smith)
Subject: OT PC architecture

Rick Murray wrote:
> ??!? Interrupt latency like that?!? IIRC the earlier x86 chips were 
> horrible, not to mention the segmented memory system that, while it had 
> a purpose, artificially crippled a lot of things 

While I won't disagree with this, I was brought up on Motorola 
processors 6809 (Dragon), 68008 (Sinclair QL).

>                                           that weren't prepared 
> to jump through the hack-on EMS/XMS loops. An example being Borland's
> TurboC that choked on source files over 64K.

This is not to do with XMS/EMS, but more to do with the segmented 
addressing having a segment size of 64K.

EMS/XMS was designed to enable the PC to have more than 640K of ram, and 
the main reason for that limitation has nothing to do with MSDOS, or the 
80x86, but more to do with how IBM broke up the memory map when they 
designed the original PC, and designated the region above 640K to be for 
addon cards / rom, after all when it was designed in 1981 640K was huge :)

As an illistration of this the RM Nimbus is also 80x86 based, also runs 
MSDOS but has up to 960K or ram available, because they put the ROM at 
the top of the memory map, leaving the rest able to be used by RAM.

>                                    I think TurboPascal does as 
> well. TurboC++ fixed that, fair enough, but it shows that it was a 
> problem. Oh, and memory models. What a pain in the ass. As a 
> raised-on-RISC-OS coder I don't give a **** what memory model it wants, 
> just give me an 80K array and don't flippin' argue or tell me a load of 
> rubbish about _far pointers. 

Whilst again I generally agree, I'm sure other processors support 
pointers of different sizes specifically I think the 68xxx series could 
use 16 and 32 bit, though admittedly it's pretty easy to convert them.

>                           It wasn't until the 80386 that it was 
> capable of running a useful version of Windows, and it wasn't until the 
> 80486 that it could run said useful version of Windows properly.

I dunno I seem to remember running Win95 on a 386 DX40 and it not being 
too bad by 1995 standards :)

> God I'm glad Acorn didn't go the x86 route!

I wil agree with that, it's a damn shame that IBM didn't pick something 
better for the PC too, like the 68K for example, but I believe that also 
had something (more expensive) based on that, which they did not want 
the PC to compete with.

Cheers.

Phill.

-- 
Phill Harvey-Smith, Programmer, Hardware hacker, and general eccentric !

"You can twist perceptions, but reality won't budge" -- Rush.