Chroma CPU

[Phillip L. Harbison]

Gerry wrote:
> Christian Hofmann wrote:
> > [...] I wonder why there's that rumor of an Intel 80186 being
> > the Chroma's main CPU.
>
> In case anyone was interested, this was Intel's First chip
> Not the 8088.

You're thinking of the 8086, not the 80186.

> IBM Thought that addressing 1 meg of memory was insane, so they
> asked Intel to make a smaller/less powerful CPU, hence the 8088.

Not quite.  IBM didn't want to have a 16-bit memory subsystem, so
they chose the 8088 which has an 8-bit data bus.  Both the 8086
and 8088 have a 20-bit address bus (both can address 1MB).

> Thats why we have the 640k barrier!

We have the 640K barrier because IBM reserved the upper 360K to map
their video cards and BIOS ROM.

Duane wrote:
> The 80186 was originally intended to replace the 8088 in 1983, but,
> it was discovered that the '186 worked better as a microcontroller,

The 80186 had some onboard peripherals (e.g. DMA controllers) that
made it a good choice for an embedded controller.  We used them in
I/O controllers.  The 80186 did not extend the addressing range of
the 8086/88, which is probably why it didn't penetrate the PC market.
The 80286 was right on its heels and added 24-bit addressing and
protected memory.  It was chosen as the CPU for the PC/AT (second
generation IBM PC).

> > Thats why we have the 640k barrier!
>
> Actually, this limitation is in MS-Dos. And I quote:
> "640k should be enough memory for anyone." - Bill Gates, 1979

Gates may have said that, but the line between the lower 640K and
the upper (reserved) 360K was drawn by IBM.  MS may have had some
input, though.  Lord knows I would love to blame everything bad
in the world on Microsoft... death, high taxes, inflation, that
pain I get in my stomach after eating too much salsa, but this is
one time the Redmond Menace is not responsible.  On the other hand,
you can blame Microsoft for allowing the user direct access to the
hardware and propagating that decision long after the PC switched
to CPU's with protected memory.

batzman at all-electric.com wrote:
> [...]
> but motorola announced their intentions for a 16 bit processor
> (68K) and intel quickly decided to shift focus mid stream. Thus
> the 64K window and memory segmenting. Often euphemistically
> referred to as "memory Management."

I believe this was just a tradeoff between address range and the
average instruction size.  In the 68K instruction set, addresses
add 4 bytes to an instruction (but you can address 4GB).  With
the 8086/88, addresses add 2 bytes but you can only address 64K
without changing a segment register.  You could use the 68000's
base+offset address mode to accomplish the same thing, but most
68000 code I've seen just uses 32-bit addresses.  At least with
the 68000 you get to choose whether to use "memory management".

--
*name: Phil Harbison      --  My home is a Microsoft-Free Zone(tm)
*path: alvitar at xavax.com  --  with no intel Inside!