Re: The Apple ][ was cool!

[pausch@saaf.se (Paul Schlyter)]

In article <3B55FA1B.5EBED348@nogui.se>, Christer Palm  <palm@nogui.se> wrote:
 
> william strutts wrote:
> 
>> "Christer Palm" <palm@nogui.se> wrote in message
>>
>>> The bit-ness is popularly defined by the largest data operand that the
>>> processor can internally operate on with "base" efficiency, or basically
>>> the width of the processors internal data bus. I.e. 8 bits for the 6809,
>>> 16 bits for the 8086 - 80286 and 68000, 32 bits for the 80386-.
>>>
>> 
>> Then the 68000 is a 32 bit computer in  your words because it
>> is internally 32 bits.
> 
> Admittedly, 68000 and 68010 are indeed fairly close to beeing 32 bit.
> However, 32-bit operations take longer than 8 or 16-bit operations, and
> thus doesn't execute with "base efficiency".
 
In addition, some arithmetic operations are available only in their
16-bit versions.  For instance multiply (16x16 bits generating a
32-bit product) and divide (32-by-16 bit divide, where the quotient
must fit within 16 bits or else an overflow exception is generated).
 
Compare this to the 80386, which has 32x32 bit multiply generating a
64-bit product, and a 64-by-32 bit divide where the quotient must fit
within 32 bits.
 
Finally, if adding an FPU to the CPU (8087 to the
8086/8088/80186/80188, 80287 to the 80286, 80387 to the 80386,
built-in on the 80486 and later) we'll get an ALU for f.p.numbers
capable of handling 80-bit quantities at "base efficiency".  So we
could argue that if an 8087 is inserted into the original IBM PC with
an 8088, it becomes an 80-bit computer!!!  <g>
 
-- 
----------------------------------------------------------------
Paul Schlyter,  Swedish Amateur Astronomer's Society (SAAF)
Grev Turegatan 40,  S-114 38 Stockholm,  SWEDEN
e-mail:  pausch at saaf dot se   or    paul.schlyter at ausys dot se
WWW:     http://hotel04.ausys.se/pausch    http://welcome.to/pausch