Re: Macintosh LC...need info on this =) TIA

[pausch@saaf.se (Paul Schlyter)]

In article <3A2C7A18.915EBCE6@inetnebr.com>,
Roy and/or Janet Miller  <millers@inetnebr.com> wrote:
 
> Paul Schlyter wrote:
> 
>> And then we had the 68008 with an 8-bit externam data bus, which thus
>> was an 8/32 bit chip.
> 
> True.
> 
>> And if you added an 8087 to your 8088, you got 80-bit internal registers
>> for your floating-point computations, making your 8088+8087 an 8/80-bit
>> chip combo!
> 
> LOL, ok, but usually the convention of the day referred to the CPU, not the
> math co.
 
But what about today, when the floating-point processor no longer is a
co-processor but is integrated with the CPU?
 
Do you know how the Pentium does an integer multiply?  Like this:
 
1. Convert the integers to floating-point
2. Do a flating-point multiply
3. Convert the result back to integer
 
The rationale behind this was that floating-point multiply on the
Pentium is so fast (1 clock cycle) that they could save some silicon
doing this.  But this will give the weird effect of an integer
multiply being slower than a floating-point multiply!  Yet the
integer multiply will require fewer cycles on the Pentium than on the
486 and earlier.  And this approach would hardly have been feasible if
the FPU had been a coprocessor.
 
-- 
----------------------------------------------------------------
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