On 8 Jan 99, at 2:29, Brent Reynolds wrote:

> After all, a DX4/100 was not a 100MHz clock-speed internal CPU, it
> was 99MHz

  This might be true.  But what if the 33/66 speed is actually
intended to provide 8.3MHz to the ISA bus?  Then we get:

8.3 x

   4  =  33.2  "33"
   8  =  66.4  "66"
  12  =  99.6  "100"  --> your DX4 example
  16  = 132.8  "133"
  20  = 166.0  "166"

and so on.  That is, you may be reporting an error based only on Intel
rounding while you truncate.  I believe that a more precise value can
(and probably does) fall closer to Intel's "100" than to your "99".

  [My point, I guess, is that the inexactitude of Intel's "official"
speeds isn't inconsistent with simply rounding to the nearest integer,
so there's no need to ascribe it to incompetence or malice.  Not
without some actual Intel marketing documents as evidence, anyway.]

> and it was not four times bus speed, but three times--clock trippled
> rather than quadrupled.

  Intel never claimed the DX4 used a 4x multiplier.  They had to
demonstrate that the number was a model identification and not simply
an attribute, or risk losing their trademark, and their solution led un-
savvy customers to assume that the number still reflected the
multiplier.  [Whether *this* was malice or incompetence really doesn't
matter any more....]
  The "Pentium II" and new "Pentium III" names aren't vulnerable in the
same way that DX2 and DX3 would have been.  Apparently Intel learned
some things over the years.


David G

                Curious about the people moderating your
                   messages? Visit our staff web site:
                     http://nospin.com/pc/staff.html