On 26 Aug 98 at 13:52, Jim Meagher wrote:

> Does this remind anyone of the DX/SX days?  And more importantly
> -- do you remember how short the "stripped down" SX market lasted?

  *WHICH* DX/SX days?

  In about 1981, Intel took the 8086 16-bit core, and multiplexed it
down onto an 8-bit bus to make the 8088.  IBM "happened" to pick that
chip for its new machine, and orders of magnitude more 8088s were
ever sold than 8086s.
  Around 1988, Intel took the 80386 32-bit core (which it now
christened "DX") and multiplexed it down onto the same bus as the
80286 to make the 386SX.  Since lots of machines had less than 16MB
of RAM, and the most popular OSes were still 16-bit, lots of
low-power/low-cost machines shipped.  Maybe not more than 386DXes,
though.
  Around 1993, Intel took the FPU out of the 80486 package (which it
now christened "DX") to make the 486SX.  Ironically, this chip was
ideal for servers (file systems and communications don't use much
floating-point math), but was mostly sold in low-cost desktop
systems; some provided the option to add an FPU later, but some
didn't.  I'm sure they sold a fair number, but none of the
clock-multiplied versions was ever offered without an FPU.

  AMD, Cyrix, and so on have been content to ride on a fairly simple
"technology curve", where this year's high-end workstation technology
becomes next year's secretarial or home PC, as new and better
technology is introduced.
  Intel believes in market "segments"; that there is money to be made
in the home/mass market by killing off old high-end products
and introducing new products that are able to benefit from current
marketing efforts, but provide less-than-current performance.  Their
track record with this approach is actually pretty successful.  [The
most spectacular failure of this approach that I can think of was the
PCjr, but that was IBM's marketing and not Intel's.]

  Is this a "bad" approach?  I don't like the fact that it pushes
first-time computer buyers off of the technology curve into dead-end
machines that generally require replacement rather than upgrade when
the owner discovers they need something better.  If you are reading
this list, and you don't already have friends or family who wish you
had warned them about this, I predict that you will.  Even though, at
the time, they seemed to be saving some money.
  The point is, Intel has worked this successfully in the past, and
probably will in the future as well.  It's not *quite* the P.T.
Barnum Business Plan ("Suckers are created faster than you can fleece
them"), but it's more about making money than about making better
computers -- whatever the ads may claim.

David G