On 14 Aug 99, at 13:19, Drew Dunn wrote:

> I'm tempted to say that the production costs are higher for a PII than for a
> Celeron, but I'm not actually sure that is the case.

  Well, it's true.  The cache that is provided in Celeron's is on the same
"chip" die as the CPU.  The PII/III are able to provide larger caches, but
must run them at lower speeds, BECAUSE the cache is in the form of two
additional silicon dies to be mounted onto the package -- and that translates
into both more silicon area (fewer complete CPU packages per silicon wafer)
and more mounting-and-connecting work.  Both of these manifest as higher
production costs.

> There's nothing "bad" about a Celeron chip.  Intel has positioned that chip
> as an entry-level processor, but its performance is still extremely good.

  The "normal" transition in this industry has been for new CPUs to be the
"latest and greatest", and for them to become affordable over the course of
a year or two as chips with higher speeds and better features get introduced.
  The Celeron got some bad press because it departs from this model -- it was
deliberately designed NOT to be a "latest and gretest", but to start life as
merely "adequate and cheap".  [Ironically, the name "Celeron", deriving from
the same root as acCELERation. deCELERation and CELERity (speed), supports
mass-marketing portrayal of this as a "latest and greatest" chip....]
  [The underlying theory is called "market segmentation".  The price of
Pentium Pro chips remained high because Intel successfully positioned them as
"server" chips but managed to keep them out of the mass/consumer/home market.]

  This is the FOURTH time that Intel has pulled this manouver within the x86
product line.  The first time, they repackaged the 8086 as the 8088 to work
with lower-cost 8-bit peripherals; IBM used that as the basis for their PC,
and it was wildly successful.  The second time, they packaged the 8088 and
some support circuitry as the 80188 to enable manufacturers to make low-cost
XT clones even as the 80286 was being introduced; IBM planned to use the
80188 in the PCjr, but in their PC/XT design they had violated some of
Intel's rules and so no 80188 could ever be XT-compatible.....  The third
time, they removed the FPU from the 80486(DX) to make the 486SX; this led
directly to the rise of Packard Bell, selling cheap 486SX-based machines for
(primarily) home use.  [Ironically, the FPU was the part of the CPU least
likely to be needed by network servers, and this broke a number of programs
that were written "knowing" that all 486 systems had an FPU built in....]

  The Celeron wasn't designed or positioned as a "latest and greatest" chip,
but its actual performance --especially when overclocked -- is quite
competitive to the PII and PIII lines, at a fraction of the price.  Of
course, Intel doesn't want to market it that way, because they need the
premium prices on PII and PIII to sustain the profits their shareholders have
come to expect.
  The initial readings, on the Celeron, then, were that it was under-powered
(by design!) and over-hyped.  [The reverse may now be true....]  That has left
them with continuing PR problems such as the comment you were responding to
here.



David G

                  Visit our website regularly for FAQs,
               articles, how-to's, tech tips and much more
                  http://nospin.com - http://nospin.org