Brian - Re:"a comment from an analyst saying that K6 would
be fine as a simple alternative to Pentium ..."
The point this analyst was trying to make is that the K6 plugs in to a Pentium socket, therefore it is a direct Pentium replacement.
The K6, with it's 64K L1 cache, MMX, and advanced architecture should outperform the Pentium with no sweat.
However, Intel's Pentium Pro, and upcoming Klamath, have a different socket/pin out with a very complex bus structure, giving it many advantages over the Pentium bus - split transaction processing, multiple bus transactions, direct access to L2 cache, GTL+ logic levels, etc.
The K6 will not be playing in this ballgame.
The hinge factor (there's always a hinge factor) is how fast (and if) Intel can switch its customers (and eventually the consumers) to adopt the new Klamath chip as the preferred chip of choice.
Intel is now incorporating in their new Pentium Pro/Klamath chip set designs AGP support to allow much faster graphics, higher external clock speeds, etc. Thus, Intel is lining up all the support for a massive Pentium-to-Klamath conversion.
By the way, Intel will be discussing a 300 MHz "Klamath-type" chip at the ISSCC in February, so they are close to offering a 0.25 micron version of this chip.
One eventuality is that the K6 does score some impressive design wins for Pentium-type applications. They will sell a lot of these. Then, as (if) the market shifts, the Klamath will become the machine of choice and Intel will regain a near monopoly position, and AMD will see declining K6 sales.
How soon can all this happen? Who knows - we all have crystal balls that are less than fool proof.
The salient point, however, is that Intel has a "bridge to the 21st century and the K6 is a bridge to the past (Pentium)". [Thank you Bill Clinton & your speech writers].
Paul |
