This article is more than 1 year old
Pentium III not long for this world
Carry on Gravy Training
Roadmap The roadmap we saw towards the end of last week demonstrates Intel's clear intent to eventually displace Pentium III processors with Pentium 4 and Foster technology. And the sooner the better, as far as the firm is concerned.
But there are some clear differences from the past.
Unlike the good old days when the Intel gravy train seemed to chug on to its next destination making stops to drop off old chips at various retirement homes for senior x86 citizens, this latest roadmap shows clear signs that the proprietary arrogance of yore is showing signs of wear and tear.
The roadmap, which was shown to PC customers and distributors in November, starts by saying that there will be an accelerated ramp rate for the P4, and that Intel has a "90 per cent high confidence" in its revised schedules.
The ramp of the Pentium 4 will "cover all mainstream" segments by the end of next year, and Intel expects a crossover sales in early 2002.
This "crossover" point is an important one for Intel, although it doesn't always make its targets. The obstacles to it managing such a crossover in early 2002 include competition, market forces, a move to larger, 12in wafer sizes, chipsets and memory availability, and also a move from .18 micron to .13 micron.
All of these are bigger variables than Intel's "90 per cent high confidence" might suggest. Throw words like AMD, huge capital investment, recession, slump in demand for PCs, ability to execute on chipset strategy and, yes, Rambus, into that cooking pot and we have to wonder about that 90 per cent figure.
Intel segments its desktop map into the performance, mainstream 3, mainstream 2, mainstream 1 and three value segments.
In the performance sector (systems without monitors costing over $2,000), it is predicting the 1.5GHz Pentium 4 will supersede the 1GHz Pentium 3. By Q2 2001, that slot will be occupied by the 1.70GHz P4, by Q3 (Hi, Doctor Yu, do you know the date yet?) the 2GHz Pentium 4, while greater than 2GHz Pentium 4s with Willamette cores will arrive towards the end of next year and the beginning of 2002.
By that time, Intel hopes it will see the virtual end of the Pentium III CuMine family, although in between times we will see the 1.13GHz Pentium III (cough) in Q2 2001, while we'll also see a 1.26GHz Pentium III briefly tip up.
Intel says in its latest roadmap that it will cut Pentium 4 prices to "reflect an accelerated product ramp", and adjust the price of its 1GHz Pentium III processor downwards as its .18 micron capacity increases.
Further, it will Carry on Gravy Training by following its usual trend of dropping prices and pensioning off its former heroes. (Intel uses the quaint term off-roadmap for this process).
Off-roadmap may not be as off-roadmap as many of us might suspect. For example, earlier in 2000, unexpected quantities of Pentium IIs made their appearance when other shortages were biting.
Prices for performance Pentium III desktops on the 28th of January 2001 will be as follows. The 750MHz will drop to $173, the 733MHz to $163, the 700MHz to $153, the 666MHz to $143 and the 650MHz to $143. (Prices for 1000s)
Parity in prices means that Intel is pensioning off chips. Sadly, the Intel CuMine 666MHz processor has had a short shelf life indeed. (Intel prefers for obvious reasons to call this a 667MHz chip).
So the Pentium III is nearing desuetude and long live the Pentium 4. But not until 2002 folks.
Tomorrow, we will look at how Intel will wrest victory from the jaws of defeat with a consolidated chipset strategy, as well as look at its Celeron strategy and its cunning plans for mobile processors.
Its mobile processor strategy still seems to be firmly on track - whither Trancemeta and Mobile Durons? ®