The release of Intel's 1.2GHz Celeron has come and gone, but this new processor is still something of a mystery. Not only did the chip continue the proud "stealth release" tradition of past Celeron processors, but it attracted a decidedly low-key level of press reviews and benchmark tests. Before delving too far into the possible reasons for this, let's take a close look at the CPU itself.

Basically, the 1.2GHz Celeron uses the same core as the Pentium III "Tualatin." This is a natural progression, since the desktop P-III was read its last rites long ago and Intel needs something to fill its shoes. This means the Celeron/1.2 core shares the same 0.13-micron process and sports a full 256K of Level 2 cache. This last point is an important one, since previous Celerons have been differentiated by their more modest 128K of L2 memory, which helped position them against the higher-priced Pentium III. With the latter out of the way, the Celeron is rising to its full architectural height -- or is it?

The hitch is that the Pentium III was a very powerful CPU design already, and shrinking its die via the 0.13-micron process opens the potential of much higher core speeds. This led to some embarrassing comparisons to the Pentium 4, as the Tualatin posted excellent performance scores and actually surpassed the P4 on a clock-for-clock basis. It seems to be just fine for Intel's flagship Pentium 4 to lose the instructions-per-clock race to AMD's Athlon, but not to a horse from its own stable.

That leads to the question of product positioning and exactly where the new Celeron fits in. It's quite obvious that Intel is looking to the Pentium 4 as its CPU of choice (especially for the all-important corporate market), so the last thing the company needs is an Athlon-like competitor in its own camp.

Put a Governor On It

So how does Intel sidestep this potential landmine? The most obvious methods are by limiting the Celeron's core and bus speeds. The original Pentium III Tualatin used the 133MHz bus, which allowed it to eke as much performance as possible out of the aging Socket 370 platform. The Celeron/1.2 is hamstrung by a 100MHz bus, which not only limits its performance for basic SDRAM implementations, but makes certain that moving to a non-Intel DDR chipset isn't even an option -- even at 133MHz, DDR offers very limited performance advantages, so don't even go there at 100MHz.

Similarly, capping the Celeron (at least until January) at 1.2GHz -- the same top speed as the Tualatin, now assigned to rack server duty -- makes sure that neither the Celeron nor P-III matches the lowest-speed 1.3GHz Pentium 4, making direct comparisons more difficult. Just by using basic math (or overclocking results) to project the performance of the new Celeron core, a 1.5GHz or higher model could certainly give midrange Pentium 4s a run for their money. It seems Intel has a real hot potato with the 1.2GHz Celeron, and a few things have me wondering whether the company wants the chip to succeed or to fail.

Online availability of the Celeron/1.2 is still not high, as a recent foray into the price search engines yielded few hits. In fact, even though the chip has been officially available for over a month, PriceWatch's CPU section doesn't have a specific entry for it. Contrast this with the 2.0GHz Pentium 4, which enjoyed widespread vendor adoption on the day of its release, as well as being stocked by virtually every chip dealer in the business.

Vendor support and promotion of the new Celeron has also been very limited. Compaq does offer a few Presario models, though I couldn't find a single one amid the scores of 1.0GHz and 1.1GHz Celeron PCs on Dell's and HP's Web sites. Even value leader E-Machines is sticking with the 1.0GHz Celeron as its entry-level option.

There are some valid reasons for the low number of vendor systems and silicon dealer lists sporting the Celeron/1.2. One of the primary deterrents may be the Socket 370 platform itself. This is an aging architecture to begin with, but to add insult to injury, the Celeron/1.2 will not work with the majority of existing motherboards.

This leads to virtually no demand for the new Celeron as an upgrade CPU (unless you want to buy a special PowerLeap adapter along with it) and may cause many chip vendors to give it the cold shoulder. The new core also requires system vendors to use a different motherboard than their existing Celeron platforms, starting from square one with integration and compatibility testing. This can certainly take time, as the Celeron/1.2 is not just a CPU-form-factor shift like the recent microPGA 478-pin Pentium 4 design, but an entirely new core requiring an enhanced chipset. This means updated drivers, new INF files, and potential issues with the new chipset not being recognized by the operating system or other software.

Building Brand Equity?

On the flip side, there is one very compelling reason for Intel to introduce a new Celeron, but still attempt to limit its overall market impact. To put it bluntly, the Celeron/1.2 looks great on paper and helps sell the lower-speed Celerons. The vast majority of PC buyers don't shop at the top of the CPU food chain, so Intel can proudly proclaim that it has the fastest value processor at 1.2GHz, while selling a ton of 1.0GHz and 1.1GHz systems (with 128K of L2 cache and no danger of upstaging the Pentium 4).

The incredibly low level of promotional support the new Celeron has received from Intel supports this idea. In fact, I've only read a couple of reviews of the processor, which pales in comparison to the press attention the Pentium 4/2.0 received. I suspect Intel is happy to see reviews of low-end, integrated PCs with Celeron chips, but less eager to have some hot-rodding Webmaster stick the CPU into a high-end motherboard with GeForce3 video card and trumpet the results.

It's a dilemma for Intel: The company has invested a great deal in the Pentium 4 architecture, and having any sort of Celeron competition doesn't fit its business plan. On the other hand, AMD continues to push the Duron to higher clock speeds, so a faster Celeron is essential to maintain economy-class market share, even as it's a potential threat to the Pentium 4.

At least until the Pentium 4-based Celeron is ready for prime time, Intel needs a value contender, but one that knows its place at the bottom of the food chain. The most interesting trend to watch will be exactly how high Intel takes the Celeron core speed, especially as AMD moves the Duron to an 0.13-micron process as well. This will be a push-pull battle for the ages. I'd expect Intel to go P4-only for an interim period rather than take the Celeron to 1.5GHz or higher, but stranger things have certainly happened.