Intel Accelerates the Desktop

Back in December 1998, I did some research on how Intel’s strategies and roadmaps were changing rapidly and made a correlation between that and the market pressure AMD had brought to bear with the K6 (see 3DNow! Accelerates Intel Chips Too!). One bit of information in that article that might be surprising is that at the end of 1998, some estimates put AMD at about 18% market share in the microprocessor market (though others put it as low as 12%), with Intel at around 80%. Depending upon whose estimates you believe, AMD today has between 17% and 23% market share, while Intel has about 80%. AMD’s gains over the past three years, it can be argued (albeit weakly), actually came at the expense of Cyrix and IDT. Revised estimates of last quarter claim that both AMD and Intel gained very slightly in market share – at the expense of Transmeta and, you guessed it, VIA Cyrix. The more things change, the more they stay the same, it seems.

Over the past three years or so, I have been paying particular attention to the seemingly divergent paths that AMD and Intel have been on with regards to executing their respective plans and market position. While doing research for this article, I found an even older write up from March 1998 (see Is Intel in Trouble? ) that documents a marketplace that is hauntingly similar to the current one. With only a few changes of dates and product names, it could be describing the current marketplace! Admittedly, however, the current slump seems much worse.

With all the recent news about the changing Intel roadmaps, I thought it would be interesting to perform a similar investigation using the three roadmaps from 2001 that I have information about (Jan 23, June 1 and July 17). Today, Intel has really segmented the market by splitting it up into three major categories – desktop, mobile and workstation/server. Furthermore, over the past year, Intel has subdivided the mobile and workstation/server categories even further, as you will see in the charts. Interestingly, the desktop category has remained almost the same for the past three or more years.

In an attempt to make it easy to compare the roadmaps directly, I made a separate table for each market segment Intel uses, and put like processors from all dates next to each other. Each roadmap date has its own color throughout to help in identifying any changes. You can see whether any particular roadmap segment has changed recently by picking a particular column (representing each quarter) and following it down.

The first set of charts is for the workstation/server category. Note that Intel has broken this into multiple segments, including workstation, dual processor servers, front end servers and multiprocessor servers. In the front-end server segment, they have further broken it down into high-density (1U/2U) and ultra-dense (introduced in Q3 ’01). The ultra-dense segment consists solely of PIII Tualatin 512K processors, and has not changed since being included on the roadmap in June 1 of this year.

As you can see, there have been essentially no changes to any of these roadmaps since the beginning of the year, with the exception of some additional details on speeds further out. This would seem to indicate that Intel does not feel any pressure in this market, and that they have not encountered any significant issues that would prevent them from releasing the various parts on time.

The next set of charts is for the mobile processor category. Again, Intel has broken this into multiple segments, including full size, thin/light, mini and sub-notebooks. I have not included the mini and sub-notebook segments in these charts because they were only recently broken out and no changes have been made since they were introduced. Since AMD does not currently have any competitive offering for the very low-end notebook market, and probably won’t until next year, there isn’t much of interest to show there.

Of particular note in this category are the additional speed grades for the mobile Tualatin (PIII-T, on these charts), and the additional higher speed grades for the P4 later this year. This would seem to indicate that Intel is viewing the Athlon 4 as being a relatively strong competitor in these two segments, though not enough to increase the speeds at the top end.

The final chart is for the desktop processor category. The various segments Intel has been using here are based purely upon price points, unlike the previous two categories, thus the single chart. Here is where the roadmap has really been accelerated, which is not surprising considering the strong competition AMD brings to this part of the market.

There are several areas that are of particular interest. First, notice that the Tualatin lifespan on the desktop has been reduced to a single quarter. After that, Intel intends to ramp the P4 volume and drive it into the mainstream very quickly. You can see how quickly the low-end ramps in speed now. Note that both the Willamette and Northwood cores will be offered at 2GHz.

The real eye-popper here is the Celeron, however. Between January and June, there wasn’t much change, but in the last six weeks Intel has turned this into a heavy-duty battle zone. As you can see, Tualatin suddenly becomes a Celeron part next quarter (with 256K cache), and ramps rather quickly. Looking at this chart, you can see that Intel is putting a great deal of pressure on the Duron at the low end, and will be way ahead of Athlon in raw clock speed (even if not necessarily performance). This may very well give Intel a significant marketing advantage going into the upcoming holiday season.

Pricing is obviously another very good indicator of how much competition there is, and a number of other publications have already provided that information. Once again, the desktop is where most of the action is, with the mobile market now coming under a bit of pressure. It will be interesting to watch the workstation market as the dual Athlon parts become available, and issues are worked out.