December 2000 Industry Update

Pages: 1 2 3 4



As noted in the Chipset/Motherboard section, some integrators have reported stronger than expected interest in the Pentium 4 from the high-end desktop/low-end workstation market. One integrator said that sales of systems using this processor were mostly to those who had previously purchased Pentium Pro systems several years ago, and saw the P4 as the proper upgrade path. Overall, however, the P4 seems to be a niche product at the current time.

Pentium III and Celeron processors have not had any interesting developments of late. Rumors have made the rounds that the 100MHz FSB Celerons will debut soon, which could spark a bit of interest, but they will still not match the performance of the AMD Duron (which, however, has infrastructure issues that must be addressed). On the other hand, the Celerons are very popular in the mobile market, because they compete very well against the K6-2 – which is the only mobile offering from AMD at the moment.


Many rumors have circulated during the past month that AMD might be having manufacturing problems, or that the Tbird core might be ‘maxed out’ because no speeds faster than 1.2GHz have appeared, as some have expected. Though consumers have now been sort of trained to expect rapid increases in processor speed, the fact is that Intel is not currently threatening AMD in the desktop market. It would not make a lot of business sense to keep increasing the MHz when Intel is not able to even match the current Athlon speeds. Obviously, there is a limit to the speed the current core can reach, but it makes more sense to try and keep a little breathing room just in case there are some delays in getting the Palomino core out the door.

This is not to say that AMD definitely could bump the current Athlons another speed grade or two higher, but my insistent questioning of an AMD representative was rewarded with the statement that if necessary, AMD would consider putting out a faster speed Athlon before the Palomino is introduced. I also believe that AMD would prefer to intro the Palomino at the lowest speeds possible, giving them a little more headroom in the GHz race with Intel’s P4 next year. The secret to this game is staying just enough ahead of the competition to claim the performance title, but not so much ahead as to guarantee running out of headroom before the game is over.

Vendors I have spoken with claim most interest is between 700MHz and 900MHz for both Pentium III and Athlons, so it is possible that AMD is down-binning to satisfy demand. The most likely explanation for the lack of higher-speed parts is that they are reserving the highest speed grades as 266MHz FSB parts for early in the year when DDR motherboards and memory are available. These would not only demand a higher premium, but the pent-up demand may be great enough that AMD needs to do a little stockpiling to ensure a smooth DDR rollout when the time finally comes. It could also be that the Dresden ramp is happening much slower than expected, so some of the manufacturing capacity for Tbird processors has been sacrificed to make the Palomino processors

Of course, this is all speculation, but an AMD representative I spoke with said that he was not aware of any manufacturing uses, nor was he aware of increased OEM demand. This leaves only the possibility of stockpiling, a shift in production or both, as far as I can determine.

Lack of a low-cost, integrated chipset was cited as a hindrance to Duron sales last month. Though VIA announced the KM133 chipset in late September, the move to this chipset has been somewhat slow, so there are obviously some other issues involved. Reports seem to indicate that Duron sales in Asia are decent, however Europe and the U.S. seem to prefer the slightly better performing Athlon. Most likely, the lower cost of the Celeron motherboards has kept the Duron from attacking the low-end, and this has definitely created a minor problem for AMD in terms of processor shipments.

While is is apparent that the Athlon has eaten into Pentium III sales, based upon the comments from motherboard manufacturers and resellers, the Duron has not done as well as AMD may have been hoping. The net result seems to be that the overall market share for AMD has not increased during the past few months, after a slow but steady gain during the first half of the year.

There is still demand for K6-2 processors in the mobile market, and there are even a fair number being sold for the desktop market. One motherboard manufacturer indicated that as much as 20% of their shipments are Socket 7 boards, with another acknowledging that their Socket 7 board was one of their top 5 sellers even now. Unfortunately for AMD, the Celeron is offered at much faster speeds, making it the preferred chip for low-end notebooks right now. This has likely reduced the market share for AMD in the mobile market during the past quarter.

Many consumers are expecting the Palomino processor to hit the market in early Q1, however there are strong indications that they will not appear until later in the quarter, along with the mobile version. Morgan (both desktop and mobile) will probably appear about a month or so later, in early Q2.

VIA Cyrix

VIA and TSMC have announced that they have successfully produced the VIA Cyrix processor using a .13u process. This will very likely make VIA Cyrix the first company to offer .13u parts in the world when they actually become available, giving them a decent chance at grabbing some mobile PC and IA market share.

One issue that was mentioned to me is that the FCPGA Pentium III and Celeron processors may not function in older Socket 370 motherboards. Intel documents this on their website. Interestingly, this situation may create a ready made market for VIA Cyrix processors.

Though the majority of systems sold in the U.S. are made by OEMs such as Dell, Gateway, Compaq, etc. there are quite a number of systems integrators who cater to a specific market using a set system configuration. Changing the motherboard requires revalidating all components, which can be a time-consuming and costly process. For this reason, these integrators really hate it when a motherboard is discontinued because usually their customers are paying for stability and reliability – and any new features can be implemented by the use of an add-on card, which is much easier to validate than a new motherboard.

Integrators who have developed systems based upon these older motherboards may very well prefer switching to VIA Cyrix processor vs. changing the motherboard. It is possible that the applications required by the buyers of these systems would not be affected by the relatively poor FP performance of the VIA Cyrix processors.

Pages: « Prev   1 2 3 4   Next »

Be the first to discuss this article!