May 2000 Industry Update

Pages: 1 2 3 4

Processor Update

Intel Processors

What has to be the most interesting bit of news for this month is the reports that Intel will begin shipping Katmai processors in volume once again. Yes, I am talking about Slot 1 Pentium III 550, 600 and (perhaps) 650MHz processors in quantity. One manufacturer told me that Intel notified them last week that as much as 60% of all shipments will be comprised of these processors until at least September.

The reason given is that the FC-PGA packaging is still a major problem that has forced Intel to push out their higher speed Celerons, and has prevented higher speed Coppermine Pentium IIIs from being made available in quantity. This seems to finally provide some solid evidence as to what Intel’s problems have been these past 6 months with regard to processor shipments.

During the last two quarters, Intel apparently was faced with the choice of shipping processors mostly in the ‘sweet spot’ speed grades, or at the high-end. The smart choice would have been the former because that is where the bulk of the revenues come from, but at the same time would have limited the quantities of higher speed parts. This would not be as much of a problem when the competition isn’t shipping large quantities of high speed parts either.

Unfortunately for Intel, AMD has now ‘come of age’ and can ship high-speed parts in volume. In order to avoid losing the market segment that provides the greatest profit margin, Intel seems to have realized they were between a rock and a hard place, and they may have been pushed a bit by Dell. Reportedly, Dell has not been happy with the availability issues and has been making noises about building Duron based systems in Q3.

Assuming that the reports are true, this is actually not a bad move for Intel. If they are able to satisfy the OEM demand for the 500MHz to 600MHz parts, and also ship sufficient high-speed (700MHz to 1GHz) parts to prevent further market share erosion, they will have given themselves more time to get Willamette to market. Motherboard manufacturers may also like this development, since their existing Slot 1 designs have now been given a new lease on life.

AMD Processors

As reported last week, wafers for the K-75 are no longer being manufactured. Sources close to AMD said that all inventory has been spoken for, however it is likely that there will be parts in distribution through Q3. Expect that these parts will become worth less after the introduction of the newer Athlon parts become available. Fab25, which is where the K-75 was being produced, will now manufacture K6-x, Duron and Thunderbird parts. All parts manufactured in Fab25 will be using Aluminum interconnects as well as the AMD developed process (vs. Copper interconnects and the HP6L process used in Fab30). I have no information at this time as to whether there will be any identification of these differences on the finished products.

AMD announced two weeks ago that they had decided upon ‘Duron’ as the official product name of the Spitfire processor. This CPU will apparently sport a 128K L1 cache (64K D-cache, 64K I-cache) and a 64K on-die, exclusive L2 cache. In case you are puzzled over how they could possibly have a smaller L2 cache, the term ‘exclusive cache’ means that the L2 cache does not mirror what is in L1, and therefore actually allows for a total of 192K of cached data/instructions. This processor will be available only in a Socket A package, and will most likely require a motherboard based upon the VIA KZ133 chipset (or possibly based upon the AMD 750 chipset, though it isn’t likely manufacturers will do this).

It is not very likely that ‘Slocket’ cards will be made available that will allow users to keep their existing Slot A boards, for reasons provided on the ‘Chipset Update’ page. There have been some reports that manufacturers have built prototypes, but they are considered to costly to build for the mass market.

The Thunderbird processor is still lacking an official name, and it now appears that it will be released to OEMs first in a Slot A package (June timeframe), while the DIY market will probably only see them in the Socket A package sometime in July, or later. As reported on the ‘Chipset Update’ page, existing KX133 based boards will very likely not support the Thunderbird, in any case, so the fact that few Slot A versions will be available to end users seems to be a moot point.

I do not expect the Duron to exceed 650MHz at introduction, and those anticipating a 1GHz Duron this year are likely to be very disappointed, as there is absolutely no marketing reason to do so. Unless Intel boost the Celeron speeds beyond 700MHz, there is little chance that the Duron will go their either. After all, it doesn’t make much sense for AMD to compete against their own Thunderbird processor.

Sources close to AMD have hinted that the Thunderbird will likely be introduced at around 750MHz, with speeds available up to 1GHz. There have been some reports of 1.2GHz at introduction, but it seems that these were more hopeful anticipation than factual information. The official AMD line is that Athlon will achieve 1.5GHz by Q1 ’01, so unless Willamette is released early, and in volume, chances of anything beyond 1.4GHz this year seems unlikely.

Pages: « Prev   1 2 3 4   Next »

Be the first to discuss this article!