As the first update of the New Year, there is more than usual to report, though you might have seen much of it reported already elsewhere. It seems that the last half of 1999 brought some pleasant surprises and some frustrating events. I can’t say that it looks like this will continue for the first half of 2000, but the first couple of months do look interesting indeed…
It appears that Intel has been it’s own worst enemy recently when it comes to the chipset market. Demand for the highly anticipated i820 chipset is extremely limited for several reasons. First and foremost is the nagging perception that the chipset is ‘buggy’, particularly in regards to DRDRAM support. In addition, the 820 is more expensive than the chipset it is intended to replace (i440BX), yet the SDRAM performance is no better, with some early testers claiming it to be worse. All in all, the i820 chipset looks to be essentially stillborn at this time, though demand could increase if DRDRAM prices fall dramatically over the next several months.
The i840 chipset has been looked upon as the chipset of choice for high-end systems, due to the dual Rambus memory channel support. Unfortunately, the pricing is very high, and when combined with the extremely expensive DRDRAM memory, it is unlikely that the DIY market will see any products based upon this chipset soon.
The i810 chipset experienced it’s own share of problems early in 1999, which prevented the quick transition that Intel had hoped for. The i810E has not had any reported problems, however the integrated video has prevented the DIY market from accepting this chipset to any great degree. The OEM market has moved more aggressively towards this chipset, primarily because of the cost savings over a solution requiring a separate graphics card. However, it appears that the overall market share for this chipset has been lower than originally anticipated.
One has to wonder now whether Intel will accelerate the introduction of the i815 (Solano) chipset in an attempt to recapture some of their lost customers. This chipset will support PC133 memory and includes integrated video – however it will also allow for an AGP slot (4x AGP), which will automatically ‘override’ the built-in graphics when a card is inserted, making this potentially attractive to both the OEMs and end users who want to have a choice regarding their graphics solution.
The i440BX chipset is still enjoying great popularity in the DIY market, primarily due to it’s superior performance with SDRAM and lower cost vs. the i820. It seems that this ‘aging’ chipset is still considered by many to be the flagship product in the Intel chipset line. Most manufacturers have indicated that availability of the chipset seems to have gotten better during the past month, though there are still some reports of minor shortages.
There have been some rumors that Intel may ‘tweak’ the BX chipset to officially support a 133MHz FSB, however there has been no mention of this from any of my sources. Unless Intel has something tricky up their sleeve, it seems that they don’t have much going for them until the Tehama chipset debuts sometime after mid-year.
During the last half of 1999, VIA enjoyed quite a bit of success with their Apollo Pro133 chipsets (Pro133 and Pro133A). Recent conversations with 5 of the top 10 motherboard manufacturers has revealed that shipments to OEMs are predominantly VIA based boards – between 60% and 70% of the total! I was actually taken somewhat aback upon first hearing this number, but subsequent discussions with others has confirmed this fact.
The reasons given for this surprising market strength is primarily price – the VIA chipsets cost significantly less than the equivalent Intel chipsets. Couple this fact with the i820 chipset debacle, and shortages of i440BX chipsets and you have a formula for significant gains by VIA. In fact, the main reason that VIA does not have even greater market share is because they cannot currently supply any additional product. Virtually every chipset VIA can produce is being shipped. Though no specifics have been given, a VIA spokesperson told me that manufacturing capacity is being taken very seriously and should be increased this year.
While shipments of the KX-133 chipset are already being made, expect an official announcement regarding volume shipments around the Jan 10 timeframe. There have been some questions about the ‘maturity’ of the chipset, however these are unfounded as it is essentially an Apollo Pro133A chipset with an EV6 bus. There have been some questions about the 4x AGP support, and in fact early versions of the Pro133A chipset worked only with nVidia and S3 graphics chipsets. VIA has assured me that they have since verified the chipsets with all major graphics chipsets, so there should be few problems with the AGP feature. Of course, users likely won’t see any real improvement with AGP 4x, but that has never stopped anyone before… ;-).
Tom’s Hardware Guide revealed late in December that AMD had fixed a small problem in their AMD 750 chipset that will improve performance by as much as 5%. This is on a platform that already has a performance advantage over it’s Intel competition. Looking forward, the next generation AMD chipset is planned to support DDR, AGP 4x, 66Mhz PCI, and offer dual-processor capabilities. This chipset will likely not be released until Q3 or Q4, and is intended for use with the Mustang processor (see the Processor Update for more information).
Be the first to discuss this article!