Some Thoughts About The K7

(11/28/98 update) At Comdex, both VIA and ALi confirmed they are making a K7 chipset and at least two motherboard manufacturers assured me they will be making motherboards. This should certainly be enough support to convince others to make products for this exciting AMD processor. Check out the 11/21 entry in the Industry News for more details

(First Update) Since this editorial was written, VIA has announced that they will be making a chipset to support the K7/Slot A processor. In addition, I have been informed that AMD has a working prototype (using their own chipset) that they are showing to motherboard manufacturers to try and convince them to build product to support the K7. The manufacturers I have spoken with are very reserved about the viability (no pun intended) of this platform in the marketplace. I think that to ensure future competition, users need to voice their opinions on this to these manufacturers. I encourage you to send emails indicating your support of the K7, if you indeed would have an interest in such a beast. In addition, if the Sharptooth (formerly K6-3) is very successful, several of these manufacturers would take that as a sign that AMD is indeed a player in the market, which is necessary for these guys to consider ‘dissing’ Intel.

After the article about the K7 appeared on Tom’s Hardware Guide many people sort of went nuts about how Intel is now on the ropes. While that may be at least partly true, there is still a long way to go before this can happen. Though the processor itself looks very promising, a number of issues still have not been resolved, and some questions still not answered.

One of the biggest issues still to be addressed is who will make the chipset to support the K7? Major motherboard manufacturers are asking that question, and have not yet been given an answer. As a result, no real plans can be made by these manufacturers. In order to guarantee a smooth rollout, a lead time of at least 6 months must be provided and time is getting short. AMD apparently has developed specs for their own chipset, similar to what they did with the K6-2 and AMD 640 chipset. In that case, they dropped the plans for production because VIA and ALi came to the table with their support. AMD is hoping that the same thing will happen for the K7, but so far there are no firm commitments.

AMD has also not decided exactly which market segments they will be targeting with this processor. While it is obvious that the high-end will be their primary market, they also have the opportunity to target the mid-range market. Because of the wide range of L2 cache sizes available, a low cost version with a small amount of cache could be offered with a more expensive flavor with up to 8MB onboard. This decision will partly be determined by how well the Sharptooth (formerly K6-3) will do in the market. Many are expecting that the K7 will be a low-cost alternative to Intel, but if AMD decides initially to only offer it to the high-end, most consumers will find it far too pricey to afford for quite awhile.

Another question is what memory will be used to utilize the 200MHz bus speed. Initial reports seem to indicate that DRDRAM will be used, however licensing and availability issues may mean high premiums for this solution. If that is the case, the processor may not be considered an attractive option for the mid-range due to the high memory costs. In this case, a lower cost option, such as DDR SDRAM or SLDRAM would be necessary. This support would be entirely determined by the chipsets available, and right now only VIA supports DDR SDRAM, while no chipset has support for SLDRAM at this time. If VIA, ALi or SiS were to produce a chipset with support for the K7 *and* had support for DDR SDRAM, SLDRAM and DRDRAM then the options would be very good indeed.

One final consideration is the assumption by many that the Alpha 21264 chip will be virtually plug compatible with the K7. While both processors are based upon the same bus protocol (EV6), they are not guaranteed to plug into the same slot. At this time, there are no chipsets to support the K7, therefore there can be no discussion of support for other processors yet!

What this really all comes down to is that while AMD may have a great processor design, the motherboard manufacturers must be convinced there is a solid plan, with definite solutions or they will simply not include the K7 in their roadmaps. You can see that this nearly happened with the K6-2, as several manufacturers either did not build SS7 boards or came very late to the party. AMD will now have the job of convincing the large manufacturers that there *will* be a supporting chipset, and will have to provide some details on the specifications.

I for one am rooting for AMD to succeed here, but they will need the help of chipset and motherboard manufacturers to truly have an impact. One positive aspect of the current level of interest is that the chipset manufacturers may see this as encouragement to get onboard with AMD. Once there is concrete evidence of such chipsets, motherboard manufacturers will no doubt start clamoring to design their offerings. With alternative processor, chipset and memory solutions, Intel will not be able to completely control the market, and hopefully this will mean that the best technology will win out.


Be the first to discuss this article!