Introduction
It is not the first time that one of Intel's competitors in the x86 market seems to have a very good upcoming product. I remember already a few occasions when the announcement of a non-Intel CPU maker was making us believe that they are going to give Intel a good run for its money. However, I cannot remember a time when it was looking almost obvious, that the performance of even Intel's high end x86 products would not only be matched, but even surpassed.
AMD's announcement of their upcoming K7 CPU at the Microprocessor Forum 98 in San Jose could indeed mark a major change of the x86 CPU market in 1999. K7 will almost clearly beat Pentium II (Deschutes) performance and it looks very likely that even Intel's next high end product, code name 'Katmai', will have a serious problem competing against K7.
This is not all though. AMD is very close to shipping the K6-2 at 400 MHz . This CPU will not just be the well-known 'normal' K6-2 core merely running at 400 MHz, but it will use a new revised core , which offers increased performance. This way the K6-2 400 will hardly fall short against Pentium II CPUs at the same clock speed.
Then there is 'Sharptooth ', the CPU core that will be used in the upcoming 'K6-3 '. This core will be AMD's first CPU core with on-die L2 cache, as already found in Intel's Celeron 300 A and Celeron 333, code name 'Mendocino'. 256 kB L2 cache running on a backside bus at CPU clock speed will improve the performance of the K6-2 core tremendously, making it faster than a Pentium II at the same clock speed already. The additional beauty of 'Sharptooth' will be the fact that it will simply fit into any of the Super7 boards, making it a perfect candidate for super simple upgrades. Socket7 platforms will then be faster or at least as fast as the fastest Slot 1 platforms for the first time in history. This should be a serious reason to not leave Socket 7! Sharptooth is no vain dream or a simple promise, it's already up and running for quite a while. One reason why AMD is not shipping Sharptooth yet is that the OEMs don't want AMD launching a new CPU right before the Xmas market, and the other reason is that AMD expects that the OEMs pay a higher price for this significantly improved product, which still is subject to negotiations. Sharptooth will ship at the beginning of 1999, maybe slightly earlier, and it will ship with clock rates of up to 450 MHz. It's certainly annoying to a lot of us that we cannot have Sharptooth now, but the outlook on K6-3 should definitely keep everyone from leaving Super7 in favor of Slot 1.
You can see that AMD is coming very strongly already, offering very high performing solutions for Socket 7 platforms. The following shall explain why Socket 7 platform owners should have no reason to ever move over to Slot 1. It could also be that Slot 1 platform owners will have to face leaving Slot 1 for achieving the highest performance. The answer is 'Slot A' and the CPU plugging into this Slot will be called 'AMD K7'.
Dirk Meyer's presentation of the K7-features at Microprocessor Forum 1998 on October 13th was certainly the most impressive presentation in the CPU section. Dirk drowned the auditorium in a flood of high tech terms at ultra high speed, whilst showing a Poker face as if he was telling us about the weather. Hardly anybody was able to follow him. I was one of the lucky people who had the chance interviewing Dirk a couple of hours later, side by side with Dana Krelle, AMD's VP marketing. I still admit that I don't understand all of the great features of K7, but I think I've got a pretty good idea.
