The future of the AMD-PC (?)
A few months ago he was publishing on these pages the news about the presentation on the part of Intel of his future scale of microprocessors, with the title of "Xeon, Katmai and Favor: the future of the Intel-PC". In that moment it looked like to me that to relate "Intel" and "PC" was slightly logical, although it is possible that in the future we have to consider to speak also about the AMD-PC...
For many of those that we move in this world of the hardware, Microprocessor Forum 1.997 (a computer fair, species of SIMO of the mikes) supposed a milestone due to the conference that on the part of AMD gave W. J. Sanders, in which he was announcing, nothing more and neither more nor less than:
This announcements bombing supposed that AMD was thinking to take the initiative in the world of the manufacturers of mikes, separating from his habit of being on tow on the Intel advances. Obviously, we show many to ourselves skeptics (although given hope) of that AMD could fulfill so many people and so good promises; well, since I must say that we are wrong.
On October 13 of this year, in Microprocessor Forum 1.998, AMD there was ratified in his plans for descabalgar the giant Intel of his position of predominance in the mikes manufacture. Nobody doubts already the success of the K6-2 with 3DNow!, that are in manufacturers' computers of the size of IBM, HP, Compaq or Packard Bell, what gives them the commercial prestige necessary to be thrown to more enterprising adventures.
The first one of them is the fiancé K6 + 3D, that with all probability will call commercially K6-3, or with his key name Sharptooth (sharp tooth, a very significant name). It is a question of a mike with a nucleus similar to that of the K6-2 but that includes 256 KB of integrated cache memory, running at the speed of the mike. This means that the above mentioned cache memory will go approximately to: 400 MHz!
So that we should do to ourselves an idea of what supposes this, we think that the cache memory of Pentium II goes to half of the speed of the mike, what does a 225 MHz maximum for Pentium II to 450 MHz, and only Celeron Mendocino overcomes this number, coming up to 333 MHz. What is more, the current ones K6-2 support a yield almost identical to that of Pentium II although his cache memory goes to "only 100 MHz.
For it, the K6-3 will be more rapid enough than Pentium II, being located at the level of future Intel Pentium III (Katmai) or even a little superficially depending on the used software. Also, the aggressive trade policy of AMD and the saving that 7 supposes the use of Super badges instead of Slot 1 allows to suppose that his price will be more under that that of the mikes Intel, what is always a weight reason...
As curious technological aspect, the integration of the cache memory of the second level in the chip (cache memory "L2", this 256 KB on that we were commenting earlier) will allow the incorporation of a cache memory of third level (L3) in the motherboard. The above mentioned cache memory L3 will not be definitely necessary, since his yield will be much minor than that of the cache memories L1 and L2, but to implant it will suppose an extra help which there lack Pentium II or Celeron (the "classic" Celeron - not Mendocinos - do not even have cache memory L2).
AMD K7 Athlon
But they do not finish here the worrying announcements for Intel, not much less. AMD has confirmed his intention of making the microprocessor K7, which characteristics it looks like that they place him not only much over Pentium II, but beyond the possibilities of the Xeon or Pentium III of Intel.
The K7 will be a chip of really resounding characteristics:
If the previous thing turns out to be amazing of for itself, it it is not less the fact that the microprocessor will be placed on a new type of socket that AMD names "Slot A", which physically will look alike very much to the Slot 1 of Pentium II but with which it will not be compatible.
Being sincere, it is very probable that AMD had preferred to use the Slot 1 for the K7, but it turns out that the technology on which it is based is registered by Intel, who refuses to license it to his competitors. In fact it is a question of a proprietary design, something really unusually in the world PC although normal for the users of other types of computer as the Macintosh. So not to fall down in legal problems for copy of designs, inverse engineering and the others, AMD has turned out to be forced to resort to another system bus, the EV6 of the Digital company (famous for his powerful mikes Alpha of 64 bits), and to connect it to a socket similar to the Slot 1 but electrically incompatibly with this one.
Although at first this might suppose a disadvantage, it is possible that in fact it avenges well the K7. After everything, to construct a badge Slot A debería to be neither more expensive nor difficult than to do a Slot 1, with the added advantage of which the control chipsets will not be necessary Intel (as it happens at present in immense most of badges Slot 1), but they might be made by several companies (ALI, ROUTE and others) and even by the same AMD, what would result in a price reduction of the prices. It would be curious that the politics pro-monopoly of Intel was ending up by costing him money...
Conclusions and calendar
Very well, we have seen a lot of technical information, but they will be wondering: and what? Will an AMD-PC be worth being bought or we will continue with the well-known binomial Intel-PC? In my opinion the answer is summed up briefly: if Intel not despabila rapid, the next sticker that will take his computer will put AMD-Inside.
The calendar foreseen by AMD for his new processors indicates that the K6-3 Sharptooth will have to be on the market in the end of the this year, almost sure than for December. The K7 will take something more, partly because it is a much more complex design and also not to eclipse the throwing of the K6-3; we will see it on the market in the first semester of 1.999, probably about the spring.
AMD has put the letters mouth above, now he has Intel to accept the challenge or raise the bet. The K6-3 promises to be a fantastic mike, with a big yield in all the applications and unbeatably in games prepared for 3DNow!, while the K7 is a bet of future capable of beating all the speed records. The only weapon foreseen by Intel for this battle calls itself a species of Pentium II Pentium III or Katmai, with new instructions for 3D calls MMX2 or KNI, like them 3DNow! but a little more powerful (again Intel refuses to use standards of the competition, preferring the proper ones although they are redundant).
The principal advantage of Pentium III is that it will go out to the market most likely with a pair of months of advance with regard to the K7, but it does not seem that advantage is going to be sufficient since seemingly it will be very lower than this one. Although it is never known, Intel has not published too much information on the final design of Pentium III; at the last hour it might inflate his cache memory L1 up to 128 KB and/or raise the system bus to 133 MHz, with what the differences would turn out to be limited. Another possibility would be to lower the price of the Xeon or, preferably, to move forward the throwing of the Xeon with MMX2 (the Tanner). Any way they must hurry up if they want that Intel-Inside keeps on being an attractive advertizing claim...