There's never been more choice for system builders or assemblers than now. Two years ago, Intel effectively ruled the micropressor roost and that meant that whether you liked it or not, you were forced to take both yourself and your customers down their way. The reasons for that were that while both AMD and Cyrix had unveiled their plans for their future strategy, the chips were not out and there was still a lot of consumer resistance to anything which did not bear the Intel name. That's changed. Both industry and senior analysts in the arena now agree that the Cyrix M2 and the AMD K6 match and in some cases, outperform the Intel offerings. The other advantage of using Cyrix and AMD processors is that both these companies are currently actively wooing system integrators. There is also a new entrant into the CPU arena called Centaur, backed by ID, which is promising, and it claims, delivering, a chip suitable for Windows-based machines. But there's no doubt that because of its size and its marketing muscle, Intel still rules the roost, and sets the trend. In January, Intel will substantially cut the prices of its entry-level Pentium II (Klamath) processor, and in so doing paving the way for the demise of Pentium MMX chips and for introducing new technology, still codenamed Deschutes. That will be followed by further price cuts in March/April, which is when the Pentium MMX chips based on Socket 7 designs make their merry way to the microprocessor Gulag. Our information from the deep throats at Intel that we talk to is, however, that the introduction of Slot 2 technology, originally slated for the same time period, is now delayed, probably until the third quarter. However, Deschutes, which uses a .25 micron technology, is still, more or less in time. The introduction of chips based on this higher performance, higher yield design, will mean it is positioned as the high-end, allowing Intel to push the PII as the entry-level. Socket 7, as far as Intel is concerned, is on its way out. But see our separate story on the latest changes to both Deschutes and Slot One. Intel is also preparing a design which will add graphics meat to its Pentium II platform. This technology will introduce extra MMX instructions and, apparently, will be called Pentium III. In 1998, Intel will introduce a number of chips with higher clock speeds and also with what it promises are higher bus rates, giving additional performance and attractive to end users who like sheer horsepower. But, according to AMD, Intel will not have it its own way in 1998. Although demand has exceeded supply for the K6 during 1997, the senior executives at AMD are adamant that they will be able to significantly increase production and are still on target to introduce two brand new processors. Like Intel, AMD is moving to a .25 micron process, allowing it to make more chips from the same wafer, increase performance, and also lower heat consumption. That will allow AMD to make 300MHz, 350MHz and even 400MHz processors, it claims. As this piece is written, it is already sampling a 266MHz K6, which it says will be in production by the end of the year. The two new chips it will introduce next year are the K6 3D and the K6 3D+. The first of these offers signficantly improved graphical performance at the high end and AMD says it has already signed deals with a raft of software developers who are making games for it. It will have a 100MHz graphics bus. The 3D+, which comes much later in the AMD roadmap, will have a large die size but wholly eliminates the need for Level 2 cache, and offers such caching advantages through its 21 million transistors that it is likely to show a large performance gain over anything the company has made before, it claims. Cyrix was acquired by National Semiconductor earlier this year and that is probably good news for the company. NatSemi has such financial reserves that it will be able to back the Cyrix engineers' plans. They, as far as we can gather, are to follow AMD to some extent by producing faster processors which also have large cacheing abilities on board, while at the same time firmly targeting the consumer market with their MediaGX offerings. The new player on the block, Centaur/IDT, is targeting assemblers and system integrators who want to keep their costs down while still being able to perform Windows compatible systems. The company stresses that it is not offering anything other than a fast and very inexpensive Socket 7 based design for Windows. For exaple, while the unit has been tested for compatibility with Windows, Win applications and games, you would be unwise to use this chip for a Unix box. As with both Cyrix and AMD, Centaur is preparing a faster 3D version of its part with large cacheing abilities and also better floating point unit (FPU). That is likely to appear in 1998. Where the Centaur part does score is that it will significantly undercut its competitors on price while being able to ramp up its production fast because IDT, its parent, has excellent fab capabilities. The trick, next year, will be to watch the pricing from Intel carefully. We are reliably informed the Pentium II entry level cuts will be very significant. AMD has the ability to still undercut Intel because of its financial model. The big question for AMD is whether it can sort out its supply/demand problems and make enough parts to satisfy the hungry assembler market. ®