Promising a future upgrade path for current socket based systems, does AMD have the reputation necessary to inspire an Intel driven world to give them the backing they need to come forth with their next generation of processors? Or must we all follow the foggy path Intel's Slot-1 and its successors will pave for us? What happened to the days where companies like AMD, Cyrix, and Intel all tailored to different factions of the market, must there be only one "winner" in the microprocessor industry? This is a capitalistic world, where anyone and anything can succeed given the proper backing, so why should one believe that it's either Intel or bust?

The Transition from the 486

If we could trace Intel's incredible lead in the race for the faster microprocessor down to a single earth-shattering action, it would have to be the transition from the 486. As a little refresher course for those of you who were into computer hardware during the days of the 486, AMD had finally begun to catch up to Intel in terms of performance, reliability, and their outstanding price produced an obvious threat to Intel's grasp of the 486 market.

The solution? Change the voltage specification for the most popular processors of the time, the 486 was on the verge of making history without even realizing it. The 486's 5V operation ran too hot for the future as Intel saw it, so by dropping their processor specification down to 3.3V and by eventually flooding the market with these new low power chips, the fault line between Intel and AMD began to tremble, preparing for a quake of tremendous proportions with only one survivor coming out on top.

After Intel's little voltage fiasco with the 486 AMD attempted to shake off the loss by jumping on the 3.3V bandwagon as well, after all, the move towards the 3.3V core voltage standard was a logical one and can't be solely considered a shutout marketing tactic. In fact, AMD will have aspired to promote two non-standard voltages with two of their processors years after the days of the 486, the 2.9V and 3.2V with the K6/200 and 233 respectively.

This is where the most controversial action of the time took place, Intel's decision to pursue a new, 5th generation microprocessor architecture obviously before the peak of the life of the 486. Intel new "586" processor, as it was dubbed, offered very little above the top of the line 486 processors of the time, other than a promise of a future upgrade path through the use of its revolutionary new socket specification. Prior to its release (and eventual failure of the 586 as it was replaced by the Pentium), 486 processors either made use of a soldered design or LIF socket (low insertion force) as an interface, with most newer 486's taking advantage of the new ZIF (zero insertion force) socket, more specifically, the Socket-3 design. Upon their release, the Pentium processor, offered in a 60 and 66MHz flavor, made use of a more "advanced" Socket-4 design which offered more pin assignments for the processor and theoretically allowing for a more advanced processor to be run on the motherboard. This would prove to be another failure as Socket-4 was quickly replaced by Socket-5 and shortly thereafter put into retirement in favor of a truly more advanced Socket-7 design.

While all of this was taking place, AMD was still stuck back in the days of the 486, eventually reaching speeds of 133MHz with their highly overclockable enhanced 486 processors. Cyrix, barely mentioned until now, was in the same predicament, they had much time and money invested in an excellent design however with one swift flick of the wrist, the market changed directions leaving both Cyrix and AMD in the dust.

Why bother you all with a lesson in microprocessor history?  It has often been said that history repeats itself, and if you were to step back from the current state of the microprocessor industry, it's almost inevitable that you would notice a definite trend forming before your eyes. 

Let's fast forward a little past AMD's race to catch up to Intel with the original Pentium, and move to the days of the Pentium MMX.   Not too long after Intel's release of the Pentium MMX, AMD finally had the answer to their prayers ready and waiting for a debut, the AMD K6.  The K6 was pin-for-pin compatible with the Pentium MMX, and on paper, it offered much more for the money, but would it be a success?  After its release, the K6 turned out to be an overall success, although it did have a weak presence among the heavy 3D gaming community it wasn't too harshly bashed for its shortcomings in the FPU area for quite some time.   Offering a viable alternative to Intel, often times at an equal or lower cost (initial production difficulties set aside), AMD had a sure fire winner on their hands...Intel's solution?

Change the interface specification for the most popular processors of the time, this time around, the Pentium MMX was on the verge of repeating history...sound familiar?  It should.  No longer than a month after AMD's debut of the K6, Intel released their Pentium II which made use of an entirely new interface design, abandoning the Socket-7 standard for a slot based interface, one we've come to know as Slot-1.  If you're beginning to notice the trend more prominently now, let's compare it to the past, a more "advanced" Slot-1 design which theoretically allows for a more advanced processor to be run on the motherboard in comparison to Socket-7 which is now entirely supported by Intel's competitors.

Slot-1: The Socket-5 of the Pentium II

More or less, one could say, that clock for clock, the Pentium II and AMD's Socket-7 based answer, the K6-2, are virtually equivalent in terms of overall performance.  All 3D Gaming and CAD performance set aside, both mainly FPU dependent in which case the Pentium II's advanced dual pipelined FPU gives it the lead over the K6-2, the Pentium II and K6-2 can be placed side by side for a comparison, at the same clock speed, and very few users would notice the difference.

Figure 1. Processor Performance Comparison as a Percentage

Does this seem reminiscent of the limited days of the Socket-5 standard?  While Socket-5 eventually reached speeds of 133MHz, it wasn't until Socket-7 was introduced that processors started to pick up in performance at a more rapid rate.  What users discovered is that their future upgrade path that they were promised turned out to be nothing more than an option to purchase a 133MHz Pentium or a blatantly overpriced Socket-5 overdrive processor.  How comforting.

It is highly unlikely that Slot-1 will continue to be the dominant interface design from Intel as time goes on, with the introduction of the Katmai and the Socket-370 based Celeron processors in the very near future, Intel will probably abandon Slot-1 in favor of a more distancing architecture, possibly, Slot-2; therefore maintaining the gap between themselves and AMD as well as the rest of the competition.  It will be interesting to see what the future upgrade path we're being guaranteed now will turn out to be in the next year, it looks like we're almost on the verge of another one of those spontaneous specification changes...wonder what it'll be...

If you can't play by the rules, leave.   And that's exactly what the competition, particularly AMD, is doing.  Tired of fighting the war for the Socket-7 platform, AMD has laid out, beautifully, in plain black and white, their plans for the future.  Allowing the current Socket-7 users a bit of a transitional period in which they can get used to some of the features of a slot based system such as higher speed memory, AGP video cards, and definite performance improvements, AMD managed to spark quite a bit of interest in the market once again.  

Instead of playing by Intel's rules, AMD will continue to manufacture Socket-7 processors which will continue until early or mid 1999 at which point they will be abandoned in favor of a truly more advanced slot design, loosely known as Slot-A.  By not following the path Intel paved, AMD has opened themselves up to a world of success as well as a world of possible failures.  AMD is pursing a completely new design of processor with their first Slot-A entry, the K7, and unfortunately, we have nothing to compare this movement to in history.  Never has a microprocessor manufacturer made such a bold move against the reigning champ, however if there is every a time to do it, that time would be now.  Let's take a look at how AMD will be turning the tables as well as the best possible upgrade path for the user looking to save money yet have a decent system.

By the end of this year, AMD's K6-2 will be available in a much more competitive clock speed.  Hopefully reaching the 400MHz mark in the next two months, the K6-2 will be able to provide users with a viable alternative to the Pentium II.  While the Pentium II may offer greater levels of performance, the cost of constructing a K6-2 400 based system should be somewhat less than constructing an equivalent Pentium II system, with a future upgrade path lying in the hands of the upcoming K6-3 which is much like AMD's version of Intel's Celeron A. 

Which brings us to the next topic of discussion, if the K6-2/Socket-7 in general, is so much more economical than the Pentium II, what about the Celeron A?   Although the Celeron A is an incredible overclocker (for the most part, there are quite a few failure stories out there), if we're not talking about an overclocked system, the Celeron A is nothing more than a lower cost Pentium II maxing out at 333MHz.  If the performance you're looking for is in the Pentium II 300 - 333MHz range, then the Celeron A is definitely an excellent solution, especially since it does offer that overclocking opportunity we've all come to enjoy.  From an even playing field however, the K6-2 at 400MHz, should give the Pentium II 400 a fairly decent run for the money.

The Future 

So you have a K6-2, and you've been guaranteed (ugh) a future upgrade path to a K6-3...but then what?  Then you make your choice.   Intel has abandoned Socket-7, however in 1999, AMD, Cyrix, and Rise, among other microprocessor manufacturers have announced new processors which should all breathe new life into the slowly aging market.  The best course of action right now, is to save money in a Socket-7 (Super7) system, keep it up to date with AMD's promised release of the K6-3 later this year or early next year, and then make the move towards a slot based system, whether it be a Slot-2 Intel system or a Slot-A AMD system, we've got around a year to wait around for the two feuding manufacturers to produce a chip that's what we want.  Until then, don't go all out expecting to live off of your system for a few years, pick up a decent priced system that does what you need it to, and prepare for another history making move in the hardware world...this time the instigator is up for grabs, will it be Intel?  Or AMD?  Or maybe even someone else...only the future will tell.