As we close the book on another chapter of our history we approach the year 2001 with a great deal of uncertainty.  If you remember, going into 1999 the world and the technology market in general was a much different place.  We were on the rising side of what would turn to be an incredible financial roller coaster and at the same time we were witness to the decline of competition in the market. 

The previous year had ended on a few interesting notes.  For one thing, in 1999, there was virtually no competition in the desktop PC chipset market.  Intel controlled the vast majority of the CPU market and in doing so they also controlled the majority of the chipset market as well.  We had been used to having VIA and ALi alternatives present in the Super7 motherboard market that it came as quite a shock when the only Slot-1 motherboards you could find were based on Intel chipsets. 

However if anything, 1999 was a year, which illustrated that necessity, was truly the mother of invention.  The 440BX chipset met the needs of virtually all Pentium II/III owners leaving no room for VIA to gain much market share with their delayed Apollo Pro Plus chipset. 

This same philosophy allowed AMD to finish off the year in a much better position than they started in.  The applications of the time deemed that there was no need for a new memory technology and a chipset that would force that upon the market was definitely working contrary to the necessity theme of 1999.  This allowed AMD to step forward with their Athlon and begin to gain quite a bit of ground. 

A non-Intel CPU that was actually competitive opened the way for a non-Intel chipset to step forth and provide that CPU with a home.  Not only that but with Intel's flagship chipset, the i820, being characterized as an instant failure because of its reliance on a memory technology that wasn't needed (RDRAM) at the time, the opportunity for a non-Intel chipset to gain some ground in the Pentium II/III arena. 

The calm and structured start of 1999 paved way to a volcanic eruption at the end of the year and kicked off what became commonly known as Y2K, the year 2000. 

Fast-forwarding to the present day, as we enter into the first days of 2001 it is important to take a look at where we've come from.  Y2K was packed with major product announcements, huge releases and mergers/lawsuits galore.  In order to kick off the New Year, we're taking a very in-depth look at the past 12 months. 

So buckle up as we take you through a guided tour of the ups and downs we witnessed during Y2K and analyze what these events potentially mean for this New Year. 

This first article will concentrate on the CPU, chipset and motherboard markets throughout the past year.  A similar article analyzing the graphics chip and card industry as well will follow it.

The Battle for Slot-A Motherboards

Just four months after the release of the Athlon, the battle for Slot-A motherboards continued.  What battle you may ask?  For those of you that don't remember the struggle AMD's Athlon had to face before getting to where it is today, here's a quick refresher. 

The initial success of the Athlon was directly dependent on the availability of reliable motherboard platforms.  Unlike AMD's previous CPU releases, the Athlon was actually available very close to its announcement putting Intel in a very interesting position. 

For once, a processor that was not only faster than the fastest Pentium III but also actually available in decent quantities was threatening their dominance.  Intel's solution?  Since they couldn't possibly stop the flow of Athlons into the market they went after the next best opportunity, put some pressure on the motherboard manufacturers. 

We noticed the first signs of this pressure at Fall Comdex 1999, where very few if any Athlon motherboards were present on the show floor while there were tons of i820-based solutions everywhere.  Later on, with ASUS' refusal to promote their first Slot-A motherboard and their refusal to even ship the boards in ASUS boxes, it was very clear that Taiwanese motherboard manufacturers were in a difficult position between a rock and a hard place. 

On the one hand, selling Pentium III motherboards was making them money and it was what was keeping them alive.  With all due respect to AMD and the Super7 platform, Super7 sales weren't keeping any of these motherboard manufacturers alive, it was Slot-1 and Socket-370 motherboard sales that were bringing in the bucks.  On the other hand, the Athlon definitely had a potential for success, unfortunately none of the manufacturers would be able to capitalize upon that success if they didn't promote their Slot-A solutions just as well if not better than the rest of their product line. 

This obviously did not please Intel and it angered them even more when AMD partnered up with a handful of motherboard manufacturers to make their Slot-A solutions featured and available at the launch of the Athlon.  Here's where the pressure came into play.  By controlling the number of BX chips and amount of support that these motherboard manufacturers received, Intel was able to quite successfully control the promotion of their Slot-A motherboards. 

At the time, talking about promoting Slot-A motherboards with the marketing departments at these motherboard manufacturers was a touchy subject; the best example being ASUS.  They knew very well that their Slot-A solution, the K7M, was the best Slot-A motherboard available at the time.  However the board wasn't being promoted at all, it was very difficult to find and even then the board was only available as an OEM solution in a white box, not in ASUS' usual packaging. 

Luckily for AMD, the first quarter of 2000 brought about some very big changes in the way the Athlon motherboard market was handled.  Manufacturers started being a little more forthcoming with information regarding their Slot-A projects, and for the most part, other sacrifices were made in order to promote Slot-A motherboards.  Eventually even ASUS, in spite of their wonderful relationship with Intel, went back and started promoting the K7M on their Web Site. 

The Athlon goes Mainstream

This wasn't the first time that an AMD chipset release was the cause of increased pressure from Intel.  Not many of you may remember this but AMD tried their hands at a chipset a couple of years back, the AMD 640, a Socket-7 chipset.  The only manufacturer to pursue a design based on the AMD 640 chipset was Shuttle with their HOT-603 motherboard which resulted in quite a few angry phone calls from the big brother, Intel.  Using the LX chipset as a negotiating tool, Intel effectively canned any support behind the AMD 640 chipset much like what was attempted against the AMD 750, the first Athlon chipset. 

A strengthening blow from VIA helped AMD's situation considerably.  With their release of the KX133 chipset, the first non-AMD chipset for the Athlon, motherboard manufacturers felt more comfortable promoting their Slot-A solutions.  The release of the KX133 chipset in February also marked another big win for AMD, this second generation of Slot-A motherboards were definitely much more mature than the first four motherboards that launched with the Athlon in 1999. 

By March we were able to round up a total of 11 Slot-A motherboards for the Athlon platform in our first Slot-A Motherboard Roundup.  Just a few weeks earlier, at the annual CeBIT convention in Hannover, Germany, we reported a total of twenty Slot-A motherboards being on display.  We took enough pictures and got enough data to put together a picture tour through the boards in our only piece of CeBIT Show Coverage.

With Slot-A motherboards gaining popularity, another little device became the focus of attention: the Gold Fingers Device or GFD for short.  This little device would fit over the "golden fingers" (which were actually copper) on the Slot-A Athlon's PCB and would allow for the manipulation of the Athlon's clock multiplier setting, making the Athlon one of the most overclockable CPUs we have had in quite a while. 

Intel Strikes Back - Spring IDF - February 2000

On the first day of Intel's Spring Developer Forum in Palm Springs, CA, Intel dropped a pretty big bombshell: they had their next-generation processor core already up and running at 1.5GHz with two of its integer units running internally at 3.0GHz.  At the time we assumed that this would give the Pentium 4, which was then known as the Willamette, a huge advantage in business application performance.  When the Pentium 4 made its official debut however, we found that even with its 2X clocked ALUs its integer performance was not anything to write home about.  However at the time, it put quite a bit of pressure on AMD since this chip, which they knew very little about, was going to be the Athlon's competition by year's end.

While we were off covering the IDF, AMD was contemplating the release of the world's first 1GHz processor in order to steal some of the thunder from Intel's recent announcements.  This would not only give AMD an ego boost however it would also prove to pressure Intel into what they used to be best at, engaging in price and clock speed battles.  But this time around, AMD was much better prepared. 

We received the phone call from AMD shortly after the announcement of Intel's 1.5GHz Willamette demonstration indicating that they were pushing ahead with the launch in less than three weeks.  When we returned to the AnandTech Labs, the 1GHz Athlon was ready and waiting for us. 

We were impressed with AMD's ability to ramp up the Athlon's clock speed on demand, however we were disappointed with the sacrifice each increase in clock speed seemed to yield.  The off-die L2 cache of the original Athlon was hindering its performance scalability, with the core running at 1GHz the Athlon's L2 cache was stuck at 333MHz.  The Athlon definitely needed an on-die L2 cache, however without any real pressure from Intel the processor could most definitely survive another three months until it would eventually gain it.

Intel's response to this was exactly what AMD was counting on, boosting the clock speed of the Pentium III from 800MHz all the way up to 1GHz.  We were lucky enough to get our hands on one of very few 1GHz Pentium III engineering samples before the launch of the processor, which took place just two days after AMD's 1GHz debut.  This sample was quite crude in terms of the nature of its cooling which we'll give you a quick refresher of below. 

Click to Enlarge

The difference between AMD's launch and Intel's that took place no more than two days later was tremendous.  For the first time in quite a while, AMD's processor was going to be available, in mass quantities before Intel's.  The 1GHz Athlon as well as the 900 and 950MHz parts that accompanied it would all be available by the end of the month in March.  Intel however would not push for the release of their 850, 866 or 933MHz parts in spite of the fact that they had already released the 1GHz Pentium III, and even then the 1GHz Pentium III was released in "limited quantities."  It wouldn't be until another quarter later that the Pentium III would be available in greater numbers. 

This was truly the birth of what Intel became notorious for in Y2K, the Intel "paper-launch".  It was obvious that the early release of the 1GHz Pentium III was no more than a marketing tactic, a paper-launch to say: "we have a 1GHz CPU too."  It is interesting to note that just one year prior to this AMD would have been listed as the paper-launching CPU manufacturer, while it would have been unheard of that Intel would do such a thing. 

Later in March, Intel released their 850 and 866MHz Pentium III processors.  The launch was very quiet, for the most part because the public had already seen their 1GHz paper-launch and wasn't very interested in anything slower. 

In search of the perfect chipset

A very big problem for Intel throughout the beginning of 2000 was that there was no desirable 133MHz FSB chipset for use with the latest Pentium III processors.  The i820 was too expensive because of its RDRAM dependence leaving only two real Slot-1/Socket-370 chipsets as available options: the Intel BX and the VIA Apollo Pro 133/133A. 

The Intel BX chipset gained a sort of re-birth in the early months of 2000 as users started to discover its ability to run at the 133MHz FSB with a decent reliability record.  The BX platforms that have been popping up since 1998 had finally matured to the point where the 133MHz FSB was a viable option, making running the latest BX boards at 133MHz the latest fad.  The BX133 as it was unofficially called was what the market wanted unfortunately Intel never delivered quite that, so the enthusiast community took it upon themselves to make what they wanted out of what was available. 

One of the major reasons that the BX133 platform was so successful was because at this time, VIA's Apollo Pro 133/133A chipset was not very mature at all.  As we found out in our February 2000 Roundup of VIA Apollo Pro 133/133A boards, these boards still had a lot of maturing to do before they were fit for use.

With the VIA and i820 platforms both disappointing for different reasons, the BX chipset seemed to be the only sensible refuge that existed in the desktop PC chipset market.  Sadly enough, the newest and supposedly the best chipsets that were out in Y2K were being trampled by the two-year-old BX chipset.  Sometimes newer just isn't better.

This put Taiwanese motherboard manufacturers in a horrendous situation.  Intel's flagship chipsets, the i820 and i840 weren't selling well at all, yet Intel was breathing down their necks to push their Intel based product lines more than anything.  With the threat of the BX chipset supply drying up, it was a very nervous time for motherboard manufacturers.  Their only escape was found in VIA, a bedmate that Intel obviously didn't approve of since more VIA boards meant potentially more Athlon motherboards. 

VIA picks up some extra baggage

VIA definitely had it good in Y2K.  Courtesy of Intel's screw ups and AMD's dependency on them to produce chipsets for the Athlon, VIA had the entire desktop PC processor market eating out of their hands.  VIA, historically a low-cost manufacturer was all of the sudden being thrust into the performance market segment and adapted quite well. 

One thing was clear however, VIA needed to return to their roots as a value market segment provider since they were drifting away from that territory.  In an attempt to build the foundation for their next-generation of value PC chipsets, VIA acquired the graphics division of S3.   To make a long story short, VIA had just gained the rights to use the Savage4 and Savage 2000 cores in future chipsets. 

We cringed at this, not only did both of those solutions have some of the worst driver incompatibilities and support, the performance of both solutions was definitely sub-par however for the value PC market segment the performance isn't really an issue. 

Intel issues more recalls

As if things couldn't get much worse for Intel, a chip that was loosely associated with the i820 chipset was recalled.  This particular chip, known as the Memory Translator Hub or MTH for short, allowed for SDRAM to be used with the previously RDRAM-only i820 chipset.  It was also theoretically possible to use the MTH on i840 boards however this was not done very frequently. 

In May of 2000, due to problems discovered with the MTH, Intel issued a complete recall of their MTH.  Intel's fix for this was to exchange all of their boards for regular i820 based boards with a stick of RDRAM.  However in many cases, vendors implemented variations on this policy from not accepting the returns to allowing the users to return their boards and purchase Apollo Pro 133A based solutions instead.  The recall of the MTH was the final nail in the i820's coffin; Intel's flagship chipset was a failure. 

Later in May Intel launched their 933MHz Pentium III, two months after the launch of the 1GHz part. 

While we didn't hear much from AMD/VIA, the public was given an early warning of the storm that was soon to come.  In May VIA announced that they would be renaming their upcoming Socket-A chipset from KZ133 to KT133 out of respect for those that recognized a dark meaning behind the initials 'KZ', referring to the concentration camps during World War II. 

The release of the first Socket-A boards and thus the first Socket-A CPUs was no more than a month away.

Computex Awaits

At the last major show, Fall Comdex 1999, Intel 820 based motherboards were everywhere and the Athlon platform was a rare presence on the floor.  By the time Computex 2000 came around in June, things had changed dramatically.  Before we get to the obvious of what made it and what didn't on the show floor, there were a couple of announcements that occurred just before the show began.

On AMD's side, the launch of the third Athlon core since the introduction of the processor back in August 1999.  That core being the Thunderbird which offered a full speed, on-die 256KB L2 cache, allowing the Athlon to scale to its full potential as it would increase in clock speed.  This also gave the Athlon virtually identical performance to the Pentium III on a clock for clock basis. 

Intel had answered the pleas of their many customers and had finally launched a successor to the 2-year-old BX chipset.  The i815 offered PC133 SDRAM and 133MHz FSB support in addition to ATA/66 and AGP 4X compliance.  The only downside being that the i815 featured an often unused integrated graphics core in addition to its external AGP 4X slot. 

With those two announcements Computex was underway.  Taiwanese motherboard manufacturers were very happy with Intel's launch of the 815 chipset as it would finally allow them to push a line of motherboards based on Intel chipsets that they could actually sell.  The i815's launch definitely came too late and by this time Intel had already lost a considerable amount of the market to VIA, however it's better late than never and the i815 still turned out to be quite a hit. 

We saw i815 boards everywhere in addition to Socket-A motherboards for AMD's recently announced Athlon using the Thunderbird core.  What we didn't see were many i820 motherboards that crowded the floor at Comdex just 7 months earlier. 

The market had seen and rejected RDRAM as a memory solution; hence the failure of the i820 chipset, however there was still a demand for a higher bandwidth memory technology.  Double Data Rate (DDR) SDRAM was plastered all over the Computex floor courtesy of VIA, but there were no demos of any VIA DDR based solutions at Computex.

Instead, ALi, a company we had heard very little from presented a DDR solution for the Socket-A platform that they would have available before the year's end.  While ALi didn't have the time to make a major impact in 2000, it will be interesting to see if they do stand a chance in 2001.

In another return from the shadows, SiS announced their 730S chipset, a value PC segment solution for the Socket-A platform.  This foreshadowed the release and the tragic flaw of the next major announcement in June.

AMD's Duron: A processor without a home

Before the end of June AMD made another release that drove the stake further into Intel's heart, or at least was intended to.  A value PC derivative of the Athlon core, uniquely named the Duron, was released in late June. 

The processor seemed perfect.  Not only did it deliver, on average, 90% of its bigger brother's performance, but also it was destined to be an extremely cost-effective solution.  For most hardware enthusiasts, the Duron became an instant hit; however as a competitor to the Celeron, it could not win. 

Intel's Celeron, throughout 2000, was crippled by a 66MHz FSB even while reaching clock speeds more than 11 times higher than that.  The Duron easily outperformed the Celeron and was often times much more affordable as well.  The tragic flaw that kept the Duron grounded while the Celeron continued to rack up sales however was that there was no cost-effective platform for the Duron like the i810E was for the Celeron. 

This simple oversight cost the Duron quite a few sales and although it did catch on quite well in Europe it was still severely hurt by its lack of a home. 

VIA's KM133 and SiS' 730S would both prove to be solutions to this problem, but as you can expect, both solutions hinder the Duron's performance considerably.  In the end, you get what you pay for. 

Just weeks later, it was discovered that the new Socket-A processors were able to be overclocked much like the older Slot-A Athlons, through multiplier manipulation.  With a slightly different twist and proper motherboard support, the Duron and Athlon became an even greater value.  Maybe you don't always get only what you pay for.

Desperate times beget desperate measures

With rumors flying that AMD will be releasing a 1.1GHz Athlon very soon the pressure, once again, was on Intel to keep up with AMD in terms of clock speed.  Unfortunately, the Pentium III core was running out of steam and the Pentium 4 launch was still months away. 

Eventually the decision was made to launch the Pentium III at 1.13GHz at the end of July, a decision that would continue Intel's streak of bad luck throughout Y2K.  After a highly investigative report published primarily by Dr. Thomas Pabst of Tom's Hardware Guide, Intel was forced to recall the 1.13GHz CPU, as it was not fit for mass production.  Intel had taken a gamble in producing the CPU and lost. 

The repercussions of this were not huge since only a hundred or two of these CPUs actually got out, however Intel did feel the pain of this decision since the community as a whole realized the 1.13GHz launch for what it was: another completely marketing driven decision. 

To top things off, the 1.13GHz Pentium III was recalled the same day that AMD launched the 1.1GHz Athlon.

The month came to a close for Intel with their Fall Developer Forum, where Intel's NetBurst Architecture, the micro-architecture behind their upcoming Pentium 4 was discussed in great detail.  While we couldn't say anything on the issue, we knew the Pentium 4's exact performance data at this time as well.  Intel was definitely not going to go out with a bang in Y2K. 

VIA grows into a leader

By this time, two very interesting things were taking place.  On one side, the first generation of KT133 based motherboards for AMD's Socket-A CPUs were surprisingly mature.  VIA seemed to have gotten at least part of the formula for a successful chipset launch right as we trusted this platform enough to outfit six of our own web servers with KT133 based motherboards. 

We also started to see dual processor motherboards based on VIA's Apollo Pro 133A chipset appear.  We thought we'd never see the words VIA and dual processor used in the same sentence; our little VIA has definitely grown up. 

Click to Enlarge

Closing out the Year: DDR and Pentium 4 provide nothing