|the soapbox @ benturner.com|
"Intel's Celeron: The Blunder or Wonder Chip?", an Essay
There is little question that Intel, the processor chip giant of the computer industry, has bullied and pushed its way into near complete market domination of its competitors. Intel's managed to stay one step ahead of companies like AMD and Cyrix, releasing the fastest chips to the public long before anyone else does. Intel has its Pentium chips in most of the PC market and a small part of the server market, and it has instant name recognition. In short, it has needed little effort to remain where it is once it got the Pentium chip in the computers of school computer labs, workplaces, and the gaming market.
So enough with the stuff we all know. Intel doesn't need explaining. Only recently have the companies like AMD and a distant Cyrix put pressure on Intel. Intel rode the wave of the rapid technological breakthroughs in the computer industry. Intel was far ahead of its competition and rarely had to consider cheaper prices and several different architectures to sell to different users. But now that time and technology are catching up with Intel, things are changing.
The Pentium II line of chips is very expensive. It starts at around $200 for a P2-333 and goes up to $700 for a P2-450 or so, much much more for the Xeon line of Pentium II processors. $200 is not so bad as it used to be, but slower Pentium IIs are becoming outdated and their prices are dropping.
Intel found, after time, that it was losing business to the other processor companies, as AMD sold its K6 line which served as a serviceable alternative to Pentium chips, for a much cheaper price. AMD did not compete with Pentium in the gaming department, however, with its much slower 3d processing. Why? The AMD did not concentrate on the floating point unit, the part of your processor which cranks out non-whole numbers for things like 3d modelling and for 3d gaming. It's the AMD floating point unit that stops it from matching pound to pound with Intel. But AMD found out that low-end computer users would be willing to pay for a cheaper chip to run their basic applications on. The fact that Intel was not the designer of the chip did not do much to hurt its success. Intel had something to worry about.
So Intel was forced to release processors which appealed to different markets. Its MMX boards, which had extra support for multimedia-based work, were geared towards audiophiles and gamers. The Pentium was replaced by the Pentium II and the PII served as the best of the best -- with MMX support inside. And then there was the Celeron, a processor Intel released without any L2 cache, allowing its cost to be substantially lower than the Pentium II.
L2 cache stands for Level 2 cache. Basically, it works along with RAM to store frequently repeated instructions so the computer could search the cache first and then search through all the RAM to find what it was looking for. It speeds up applications that have redundant runtimes. It really only affects office-based applications like Excel, but can also affect how your computer handles input and output (keyboards are input, monitors are output, for example).
Intel was probably thinking pretty highly of itself. It came off as being more flexible for users, offering different kinds of chips for different people. It came off as being cost-friendly, releasing high-end CPUs and low-end CPUs. It had products with the Intel name to compete with whatever other companies could throw out on the market.
But its Celeron chip failed miserably. The lack of L2 cache meant that the system would take a performance hit, particularly in office-based applications. Although it was cheaper, which companies would buy an inferior product? The Celeron had the same architecture as the Pentium II, but no one cared because of all the negative articles and exposure. Certainly no gamers were using Celeron chips. About the only people with Celerons were the people who got it first, and the people who really couldn't afford to spend a heap of money on a new chip.
It would seem to most everyone that Intel made a big mistake releasing the Celeron. Why buy a crippled Pentium II when you can get an AMD K6 that only performs slightly worse than a real Pentium II?
Intel reacted quickly and began work on the next line of Celerons, chips this time having L2 cache. Technology web sites were skeptical, even more so when Intel announced the next line would have the same product name. Why make a sequel of a product that failed big time?
The Celeron was, and still is, extremely unpopular among the reviewers and hardcore computer users. They've dismissed it as slop, not worth further investigating. People looking for information on what to buy of course read what these experts say and believe them. Not a bad way of going about things at all. But this would prove bad later.
While the Celeron fiasco was taking place, some other interesting events were taking place among owners of 3d video cards and CPUs. A practice called "overclocking" became the new thing to do. What it involves is basically increasing the power going into your hardware so that it has a higher clock speed -- in other words, people were making their 200MHz computers overclock up to 233MHz. More hardcore people were overclocking their video cards too, for better performance in gaming.
Intel, you can understand, was not pleased with people getting more out of their chips than what they paid for. Intel had been overclocking chips for awhile, since it has to test all its processors to see how fast they'll go. Remember the 150MHz processor? Well, all that was was an overclocked 133Mhz processor, according to Tom of Tom's Hardware. To make sure that all the processors it made would get it the most money from customers, Intel tested its CPUs to see if they could go faster and faster. Not all processors that are made are the same, you know. Processor companies throw away huge quantities of chips which didn't work.
Intel took steps against overclocking by releasing chips which didn't overclock as well, and were crippled for overclocking, since they didn't allow people to mess with multiplier settings, the key behind overclocking.
Thus the Celeron didn't receive much attention from the overclockers, for this reason. The public had already cast out the Celeron and no one cared to even hear about it anymore.
Computers have thundered to faster and faster speeds, and now even the motherboards and memory have to be upgraded if you buy new processors. It used to just be that you could buy a processor upgrade, get it installed, and you'd be done. Now that all parts in a computer are operating faster and require more wattage and speed to run, upgrading from an old chip to a new chip involves buying a whole new system, basically.
If you weren't turned off by the technical details earlier, you will be now. Motherboards have been running faster than they used to, and now they clock up to 133MHz, from the 66 and 75MHz motherboards of only a year ago. The motherboard is the part of the computer which acts as the brain center of the system. You plug in all your components into your motherboard and it integrates all the action. Motherboard makers have released boards which clock faster and which still allow older chips to work on them.
Intel locked its chips so that they could not be overclocked. But part of overclocking depends on the bus setting (basically, the MHz of the motherboard I talked about above). A computer running at 66MHz on the bus might have a multiplier of 3.5 on the processor chip. What Intel locked was the multiplier. So 66 multiplied by 3.5 would be 231MHz, the speed the processor would be running at. Now imagine if the bus speed was increased to 133MHz. 133 multiplied by 3.5 would be 465.5MHz. Yes, I think this example is impossible, but I'm merely demonstrating what's involved here. Lower bus speeds might be more realistic for that example, but...that's not the point.
Here was a workaround to thwart Intel. :)
Recently, a lot of gamers, particularly those who played Quake 2, found out about people who were overclocking Celerons with the higher bus speeds. This group of people began looking seriously at what the Celeron could actually do. Sure, a Celeron running at 266MHz was unimpressive. But how fast could it be if it were overclocked? Some folks had managed to overclock Celeron 266MHz computers up to 496MHz in their private tests! And the reports and benchmarks had shown that, at these higher speeds, the Celeron was competing with the Pentium II in gaming. The lack of L2 cache on the Celeron hindered it a bit from reaching the Pentium II's office-based results, but it was still competing with mid-range Pentium II chips.
Bulletin boards and IRC chatrooms were feverish with getting more details. A lot more people were buying Celeron 266s and finding that the Celeron could run quite happily at 448MHz, still a wonderful improvement from the default settings. What was best was that the Celeron 266 cost only about $100, based on its horrible reputation in the market. You'd be saving $200 to $300 on a chip with faster overall performance than a P2-350.
I bought a Celeron 266. Regardless of whether its results are slightly slower in office-based applications, it wouldn't be noticeable to most people. How often do you have to wait for Word to get its work done? It's usually waiting for you to do something, these days. I had to buy the suggested components and the right model of Celeron 266. I will see if this thing can get up to 448MHz. I hope so. This chip is a steal.
Intel was lucky that people didn't find this out sooner, or else Celerons would be far more popular. The Celeron has such a negative image that people don't even consider it, and it's easy to be skeptical of an overclocked Celeron's benchmarks when everyone in the industry has claimed how terrible it is. Intel lost a bunch of money on the Celeron, I'm sure. But it could have been worse if people found out that the Celeron could be set up to the power of a Pentium II.
Intel just released the next line of Celerons that I talked about earlier, the Celeron with a small amount of L2 cache. It's not as much cache as a real Pentium II, but it's more than the original Celeron. The verdict is still out on the new Celerons, but most people figured Intel would try to handicap the new Celerons for overclocking. Tom's Hardware, however, overclocked a prerelease of the new Celeron to very tempting speeds.
So what is Intel thinking? Will it cripple the new Celerons so people can't overclock them? It would mean, for them, that people would have to buy Pentium II's, thus making them much more money. But Intel cannot deny that the Celeron easily holds the place for the most overclockable chip in history, and maybe it will allow people to overclock its new Celerons if that means people continue to buy Intel instead of the highly competitive, more cost efficient AMD K6-2 chips.
Obviously, the Celeron turned a lot of heads one way or another for Intel. Most say it was a strategic disaster for Intel, but it turns out to be a terrific steal for those who are savvy enough to overclock. How can a chip that's so good be considered so bad? This sort of thing is not common in the computer industry -- products usually are recognized to their full potential.
Maybe it signals a crack in Intel's walls. They let this chip get by. All in order to buy out the market for cheaper AMD and Cyrix chips. But AMD and Cyrix are not looking to top the Celeron -- they're looking to top the Pentium II and everything else soon to come from Intel. When AMD and Cyrix finally get their products up to Intel speeds, you'll be able to buy one of their chips with Pentium II performance for a little more than the price of a Celeron. Intel's strategy has failed with the Celeron. Intel hasn't reacted well to the growing pressure from competitors, and it could just be that the company may lose a lot of ground when the next generation of chips comes around.
Intel has no revolutionary products coming out in the near future, but AMD and Cyrix have improved their chips dramatically, almost to the point where consumers would be stupid to ignore buying them. Intel was not even one of the first companies to announce its work on the new copper style of processors. Copper will be replacing silicon on processor chips soon, and since copper's thermodynamic properties allow it to be heated to higher temperatures without becoming unstable, it will allow processors to get past the 1000MHz barrier. Can you imagine that? Intel has been slow to work on the new technologies, and that's where I see it falling down to just another competitor in a sea of many.
Intel's also beginning to feel the Microsoft curse, the problem of being labeled a monopolistic company that pushes other competitors out of the scene to get what it wants. AMD and Cyrix are the underdogs, the morally acceptable alternatives to buying processors from the "Empire". Imagine Mac loyalists, but not quite so rabid. These people refuse to buy Intel, so they get AMD boards for cheaper prices and only minimal performance hits. I personally want to see there be more competition in, specifically, the processor market, and I want AMD to gain more market share. But I bought Intel this time around. Another generation and AMD will be one of my final choices, I'm sure.
So what's the point of this Soapbox? I just thought it was interesting to describe what I've observed going on in the processor market. Intel is losing its stranglehold on everyone from gamers to corporates, and the Celeron will never receive the positive attention it deserves. No credible news site will report on how people are snatching up old Celerons and overclocking them up to Pentium II levels. Overclocking is still only for the geeks, and not for the regular Joe, and I guess that's why this shift in opinion over the Celeron has not made it to the headlines.
But it's still there, and there's a small group of people who know about this and have taken advantage of it, and I've seen it unfold.
This is how cruel the industry is -- you get lazy and you die. You try to scam the customer and you eventually get scammed even worse in return. Even if you're Intel, if you're sitting on a wonder chip and you fail to realize it, you're bound to lose ground against your rivals.
The customer wins again.
|RATE THIS SOAPBOX|