Geeks declare war on Intel

Chip-heads say flaws in the Pentium 4 prove the high-tech giant is sacrificing engineering principles for marketing goals.

Published March 2, 2001 8:30PM (EST)

On Feb. 19, Intel launched a $300 million advertising campaign to push the new Pentium 4 processor. Amid the now-standard TV ads featuring blue-pigmented men in skintight body suits and the drumbeat of hype pushing the chip's supposedly super-fast 1.5 gigahertz speed, chip aficionados are also being bludgeoned by a stream of new Intel trademarks. The Pentium 4, we are being told, features Intel's "NetBurst" micro-architecture, which in turn includes neat and dandy stuff like "hyper-pipelining" technology.

And it's all in the service of helping your computer scream along as you immerse yourself in 3D gaming and MP3s and streaming video. What could be better? One would imagine that hardcore chip-heads would be ecstatic. But guess what -- the geeks are not impressed.

What's more, they haven't been impressed for quite some time with Intel. When Intel debuted the Pentium 4 on Nov. 20, 2000, the computer site tüplay.com commemorated the occasion by posting a list of Intel's top 10 sneakiest moves and greatest screw-ups of the Pentium era. And in the intervening months, a small cabal of geek-oriented Web sites have kept their sights focused on Intel, documenting every slip-up and deconstructing every piece of marketing jargon the company disgorges.

The gist of their attack on the Pentium 4 is straightforward: Intel is no longer a company where engineers make the decisions, say the geeks. The marketing department calls the shots, and in its effort to market a chip that will supposedly make your computer an Internet superstar, it is telling Intel engineers to do things that don't make much sense. Ultimately, that could lead to the undermining of Intel's current stranglehold on the all-important microprocessor market.

Some of the sites are relatively well known -- the infamous Slashdot needs little introduction. But other watering holes where the most knowledgeable computer geeks on the planet meet to distribute and discuss cutting-edge technology information include TomsHardware.com, Sharkey Extreme and Emulators.com -- a site that is both the corporate front for an emulation software company and a node for Intel-watching.

All these sites confront the quandaries of the computer age from a most original perspective: one that is informed. And while their conversations don't obsess on the financial bottom line that fills most news sites, followers of the computer-based economy could do worse than to heed their advice. It's a case study in how the Internet works -- on one side, you have Intel, one of the most successful high tech companies in the world, fully able and willing to spend hundreds of millions of dollars to brainwash consumers into believing whatever it wants them to believe. And on the other, you have a motley group of Web sites populated by itinerant geeks, armed only with knowledge and a burning desire to spread it.

It's hard to imagine how Intel (or ChipZilla, as the company is affectionately known), the world's largest manufacturer of silicon-based microprocessors, could have fallen into such ill favor with the geeks. After all, these are people who generally treat computers with more adoration than their significant others. And for years, Intel has been one of the industry's key innovators. How is it that the people who know most about computers seem to like Intel the least?

The most virulent anti-Intel faction is led by Darek Mihocka, founder and president of Emulators, Inc., a prominent emulation software company. When it comes to his computers, Mihocka is Old School. A 15-year industry veteran, he has worked continuously through the days of Atari, Macintosh and PCs, and still works with Atari emulation software to this day.

In the not-so-distant past, Mihocka was an Intel advocate. Now he's an Intel advocate in the same way Dr. Laura is a gay rights advocate. On his company's Web site, Emulators.com, he's twice critiqued the new design of the Pentium 4. "Intel's new flagship Pentium 4 processor," he wrote, "[is] an engineering disaster that will hurt both consumers and computer manufacturers for some time to come."

He blames the design flaws not on lack of ability, but on confused priorities. "Just like at Microsoft and just like at Apple," wrote Mihocka, "the marketing scumbags at Intel have prevailed and pushed sound engineering aside." He was so upset by the Pentium 4, he urged his Web visitors to boycott the company until Intel makes a better processor.

Intel spokesman George Alfs denies the accusation that marketing interests have affected Intel's engineers, but Mihocka is not alone in his assessment. Mihocka and other observers say that Intel's engineers have been asked to accomplish goals attached to fuzzy concepts such as "Internet SIMD Streaming Extension" and "NetBurst Architecture," -- the design names of the Pentium III and 4, respectively. But the mandates for these ideas are grounded in market research, not in the physics of semiconductivity.

Intel's Alfs says the new design name is a reflection of the chip's awesome power. "The Pentium 4 has 1.5 gigahertz and a 400 megahertz bus, what we call a system bus," he said, referring to a piece of chip hardware that connects the central processing unit (CPU) to the main memory chips. "We believe that this allows for a certain explosive power, or 'bursts' if you will."

Mihocka and other experts scoff. To them, the design names are blatant marketing ploys intended to capitalize on the Internet and MP3s. On Intel's Web site, the Pentium 4 is described as "technology for the Internet and beyond." Thus "NetBurst Architecture"; the idea that microchips can be designed specifically for the Internet, which some hardware-watchers consider a dubious claim.

One of the first to challenge Intel on its "NetBurst" marketing strategy was Thomas Pabst, CEO and founder of Tom's Hardware Guide. Pabst, a German-born doctor in his 30s who lives in the U.K., founded Tom's Hardware in 1996 with the goal of creating an independent, unbiased testing agency. Pabst developed his uncanny computer hardware savvy by disassembling and reassembling his old Intel 8086 CPU in the early 90s. Now he's one of the most trusted hardware gurus in the world.

When geeks want to know which toys to buy, they go to him. On the release date of the Pentium 4, he published a 20-page dissertation on what he dubbed "the most controversial ... processor on the market." Traffic at his site hit an all-time high, quickly scaling up to 1 million visitors.

In his professional opinion, the Pentium 4 isn't a bad processor, says Pabst. His article depicted it as a powerful chip with extraordinary room for growth. But he also observed serious flaws, and expressed dismay at Intel's inability to live up to its own hype. "Whatever Pentium 4 is right now," says Pabst. "It is certainly not the greatest and best performing processor in the world."

He has a particularly hard time understanding what Intel means by the term "NetBurst architecture": "Believe me, your Web pages won't pop up any faster, downloads will take just as long, and the Internet won't 'burst' either." He believes marketing is to blame for Intel's desire to lead consumers into thinking that the Pentium 4 is the premier Internet microchip. "Since the Internet is still hip, it is a perfect vehicle to market Pentium 4," says Pabst. "Again, 'NetBurst' shows its roots in Intel's marketing department."

The big surprise of 2000 was the fact that AMD, an Intel competitor with vastly inferior technical and financial resources, bumped Intel out of position at the top of the chip heap. When AMD launched its 1.2-GHz Athlon processor last summer, it opened a huge can of silicon whoop-ass on Intel's 1.13 GHz Pentium III. That was also about the time Intel's stock price went scuba diving, plunging almost 60 percent. Since then, Intel has been obsessed with striking back.

Part of its counterattack has included an obsession with increasing its megahertz stats. Meanwhile, AMD is moving more sedately, probably because it is aware that more megahertz does not guarantee a faster processor. More megahertz translates into greater speed within the same processor, but different processors react differently to different stimuli. Hence, a 1.2 gigahertz processor can outperform a 1.5 gigahertz processor, depending on what it's being used for, or if it has a more efficient design for a given task.

Which, in the case of Athlon vs. Pentium 4, is exactly what's happening. According to the Microprocessor Report, an industry trade journal, the Athlon performs between 5.3 and 12 percent faster than the Pentium 4 on typical systems.

In the quest for the Holy Grail of megahertz, Intel neglected other areas of the chip's design that have affected its overall speed -- to the point that the Pentium 4 has actually lost ground to the Athlon and even the Pentium III.

Pabst and other hardware geeks say that Intel's engineers were asked to place disproportionate emphasis on aspects of chip design that catered to the goals of mucho megahertz and the bazooka-like bursting system bus, and they were willing to forsake anything that got in its way. It's made for an awkward processing design -- and, for the most part, a majestically unpopular final product.

For example, to accommodate its goal of a massive bus and extended "hyper pipeline," Pentium 4 engineers were forced to reduce the size of something called the L1 data cache -- a piece of hardware used for storing and processing frequently used information -- to a meager 8K. This shrunken size is half of the Pentium III's, and an eighth the size of the Athlon's.

The diminutive cache size has had, according to Darek Mihocka, "crippling effects" on the chip's ability to handle basic office software. Which is why the Pentium 4 has lagged behind the Athlon and the Pentium III on most basic office software benchmarks.

Intel doesn't see this as a problem. "Once you get past one gigahertz, you're going to have more than enough power to write your letter to Grandma," says Alfs. "We wanted to focus on some of the more exciting aspects of processor design." This means streaming media, gaming, MP3s, video encoding and the like. These are the areas where the Pentium 4 does beat the pants off any other processor.

In that vein, the Pentium 4 is a great processor. But it's not what Intel claims it is: the best in the world. PC World, the New York Times and sharkeyextreme.com all agree with Mihocka and Pabst's estimation that the Pentium 4 is not the same quality processor for your buck as AMD's Athlon. The Microprocessor Report also awarded its Processor of the Year 2000 prize to the Athlon.

In fact, the Pentium 4 has grabbed only an anemic 1 percent market share in its first couple of months (as compared to 10 percent for the Pentium II and III on their debuts). It's possible that the Pentium 4's bloated price has something to do with its lack of success -- it retails at two to three times the price of its comparable rivals. Overall, Intel still maintains its stranglehold on the overall processor market, floating around a 90 percent market share.

Could the skepticism expressed at sites like TomsHardware explain Intel's failure to hit a home run with the Pentium 4? Hard to say. What is for certain is that if you care about finding out the truth about a certain chip, the Internet is there to help. There are plenty of computer lovers, like Pabst and Mihocka, who know their stuff and are more than willing to share it with the rest of us. You can rest assured they'll be typing away furiously on their Web sites every time Intel, AMD or anyone else does anything computer-related worth noting.

The only question is whether the rest of us are willing to listen.


By Kieran McCarthy

Kieran McCarthy is a freelance writer.

MORE FROM Kieran McCarthy


Related Topics ------------------------------------------