Posted on 07/26/2002 11:11:31 PM PDT by JameRetief
Why the Buzz on Hammer Just Won't Quit
Just in case you haven't been paying attention, AMD's upcoming 8th generation processor has been generating a steady buzz of discussion almost since the first of the year. This has led to an occasional outburst of exasperation from editors at various websites, some of whom have complained the web community in general is too focused upon Hammer. Hammer is, their argument goes, an unproven product with a distant and uncertain launch date, and should not be focused on so heavily. It's also been suggested by some that much of the Hammer hype itself has been driven by fanboy enthusiasts with little true understanding of the hardware market in general, or by websites willing to serve as PR engines for AMD's marketing machine.
The problem with the above argument, however, is that it only explains the theoretical behavior of a single website, not the general discussion of Hammer that's occurred across dozens of sites and throughout the web in general over the last eight months. Unless the Internet and technical sites have become overwhelmingly populated with anti-Intel bigots (a conclusion which is both insulting and idiotic) there must be other reasons for why Hammer has caught the technical community's interest. Rather than discuss the biases or honesty of any particular site, therefore, we've chosen to look into the greater issue: Why do both Intel and AMD users keep talking about K8?
To understand the community interest in K8, it's necessary to look back to 1999 and the launch of the K7. The Athlon looked good on paper, but there were plenty of reasons to be skeptical about how the shipping product would perform. While AMD had achieved some measure of success with the K6 and K6-2 line of processors, the K6-3's were scaling poorly and the Socket 7 platform was running out of steam. The company also had a reputation for promising much more than it could actually deliver and suffered notoriously from manufacturing problems and delivery issues. As Athlon neared launch, no one quite knew what to expect. Would K7 deliver—or would it sink like a stone?
Luckily for AMD, Athlon not only delivered, it delivered in spades. Despite some initial compatibility issues, the K7 became increasingly popular for its low price, high performance, and intermittently higher clock speeds (the last depending on whether Intel or AMD happened to have released a chip that week). Athlon adoption was further boosted when Intel began having major yield problems with high speed Pentium 3's. This, in turn, forced high-end Pentium 3 prices up even more, and made the Athlon an even better bargain. Intel's release of the i820 motherboard and its reliance on amazingly expensive RDRAM made an Athlon-equipped PC an even more cost-effective option—particularly at the high end, where, starting in January 2000, AMD began shipping 1 GHz chips, which Intel couldn't ship in volume until much later in the year.
What's important to see about the events of late 1999 / 2000 is that AMD's aggressive launch and Intel's stumbles over the same period of time both combined to produce an environment particularly conducive to K7 adoption. While the K7 would've been a success regardless, Intel's combinations of motherboard and CPU difficulties created vulnerabilities both AMD and VIA were quick to exploit. Whether you were an Intel or an AMD fan, late 1999 was an exciting and unpredictable time—and the entire cycle was kicked off by the launch of the Athlon.
Fast forward to the present day, and the parallels become easy to see. Again we see AMD preparing a new core at a time when Intel seems on the verge of decisively establishing its superiority over AMD's best offering. Just as the P3 left the K6-2 and waning K6-3 behind, the P4 has slowly but steadily overtaken the Athlon, and while the upcoming Barton core could give the Athlon a temporary boost with a potentially faster FSB and a dramatically increased amount of cache, its unlikely to provide the chip with the power it needs to catch or surpass the P4—at least not for very long. Athlon's days as a competitor in the high-end arena are numbered, and the Sunnyvale-based company knows it. For months it's been made clear: AMD's future depends on K8.
This time, however, the stakes are even higher. Hammer isn't just AMD's next-generation 32-bit processor to compete against the P4; it's also the company's first 64-bit capable part. The x86-64- implementation AMD has chosen, moreover, is entirely different from the IA-64 technology Intel is backing, making this the first time Sunnyvale has ever challenged Santa Clara for control over such fundamental computer standards.
The idea that AMD could even hope to succeed with such a challenge would seem ludicrous in 1999, but we're headed for 2003, and Intel isn't laughing. After watching the i820 tank, the P3's inability to scale past 1 GHz, and the P4's utter failure to impress at launch, Intel learned their lesson. Now we see chipset development back on track, the P4 is scaling like a bullet, and Intel's looking for blood. Not only has Santa Clara announced plans to ramp the P4 to 3 GHz by the end of the year, but they've recently launched Itanium 2, and dedicated tens of millions of dollars to developing software for it. From desktop performance to high-end servers, Intel is pouring on the speed and the cash; determined to neutralize K8 before it even has a chance to become a threat.
Even in the face of Intel's aggressive ramps, however, AMD isn't blinking. The Sunnyvale-based company has come a long way from 1999, when 2nd and 3rd tier motherboard manufacturers shipped Slot A motherboards in unmarked white boxes and AMD was unknown as a high-performance competitor. Since then, AMD has seen its share of the CPU market jump to from approximately 5% to 20%, successfully entered the server, business, and mid-range mobile markets, and has dramatically increased the visibility of its partnerships with key manufacturers. Far from keeping potential K8 solutions in the dark, manufacturers have been prominently displaying them all year.
Of course, ultimately, everything is resting on Hammer's ability to perform as advertised, but even there, early indications are positive. Hammer is, essentially, a super-charged Athlon, with an integrated memory controller for vastly decreased latency, possible larger caches, a slightly longer pipeline for superior scaling, and a faster front-side bus speed, as well as, of course, support for 64-bit x86-compatible computing. Overall performance is expected to be between 15-25% higher than the AthlonXP at an equivalent speed, which could make K8 an extremely attractive product. Later on in the year, we'll have Opteron, AMD's new 64-bit server processor aimed directly at Itanium's market.
It's easy to see why people are talking about Hammer, and the reason has nothing to do with biased editors or enthusiast websites. On the one hand you've got Intel with its billions in capital, immense manufacturing capabilities, the rapidly-scaling P4, and an oft-stated determination to make Itanium a success, no matter the cost. On the other,
AMD has a highly successful core design in the K7, far stronger relationships with manufacturers and vendors than ever before, and such a strong belief in their upcoming K8 that they've hung the future of the company, from mobile systems to big iron, squarely on its performance.
The reason both Intel and AMD fans are talking about Hammer is because its exciting, no matter which company you believe has a better approach to 64-bit computing, or a stronger and more-proven technology. It's not often, even in the fast-paced technical world, that we see computing history being made, but the struggle between Intel and AMD over x86-64 vs. IA-64 could very well be such an event.
Neither side has a foolproof strategy, neither company an unassailable position. Both are watching the other for any sign of weakness, any hint of distraction. At stake is the future of CPU design, and which company will profit most from it. One company stands to lose an immense amount of face; the other could lose every hard-won inch of progress made over the last four years. Place your bets, ladies and gentlemen, place your bets. P4 / Itanium or AthlonXP / Clawhammer?
My only regret is that I can't sell tickets.
I don't know if anyone is up this late but we'll see:
To find all articles tagged or indexed using tech_index
Click here: tech_index
Combined with being iterations of the same hardware (less production diversions), the hammer lines will be able to scale well, thanks to the pipeline changes, and be even more competitive in the raw computer performance, thanks to a variety of innovative technological achievements.
From the roadmaps that I have seen from Intel, nothing is looking promising. And unfortunately for Intel (not for consumers), they let AMD gain a considerable foothold and gain credibility. Intel's FUD machine isn't as able to stigmatize AMD as the low end anymore.
But fortunately, Intel does have the fab (chip fabrication facilities) advantage. They will have to rely on volume of supplies of their chips and hope that AMD will not be able to deliver high quantities of hammer chips from their fabs.
There are other reasons why I think that AMD will be the market leader for the foreseeable future when hammer is released, but the above are good indications.
What are we going to do with all of the processing power that is headed our way?
I can't fault Intel's chips either. I'm just glad to see serious competition.
Intel's attitude about microprocessors was similar to the NEA's attitude on government schools.
Intel - You can use any chip you want as long as it is an Intel chip.
NEA - You can go to any school you want as long as it is a government school.In both cases, competition is good.
While the AMD does give a slightly better performance at equal speeds (especially gaming aps), I had to choose the stablilty this time around. But, if AMD can rebound, I wouldn't hesitate to buy another CPU from them...
But the upcoming Hammer lines will have the chipsets integrated with the CPU, so the chipset issue won't be a problem.
my (celery 633) box isn't worth jumping to an athlon, but the amd site has durons for dirt cheap ($64)...
Depending on where you buy, you can get a much more powerful Athlon for a few dollars more than the Duron. If you do a search on Pricewatch.com, you can find even better deals than the one you have listed. Never go by the official price listing from AMD. The prices you can get them for in the market is alot less. You can get an Athlon 1.33GHz 266 FSB for $62, only $7 more than the $55 Duron 1.3 Ghz. And if you get the Athlon 1500+ (an upgraded version of the 1.33 Ghz listed above), it's only $63.
Obviously, you will want to search through a few sites to find a company you would trust to order from, but you shouldn't have to pay much more. Either way, you will be better off spending $10 or $20 more for an Athlon rather than a Duron. It makes your purchase more future proof.
I also have very good reason to believe IBM is going to ship eServers with the Hammer,
However I remain frustrated that we are unable to purchase workstation quality Athlon systems from a top tier vendor, such as HP or Dell, however my daughter doesn't seem to have a problem with the situation as we are paying her $50 a box to assemble Athlon systems and upgrade older machines.
Basically, the articles contents described why AMD can undersell Intel, spend less on R&D, and not make as much money in general and still be able to gain marketshare and be able to satisfy their profit needs.
It was a well researched and in depth article. I will see if I might be able to find it.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.