Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?

To: richard surckla who wrote (128882)	11/20/2000 2:19:36 PM
From: tejek	Respond to of 1583404

From the Motley Fool: ____________________________________________________________ Why Should AMD Drop Mustang? Email this page Format for printing Become a Fool! Discussion Boards Advanced Micro Devices Intel By DeepNorth November 20, 2000 The following perspective was written by a member of the Fool community and regular contributor to our discussion boards. The opinions of the author are not necessarily representative of the opinions of The Motley Fool or its editorial staff. To beat Intel (Nasdaq: INTC) to the marketplace and take market share, Advanced Micro Devices (NYSE: AMD) needs to balance maximum profits with competitive advantage until its Hammer chip can be leveraged to dominate the processor market. Plotting the roadmap appropriately is one of the techniques at the company's disposal, and one of its competitive advantages over chief rival Intel. AMD management has executed brilliantly over the past year, and there is no reason to expect that this will change. The decision to drop its Mustang chip [The Wall Street Journal, 11/14/2000] appears to be a wise one, a move that will maximize both profits and competitive advantage. AMD should be attempting to maximize profits over about a five-year horizon. This means working out a series of max/min problems with respect to demand, "mind-share," die sizes, and competition. It also means maximizing overall competitive advantage to the extent that AMD's competitors make mistakes such as incorrect allocation of resources. Certainly, the fact that AMD can and does alter its roadmap so dramatically has an unnerving effect on the competition. What are they thinking at Intel HQ? Clearly Intel is in trouble and has been for about a year. It has two new technologies: Pentium 4 and Itanium (dubbed "Itanic" by critics). Both appear to have failed.¹ What now? Intel has to try to protect its market share from AMD, but how? Currently, it doesn't even have a clear idea of what it's facing. Strategically, AMD is putting the squeeze on to completely hijack the 64-bit arena. That is the obvious long-term goal here. When Hammer becomes the de facto 64-bit standard, AMD wins all the marbles and begins to switch market cap with Intel. Tactically, AMD is beefing up its war chest and taking market share from Intel. How best to do this? By taking market share in the "sweet spot." The sweet spot is where maximum profits (a combination of volume, ASPs, and cost) lie. Is AMD in the sweet spot? I think so. Is a Mustang release a good way to maximize this sweet spot? I think not. Apparently, AMD management agrees. There is a very real competitive threat from Intel. How best to disarm it? Place Intel under short-term tactical pressure and give it as little information as possible to respond strategically. Right now, Intel is losing the tactical battle. AMD's goal is to make it lose the strategic war. Intel cannot respond as quickly as AMD. By keeping it in the dark about its medium-term strategy, AMD hampers its ability to develop an effective counter-strategy. AMD has the advantage of being nimble and focused, but the disadvantage of having limited production capacity. By leaving Intel as much in the dark about the future, AMD forces Intel to commit capacity (where it has the advantage) before it knows what AMD will mount as a competitive threat. Intel is stumbling because it is being pushed too hard. The recall of the 1.13GHz Pentium III is eloquent testimony to this. The Pentium III is already being pushed some 20% or more beyond its proper engineering limits for the current process technology. This is obviously because Intel does not have any other answer to the current AMD Athlons. Is the Pentium 4 an answer? Short-term, no. Medium-term, maybe, but this is where the silence about the roadmap comes in. Let Intel commit its capacity to ramping the Pentium 4 or to moving McKinley ahead. Then, AMD can either dominate the medium-term with higher-clocked Athlons or execute the coup de grace by simply switching the whole world over to the 64-bit Hammer. Obviously, execution is and always has been vital. However, the advantage always goes to the leader, and right now that leader is AMD. Does AMD have the ability to field a 1.5GHz processor in volume? I honestly do not know. Neither does Intel. To be honest, I really don't see any way that Intel can properly respond to the current situation, except by marketing the hell out of the Pentium 4, touting its clock speed and basically buying market share with subsidies. However, even that has limited short-term benefit, since it's pretty clear that even if Intel produces 20 million 1.7GHz Pentium 4s this quarter, there is no way to get them into shipping systems with motherboards and RAM. How many manufacturers or dealers will inventory 1.5GHz Pentium 4 CPUs waiting for system components? Not many. This quarter belongs to AMD. Next quarter, Intel might be able to recover, but needs a lot of unlikely events to do so. It needs RDRAM to ramp in volume and prices to drop. It needs working motherboards in volume. It needs to be able to produce the Pentium 4 in volume. It will have to produce Pentium 4s on the order of 1.7GHz to 2GHz. It needs AMD to stay at or near 1.2GHz. Since this is an all-or-nothing proposition, I would predict Intel will be losing next quarter as well. The second half of next year will be interesting, to say the least. I still think Intel will be struggling, but it's harder to call. If Intel can ramp enough fabs to produce 2-3GHz Pentium 4s in volume, switch to DDR, and AMD fails to exceed 1.5GHz Palominos in volume, Intel may regain the "sweet spot." I think that's unlikely. Even if Intel prospers in the second half of next year, it does so at the expense of fully committing to the 32-bit Pentium 4. This leaves a full ramp of the Hammer a completely open playing field. Again, back to the Mustang: keep 'em guessing and make 'em sweat. If Intel wants to know what our plans are, let them go down to the computer store and ask if there are any Hammer systems for sale. For the first half of 2002, Intel needs for McKinley to have such compelling performance advantages over 32-bit systems (and 64-bit Hammer) that it is worth doing a software rewrite for the majority of developers. My 20 years in the industry tell me the chances of that happening are virtually nil. AMD has been on-track for more than a year. Intel has been stumbling for about as long. Recent AMD road-map changes have not changed this status quo. __________ ¹There is considerable argument about whether the Pentium 4 has/will fail. Intel apologists keep saying that the (as yet unreleased) Pentium 4 "beats" the currently shipping Athlon 1.2GHz. What benchmarks we have show that, overall, the 1.2GHz Athlon, in fact, beats the Pentium 4 in 10 out of 13 benchmarks. Whatever. The point is, if there is even some debate about this, then the Pentium 4 has lost. The Pentium 4 should be unambiguously faster, clock-for-clock, than the Athlon. The fact that everyone agrees it is not, is proof of failure. Whatever the eventual volume shipments of the Pentium 4 compete against, it is a foregone conclusion that it won't be a 1.2GHz Athlon. If Intel continues to experience problems, the volume shipments of Pentium 4 might be competing against dual-processor Athlon descendants, or even the 64-bit Hammer. I don't believe that there is much argument that the "Itanic" is a failure, even from Intel. Will this processor ever be anything other than a historical curiosity?

To: richard surckla who wrote (128882)	11/20/2000 5:05:16 PM
From: tejek	Read Replies (1) \| Respond to of 1583404

<font color=green>Pentium 4 fails to close gap on Athlon, say testers By Michael Kanellos Staff Writer, CNET News.com November 20, 2000, 1:20 p.m. PT Intel's Pentium 4 chip released Monday doesn't provide a real performance advantage and is often slower when compared with the fastest Athlon chip from Advanced Micro Devices, benchmark testers and analysts said. The Pentium 4 doesn't seem to be worth its price right now, they said. Benchmark tests posted by review sites such as Sharky Extreme on Monday indicate that the 1.5-GHz Pentium 4 does outscore the 1.2-GHz Athlon on the "Quake III" game, some video and media editing applications, and relatively theoretical tests on memory bandwidth or scientific calculations. But when it comes to many real-world applications and games other than "Quake III," the difference is inconsequential. On a number of benchmark tests, the first version of Pentium 4 underscores Athlon and even the Pentium III. "For today's buyer, the Pentium 4 simply doesn't make sense. It's slower than the competition in just about every area," wrote Anand Lal Shimpi in Anandtech, a review site. The Pentium 4 "is not a body blow to AMD by any means," said Nathan Brookwood, an analyst at Insight 64. "The Athlon will be better for productivity applications, which is kind of funny because AMD has yet to establish a base in the commercial market..the personal productivity applications, the Pentium 4 is showing no benefits over the Pentium III." The relatively uninspiring results that accompanied the launch of the chip will be debated for weeks and set the stage for another rousing year of competition between the two companies in 2001. Because of the new architecture behind the Pentium 4, Intel will likely rapidly increase the speed of the chip. AMD, however, will be difficult to outrun. The product roadmaps, combined with the performance results, could also pave the way for AMD to finally get into the corporate world. AMD will have chipsets to make multiprocessor computers out in the first quarter. Intel won't be able to offer a two-processor solution for Pentium 4 until the second quarter. Workstation buyers tend to scrutinize benchmarks more closely and often demand multiprocessing. As a result, some major computer makers may shift from an all-Intel lineup to get a jump on the competition. "I'd be disappointed if AMD didn't get a workstation win," Brookwood said. "What AMD has lacked is dual processing." A major contributing cause of the flat benchmark results comes from the 20-stage pipeline of the Pentium 4.The pipeline is the processor's equivalent to an assembly line. At 20 stages, the Pentium 4's pipeline is twice as long as the one found in the Pentium III and longer than the 15-stage pipeline found in the Athlon. With a longer pipeline, data simply has to travel through more steps; and, if a mistake occurs, the processor has to do more backtracking. Although a long pipeline is internally less efficient, it allows designers to push the clock speeds faster, noted Dean McCarron, principal analyst at Mercury Research. The Pentium 4 is "a lot more forward thinking," he said. Intel, in fact, is already talking about goosing the system. A 2-GHz chip is expected to be out by the third quarter next year. Still, that will likely lead only to relative performance parity, predicted Anandtech. By then, AMD will have a 1.5-GHz Athlon. Part of AMD's ability to keep the performance gap tight also comes from the company's adoption of double-data rate (DDR) DRAM, a high-speed form of today's standard computer memory. Without DDR, Athlon would be lagging, agreed Anandtech and Brookwood. The Pentium 4, meanwhile, will be coupled only with Rambus until toward the end of 2001. While Rambus memory isn't creating a performance problem, it does add expense, a problem with the Pentium 4-based computers in general. Even though the Pentium 4 costs less than the typical new processor from Intel, the competitive landscape is harsher than in the past. Pentium 4 chips costs $819 each in quantities of 1,000. "For the rest of us who pinch pennies each month just to make rent, the Pentium 4 makes for great reading material," wrote Sharky Extreme.

To: richard surckla who wrote (128882)	11/20/2000 5:40:13 PM
From: tejek	Read Replies (1) \| Respond to of 1583404

Don't be afraid to read the last paragraph in this post....it makes for very interesting reading re P4 and RMBS! anchordesk.co.uk ____________________________________________________________ P4 willy waving I suppose you thought that the Pentium 4 was going to be the fastest chip in the world? Well, Monday marks the day when the sad truth about Intel's new Pentium 4 "Willamette" processor (Willy for short) will become public news: it's really not worth buying. At a clock speed of 1.5 GHz, it's barely faster than a Pentium 3 at 1 GHz. Intel, in short, has a little Willy. This time next year, the 2GHz P4 will be the fastest chip Intel makes for 32 bit software. It will kick. It will absolutely eat 3D software and other multimedia applications. But I'm afraid this time this year, it is merely Intel's most expensive bit of silicon. And at Windows, it's a slug. And nobody was allowed to know, because the fact that the P4 is a Windows slug has been under embargo. Everybody who knew, was gagged. The chip itself is no secret -- Intel didn't mind telling everybody all the clever things about the new processor architecture -- but the corporation has strongly prohibited anybody from announcing the results of any bench tests comparing it with other processor chips. And the reason is simple enough: if you're running Windows, there's simply no point in buying one. But the smoke-screen was bound to blow away eventually, and you may think it's a mystery that Intel tried to keep it secret at all. The question you have to as yourself is: "Who didn't know about Intel's Little Willy problem?" Who was it who had to be kept in the dark? It certainly wasn't Intel's rivals. I've been talking to them; they not only know that it's slow, but they know how slow. And they know why it's slow, too. The people who didn't know, sadly, are the suckers. Not you and me, hopefully; but the people who are in Intel's pocket. That's the PC makers. I've been talking to them, asking them about why they are launching a machine which is vastly more expensive than a Pentium III, but not noticeably faster. "But it's 1.5 GHz!" they exclaimed. "Obviously it's faster! they all said. Exactly how it performs, you'll find out from a whole host of sources come Monday morning. I haven't been able to run any bench tests, but I've been able to speak to people who have, and the tests are commonly available enough that the news hasn't been entirely suppressed: and if you want it in simple terms, then it's going to be something like this: "The Pentium 4 at 1.5 GHz will run ordinary Windows at roughly the same speed as a Pentium 3 at 1 GHz." Rupert Goodwins explains the technical details of why; but from most people I've talked to amongst designers of silicon, it is clear that nobody doubts the basic fact: there is almost no conceivable reason to buy today's 1.5 GHz Pentium 4 system -- unless you are a mad games player who will replace it in six months or so with a 2GHz Pentium 4 system. Now what you have to understand is that this is really no surprise to anybody who knows about chip design. It's happened before often enough; you'll remember the original Pentium Pro launch, which was embarrassing because the P Pro was actually slower than the Pentium at Windows 95 (it was optimised for 32-bit software). And it happened with the original Celeron, which was launched without any level 2 cache, and was noticeably slower than a Pentium 2. So in a sense, it should have been expected. Intel may have put journalists under a gagging embargo; but its own chip designers have been discussing the architecture with other chip designers for months; and to those people, the problem is as obvious as it would be to you that there would be a problem with a double-decker bus with a 50 cc engine. And the problem is Rambus. Intel delayed the launch of the P4 twice because the only motherboard chipset it has is the 850 design, which uses Rambus memory. And Rambus memory is fine (if costly) up to about 250 megabytes and up to about 1,000 MHz clock speed; and after that, it simply can't be made to go any faster. The problem is cache misses. If the data you want isn't on the processor, it takes too long to retrieve it from the lower levels of memory. NEXT