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.

Technology Stocks : Advanced Micro Devices - Moderated (AMD) -- Ignore unavailable to you. Want to Upgrade?

To: hmaly who wrote (73163)	3/4/2002 11:33:46 AM
From: wanna_bmw	Read Replies (2) \| Respond to of 275872

hmaly, Re: "The message I am getting here is that while you agree mhz is a poor measurement, you also dislike every other measurement. Aren't you sounding like Daschle the other day when he said that while he liked the way GW was handling the war in Afghanistan so far, he could do better. Sure he could." Not quite. I am not trying to say that I could do better. I am trying to say that AMD is setting themselves up for a long-term losing battle. But please let me explain why. For the reasons you have given, QuantiSpeed is great short term solution. The effectiveness lies in the results, and as you say, benchmarks and various product reviews don't lie - the Athlon is at the very least on par with the Pentium 4 to which the ratings relate. Of course, I believe that there is something inherently wrong with QuantiSpeed. As one of the more open-minded people on this board, you might be one of only a handful of people that can listen to and understand what I'm about to say. I only ask that you at least give me the benefit of the doubt. Basically, I see QuantiSpeed as AMD's effort towards two goals. The first is to debunk the notion that Megahertz relates to a processor's performance, and the other is to establish another metric for assigning performance to the Athlon, which gives it a meaningful comparison against competitive products. The fault of QuantiSpeed isn't in the metric itself, but rather in what AMD uses as a baseline. As you have already pointed out, the baseline appears to be the Pentium 4. In almost all of AMD's comparisons, they try to bench themselves against a Pentium 4 with megahertz that equals their processor's model numbers. This in and of itself is a faulty comparison, and goes against their Goal #1 (as I mentioned above). By setting megahertz as their baseline, they have effectively given credibility to the notion that it can be used to measure performance. But let's look at it another way. They named their 1.67GHz processor an Athlon XP 2000+. Compared to the Pentium 4, this would seem to suggest that this Athlon is capable of offering performance that meets or exceeds a Pentium 4 at the 2GHz rating. 2000+ to AMD is a performance rating. However, it is based on the original 2GHz rating from Intel, and therefore it is based on the assumption that 2GHz somehow relates to performance. A more meaningful model rating would not compare itself to Intel's megahertz rating. For example, let's say that AMD decreed that SpecInt would be the meaningful way to benchmark processor performance. We all know, however, that SpecInt is hardly the quintessential measurement for CPU performance, but for the sake of argument, let's pretend it is. What would stop AMD from, say, taking the SpecInt score of their processor, and adding a zero at the end? Therefore, their fastest 1.67GHz processor would be an Athlon XP 6970. Their 1.4GHz processor would be an Athlon XP 6330, etc. In my opinion that would be a meaningful metric for performance, and the consumer would be able to scale performance linearly by looking at the model number. Instead, however, the average consumer looks at AMD's model numbers, and instead relates it to Intel's established megahertz metric. QuantiSpeed continues to give undue credibility to megahertz, and it also presents other challenges for AMD in the future. With Hammer, the rules are going to change, and it's very probable that AMD will be altering the model numbering algorithm. They will again compare their processor to whatever the comparable Northwood is at the time. However, Northwood is certainly not the last core from Intel. Presott is aimed for the second half of 2003. Will Hammer have to again adjust model numbers to compare to Prescott? Or will AMD make their model numbers conservative in advance so that Prescott matches up with Hammer's performance much like Northwood matches up with Athlon XP? I think this decision would hold a lot of risk, since it's hard to predict what kind of micro-architectural enhancements will be present within the Prescott core (and Intel has already said that there will be some). AMD might be able to second guess, but here lies the long term problem that I am talking about. As long as QuantiSpeed is based on the Pentium 4, rather than on absolute performance, AMD is chained to Intel's future product lines. They are unable to make a story on performance, since their baseline will constantly change. AMD needs a fixed baseline - one that doesn't alter from year to year. So that's my point, in a nutshell. I don't claim to have the answers, but frankly it's not my problem. I am, however, an AMD investor, and I would appreciate from this forum a little less heckling when I express my criticisms. I don't criticize to give AMD a black eye, but rather to offer my opinions here when they matter. Take them or leave them. wbmw