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: Elmer Phud who wrote (118312)	4/4/2004 7:07:23 PM
From: zx	Read Replies (1) \| Respond to of 275872

You have been completely disruptive of the flow of this thread. You are now ignored. Not that you should or would care. This is a warm and friendly place again.

To: Elmer Phud who wrote (118312)	4/4/2004 7:42:51 PM
From: Tomcat	Read Replies (1) \| Respond to of 275872

Their debt is larger than ever and they have only 1 logic fab. Yes they have debt and only 1 fab. They may only need one fab for awhile. There is always the IBM rumor. Their die size is huge and it runs on a very tempramental process. I don't know the die sizes off hand, but I didn't think there was much difference between Prescott and the 90 nm Athlon 64s. I think Intel is currently experiencing more process problems than AMD. ... there is now competition for x86-64 from a much lower cost supplier... Where can I buy one? ... and you think they (AMD) are in the best position in their history? In the past I don't think AMD had any presence in the server space. Today HP, SUN and IBM are shipping Opteron servers. AMD's presence in the notebook space has never been better with HP and emachines as OEMs not to mention Acer, Fujitsu, Voodoo and others. AMD has also kept competitive in the desktop space. They still need to work on commercial wins though. Intel has a much smaller die... You already said that, but I don't think that is true. ... and 2 300mm fabs going to 5 over the next year or so. That is a lot of capacity. It will be interesting to see if Intel can keep those 60% gross margins. If you want to think of Intel now as the second source then I guess that's ok.. I never said this. As far as processor performance goes I don't think Intel has much if any advantage over AMD. They are as close as they have ever been. And don't bet that AMD is resting on its laurels just as I am sure that Intel isn't either. I believe that AMD is just getting started and will continue to take market share and increase profitability. At least I am betting on it with my AMD shares. Good luck!

To: Elmer Phud who wrote (118312)	4/5/2004 11:20:50 AM
From: pgerassi	Read Replies (2) \| Respond to of 275872

Elmerp: Intel's 2 300mm logic fabs can't seem to produce much at this time. Less than 1 million Prescotts seem to have been produced last quarter. AMD's single 200mm logic fab seems to have produced as many 193mm2 dies as Intel has 112mm2 dies on these two huge 300mm fabs. And that single fab also produced over 6 million 100+mm2 dies as well. It seem yields must be 20 or more times lower for Intel by your previous arguments. Even if the yields somehow ramped to be roughly equal, Intel can't seem to sell 100 million CPUs a quarter at any price. They are desperate for anything that could use the output from these new fabs that they are willing to make TV chips at far lower ASPs relative to die size. You know what happens when you add too many factories producing products that can't be sold in the quantities to make them profitable. Yes, most companies have to shutter and close those factories taking the hit to the bottom line. Usually current management gets hit as well. That is why Intel doesn't want to admit that they built 2 to 3 times too much capacity. A $5 billion dollar loss or so is hard too explain to stockholders to keep their jobs. And this capacity where most of the costs are equipment, quickly depreciates to almost nothing even if it is not used. In two years, if those fabs are not used, they become obsolete and not worth much except as a shell. Thus any argument about the size of the die has no real meaning for Intel because they have too much capacity in any real case. Intel needs $5 to 6 billion dollars worth of production every quarter to stay profitable and that means $130 a die, with doubling the size only adding $5 to 10 dollars more per die. AMD is profitable at $500 million dollars worth of production which can be achieved with 2 to 2.5 million Athlon 64 dies which is possible even at 130nm with less than 50% of the fab. At 90nm, 8 to 10 million are possible. Since AMD has taken the high ground in performance, Intel over time must either let AMD make money (over $70 ASPs) or lose huge amounts itself every quarter (ASPs under $70). As far as SPECmarks, Xeon does not do very well when using the only 64 bit integer part of itself, x87 code (SSE2 only does 52 bit mantissas where x87 does 64 bit mantissas). It has half of the performance of Athlon much less Athlon 64. What would a customer do when Xeon loses against Opteron in 32 bit code typically by more than 50% but, gets less than 25% of Opteron in 64 bit mode? Where in 64 bit, 4 way Xeon is slower than 1 way Opteron. Having 64 bit mode doesn't help much if the performance is a snail when it is used. The big problem with SPEC is that the benchmark uses optimization by skewed means. It also does not seem to catch many abuses typically to reduce the work needed to be performed. Using a different benchmark like SciMark, Itanium and Xeon both fail miserably (Itanium more than Xeon). A preponderance of benchmarks show that Xeon is much slower than Opteron in 32 bit mode. No 64 bit ones exist for Xeon and given that they are supposed to be released less than 3 months from now, that is very troubling. Most likely they are worse than 800MHz Opteron's right now. AMD gave those out 6 months before release (even when floating point was weak) and 1.4GHz ones, 2 months before. So the longer we wait before any benches show up, the worse it must perform. Pete