Poor AMD

[pm]

Seewhy:
<<Let's say AMD introduce 1.5 Gig Athlon today, which has to be limited in quantity because of the yield. How much can they charge for 1.33??>>
In this scenario, I would say that the 1.5GHz CPU is not ready to be shipped. Why not? Because there's not enough bin split at 1.5GHz to allow it to be released. If you are only getting a few parts, then the product is not ready to be shipped. This is what I've been trying to say all along.


Bremen (on the subject of chips slowing down with time):
<<How much of a concern is this, and how fast does it happen? I've been under the impression silicon is mostly forever>>
It's an effect mostly caused by hot-electron gate impact ionization (AKA &quot;hot-e&quot. Hot-e gradually leads to increased Vt (threshold voltage, the voltage level that causes a transistor to turn on), and so waveforms need to reach a higher threshold before they are considered &quot;on&quot;. Since this threshold takes longer to achieve, the transistor is effectively slower. Contrary to common sense, the &quot;hot&quot; in &quot;hot-e&quot; doesn't refer to temperature but to electron energy level.

The specifics of time vs. effect are probably confidential (as well as very process dependent), but I will say that it occurs at any voltage, but the effect becomes substantially more pronounced as the voltage increases (since the electrons have more energy) and thus the problem slows down chips that are run at higher voltage much more quickly than those at lower ones.

And chips definitely do not last forever if you run them. OTOH, if they are sitting on a shelf in a hermetically sealed, humidity controlled, temperature controlled, non-oxygen, low-radiation environment, then they probably will last an extremely long time - like hundreds of thousands of years. Silicon chips are essentially crystals. If you subject a crystal to stress, it will eventually fall apart. If you leave it alone, it will pretty much last 'forever'.

<<Oh, and I find it hard to believe that in the 5 months? since the P4 was introduced that Intel has been unable to produce a faster version. The thing was designed to easily scale the MHz ladder... >>
That's what the press said about it. But the truth is that it's the same problem as it has always been. A 15% frequency increase on a Pentium 4 is the same level of difficulty that it was on the original 486. They are both CMOS process devices and they obey the same rules. The transistors are smaller, but they basically behave the same. If you want to increase the frequency, then it's the same level of difficulty.

So you really think that Intel is sitting on a 1.7GHz Pentium 4? That they could release it, but aren't for no real reason? I guess we are back into this discussion again, but (no offense intended) I'm actually having a problem understanding why this scenario makes any sense at all to you.

 

[TravisBickle]

yah, AMD chips are quite cheap. doncha all remember how expensive Intel chips were years ago? if AMD they could make 1.5GHz chips work perfectly, they'd ship them now at $500 or something. why not, it would be way faster than anything else around.
about a 733 being the same to produce as a 1000; yes, that makes sense, but let's think about the Celeron. because Intel choose to mutilate a chip to make it worth less, it costs more to produce. I bet they don't sell them at a loss though. which means that a coppermine core takes $60 or less to produce. so the rest on ALL chips is pure profit dude.
how can Intel be suffering in such circumstances? nobody else has margins like that in computers! except MS :disgust:

 

[seewhy]

<< And the comment that AMD and Intel are working together to slow down progress - this is almost laughable. If they did, then some startup would show up out of nowhere with a new wonder-product, have TSMC and IBM fab it for them, and they'd make a fortune. I can assure you that Transmeta did not originally intend Crusoe to be a mobile/laptop part, they wanted something that would compete in the desktop space, but they weren't fast enough and are now in the mobile arena. There are other Transmetas out there. Even so, if AMD and Intel somehow came together to slow down progress, Transmeta and Via would keep moving forward and would improve their products and their products would eat market share >>



Right now, in this kind of economic condition, both companies cannot afford losing more profit than they have to be. Yes it won't happen for long term, but it is conceivable for the time being, that both company agree on a silent pack to slow their pace (just my theory, nothing to back it up alright!!). If AMD introduce the fastest CPU they can like they did a while back, Intel could do the same now with their P4, or at least do some sort of paper launch they are so good at. Agian, doing that would reduce profit for both company. By the way, it is way too hard for a startup to design chips to compete with Intel or AMD in a short period of time, Transmeta took almost five years and their product isn't nearly as ready. VIA with their Samual is still a bit behind. So both company wouldn't worry about competition, at least in the short term.

Again, you have to look at the economic condition right now....I am stressing that. I agree that for the long term, both company will do whatever it's best for themselves. But sometime when faced with external problems like the ugly economic condition we have now, competiting company can cooporate to limit their loss. If I am running strategy for either company, I will do that. But that's just IMHO.

It is that hard to believe??

 

[seewhy]

PM:



<< Let's say AMD introduce 1.5 Gig Athlon today, which has to be limited in quantity because of the yield. How much can they charge for 1.33??In this scenario, I would say that the 1.5GHz CPU is not ready to be shipped. Why not? Because there's not enough bin split at 1.5GHz to allow it to be released. If you are only getting a few parts, then the product is not ready to be shipped. This is what I've been trying to say all along. >>



Well, then Intel introduced their 1Gg P3 when they had lots of supply?? We all remember Intel introduced 1G P3 to save face when AMD introduced their 1G AMD, even though they didn't have the yield. What you are saying is a good business practice and I hope all chip maker does that. But there are also marketing and other reasons behind introduction of a chip like what happened to Intel a while back. That is also my point. Both Intel and AMD could contiune to play game like they did a while back, and that will only cause slimer profit margin. Let say AMD have 10% yield on their 1.4 G, 50% yield on 1.3G, they can certainly introduce 1.4G for marketing reason to make intel look bad, but that if that reduce the price of 1.3G by 80%, that might not be worth while. Plus Intel might introduce 1.6G P4 even if they only have 5% yield just to save face again, and that make Intel that much harder to charge high price on their 1.5 P5.....so on.

That's my whole point of instead of doing the macho thing like a while ago, they are slowing down all together.

 

[pm]

Seewhy: I guess I see what you are saying. But the other part of the equations are personalities. The reason I found it &quot;almost laughable&quot; (and I realize that was an insultingly poor choice of words, and I apologize and can only claim fatigue for my lapse) is the personality of J.Sanders (CEO of AMD). I really can't see it. Maybe - possibly, but I doubt it.

But the fab tweaks will continue, and products will continue to be designed and released. Perhaps the rate will slow down slightly but this is simply economics (stop overtime for hourly workers, reduce expenses by not upgrading workstations as often, layoffs, reduced hiring, etc., these and plenty of others reduce expenses and yet slow productivity). I guess they way you phrased it above, it seemed sinister (like a secret agreement between the two companies) whereas the way that I see it when economics are in play is that it's merely a natural event.

Ok, I'm headed to bed. Let the flames begin! I'll see you guys in 8 hours.

Edit: one last bit... the Pentium III 1GHz was a bit of an exception. It's a good example for what you are saying, but I consider it to be an exception to the rule. Can you think of another that was like it?

 

[seewhy]

Naaa, Jerry is getting old and he's gonna retire soon. Now AMD is getting to a point that profit and Wall St. expectation is more important than just survival and personal ego...at least I hope so for the sake of my stock.

Anyway, it's been fun, time to sleep for me too.

 

[MadRat]

Intel has always catered to the speedking niche of the market, where profit per part is high. The bad part of this market is that it consumes an unholy amount of the critical resources necessary on the bigger picture. AMD has decided against aiming for this niche so that it can nail the intermediate crowd, where profitability has always been effective. On the flipside to Intel you have VIA, which is catering to the cheap crowd, where we all know there isn't enough money to bother.

 

[NOS440]

Rectalfier



<< If I try hard to remember stories of processors unfit for release it's not AMD that comes to my mind >>



Then what is this ????


PM very informative post as alway's

 

[jeffrey]

Is AMD delaying the Palomino processors because they want to keep the shorter pipeline or because they are reworking/adding to the pipeline? I can see why if they would want to keep the shorter pipeline because of it's strong performance. I could also understand it if they simply need to add more to get more mhz.

The Apple platform has gotten a lot of heat for being beatten badly in the mhz war. Their processors have a very short pipeline and the top G4 part was stuck at 500mhz forever. Motorola finally added two stages to the pipeline and calles it the G4+. This processor improved available clock speed ~50% at launch to 733mhz. However, there were reports that it was not as fast at 500mhz as the old processor.

Now, Palomino is AMD's new core, can they take the current core to the speeds necessary and expected of it, or will they have to add? The heat generation is going to have to be solved with Palomino for it to advance, but with heat issue taken care of can they continue upwards? Maybe AMD is delaying the part because they were originally going to add stages, but backed-off because they saw the p4 results. Now they want to advance with the new core with the same pipeline, but .18 isn't going to cut it and are waiting on .13?

 

[etech]

For what it's worth, AMD stock price crossed above Intel's today. It doesn't look like it will stay there, at least not today.

Poor Intel.

 

[jeffrey]

Damn, I was going to post the stock price bit.
Anyway, you can't really say poor Intel on the basis of AMD shares being higher than INTC's because of the fact that it's not relatively meaningful by itself.

Intel has a market cap of 187+ billion and is trading with a pe of 19

Advanced Micro Devices has a market cap of 8.8 billion and trades with a pe of 9

Even though AMD's share price is higher INTC is worth one heck of a lot more and the street values the company's earnings much higher.

*****corrected figures

 

[MadRat]

Well, they may be worth more overall, but six months ago they were worth four times as much...

 

[Rectalfier]

NOS440

You misunderstood. I was quoting Tom, from Tom's hardware guide. I should have made it more apparent.

 

[NOS440]

Rectalfier NP but quoting him is even worse. He has a hard On for AMD these day's all his reviews are bias as hell. He just wrote one on the 1.3 1.33 Tbird and over clocked it with a Frig. and compared that to a un OCed 1.5 P4 when the P4 will oc more with a $20 Air Cooler. What a loser. I used to think he was alright but these day's he just sucks.

 

[Bremen]

Well you gotta admit he (Dr. Tom) has justification. Intel was very rough on him for reporting the truth about the 1.133 GHz P!!!. I think he is well justified in his stance towards Intel.

 

[Degenerate]

<< Now, Palomino is AMD's new core, can they take the current core to the speeds necessary and expected of it, or will they have to add? The heat generation is going to have to be solved with Palomino for it to advance, but with heat issue taken care of can they continue upwards? Maybe AMD is delaying the part because they were originally going to add stages, but backed-off because they saw the p4 results. Now they want to advance with the new core with the same pipeline, but .18 isn't going to cut it and are waiting on .13? >>



if its made with the same pipline, then i would say its not very rampable interms of speed. Intel, and AMD for that matter of fact, know fully well that the P4 is here to stay and (with a little personal opinion) will be an excellent chip ooonce the spftware can catch up to it. as evident in the GF3, the Dx8 inproved the gf 3 by miles. I am expecting future aplications and games (probably now developed) will ahnialate AMD. (of cours thats just my opinion.).

The release of new CPU is really slowed due to economical reasons. bother companies know that todays people (the majority at least) dont need to spend AU$5000 on a computer to run ms word or IE or games.

anyway i would expect intel to release the p4 1.7 very soon.

 

[MadRat]

People without computers have no compelling urge to buy brand new. People with computers beyond 200mHz are generally happy with the speed for non-gaming applications. There is very little market share to compete; Intel is mis-aimed at a niche market and its been shown by its sales. AMD is aimed at the consistent sales mid-market and their results have been exactly that, steady.

I can't remember hearing the last time anybody around me bought a Celeron, but would guestimate its been six months or better. Duron and Thunderbird is what has stuck on their lips. AMD is really impressing me with their market penetration of the technological illiterate. Mate that with their inexpensive, yet reliable, product lines and you have alot of happy non-Intel customers.

 

[sandorski]

Hmm, this also is MO and you can hold it against if I'm wrong:

P4 optimizations will *never* negate the sheer power of a strong fpu unit at mhz-mhz comparisons. There will be a few apps that will clearly be faster on a P4, many will be close to being as fast on a P4, but most will lag behind on a P4. We've been down this road 4 times now and only 1 app(Quake 2) has shown any kind of significant benefit compared to a strong fpu unit. Intel needs to keep the P4's mhz higher than current AMD's current cpus. IMO again, Intel will abandon the P4 design long before it reaches it's maximum speed potential, once they realize it's inability to bridge the performance gap.

 

[NOS440]

Bremen Justified in his personal veiws or dislikes yes. But blatant one sided reviews that insult the intelligents of his reader's NO

 

[Soccerman]

I'd like to add something (it might have been mentioned already).

AMD isn't holding back small speed increases, we see them in our T-Bird @ 1.2+ ghz, and possibly Durons @ their higher speeds.

they're holding back Palomino, for business reasons (profit will be larger becuase people will be buying their 'new' CPU's today, and when Palomino is released, they'll want to buy that, thus increasing profits by a good margin (and not spending anything to get them).

do you understand? simply becuase the part is measurably better, is why people will want it over their already fast CPU's. AMD can have their Cake, and eat it too.

People who have been holding back on buying AMD might go for this one for some reason (lower power requirements??).

 

[Grminalac]

Well i believe that it was intels plan to release a 1.13 ghz and faster pentium 3 and then step up to the pentium 4. Unfortunetly Intel did exactly what pm said they did not want to do, which was release an overclocked PIII and YES it did stumble on bits of code and was recalled.

 

[Rectalfier]

P4 accounted for about 1 percent of all processor sales last month. I don't think that SSE2 will be implemented into the majority of software untill early 2003. That's about a year after Clawhammer comes out.

 

[Degenerate]

well, intel is the one who brings the new technology for other to follow..
they are the ones who tryies things out and then gain or suffer

i certainly think the palomino is not released becaues amd is not capoable of it.

 

[blaster31]

Hey Degenerate.. I guess you dont believe all of the interviews out there and the opinions on other hardware sites that AMD is more then capable of it.. but just doesnt want to for reasons already specified... I think you're being a tad Intelabee there

Intel makes the technology for others to follow eh? HOw about AMD's LDT/Hypertransport technology? HOw about the fact AMD was #5 in patents, while Intel was what, 15th? worse?

 

[Snoop]

Not to mention that AMD was also the first to bring out a fpu enhancing instruction set, 3DNOW, before intel with its SSE.