Thursday, December 28, 2006

2007 -- A Year Of Promise

Tomorrow will be a new year. A lot of people will be looking for the Q4 06 results and making predictions for Intel and AMD in 2007. However, given what I've heard in the last couple of months it will be difficult to sort the wheat from the chafe. I'd like to look at where things are now and where they are going.

Reality has been a bit at odds with appearances in 2006. All of the fanfare has been from Intel. Revision F from AMD was a minor announcement as was 4x4. In contrast, Intel started the ad campaign blaring back in Q2 long before Core 2 Duo was actually available. And, it hasn't stopped. We have seen review after review of Woodcrest and Conroe and Kentsfield and Clovertown. Each review has been so hopeful and practically giddy with all of the incredible speed of C2D over K8 as well as lower power draw. Add to that, all of Intel's production capacity and AMD's capacity limitations and many Intel proponents were ready for announcements that Intel had taken back significant share from AMD. Except, it didn't happen.

The fourth quarter of 2005 was the best quarter that Intel has ever had in terms of processor sales. Yet, this didn't stop AMD from taking 4% volume share from Intel. I was a bit cautious about this in the first quarter of 2006 because I thought the gain could have been temporary and the volume share could have switched back. However, AMD not only held onto its 4% but managed to add another 1.5% giving it a total gain of 5.5% by Q3. Intel has gained none of this back nor has Intel gained any microprocessor revenue share. I'm guessing that some are waiting for the Q4 reports to see if Intel has gained anything. However, with AMD's capacity constraints Intel could have temporary gains if the market volume surges more than AMD can match. This wouldn't be a real gain because the % share would fall back again in Q1 07 when the volume drops. The best time to see volume share will be the second quarter when volume is lowest. Sales at that point should reflect a true preference in hardware for the market unless the total revenue falls. With falling revenue, share increases by AMD may not be accurate which is why I wasn't sure about AMD's volume share increases until Q3 of 2006. However, if the market stays up then Q2 should be accurate.

That Intel has not gained share is simply not believable to some. They point out that C2D is better and therefore Intel should gain. The problem with this idea is several things. First of all, Intel started producing C2D in mid year and then has ramped to 25% of production by today. This puts C2D at only 6.25% of production in Q3 and 18.75% of production in Q4. So, C2D accounted for only 12.5% of Intel's production for 2006. This has simply not been enough to have much effect on share. AMD will leave 2006 with the best year it has ever had with twice the processor revenue that it had 2 years ago. In contrast, Intel will leave 2006 with about the same processor revenue as 2 years ago. I think anyone with common sense can see that AMD has done much better than Intel in 2006.

Seemingly to fix what some must perceive as a devine right by Intel to take back share, some have invented paper gains for Intel by including the X86 sales from VIA with the volume figures. With one of their processors being end of life, VIA dumped these on the market. If we include these processors in the numbers then this lowers Intel's Q2 share even further which then makes the Q3 numbers appear to be increasing because the VIA share is gone again. However, these are only paper gains and don't reflect any actual increase by Intel. If we look at the share between AMD and Intel only then Intel has lost more share since Q4 05 and there has been no gain for Intel in Q3. Those who stoically ignore such attempts or perhaps who find the small paper gains to be insufficient then turn to early 2007 with great hope for share gains by Intel. The basic argument is that Intel is still ahead with C2D and with its higher percentage of production they argue that Intel should be able to gain. However, AMD is not actually that far behind. The latest reviews that I've seen suggest that K8 is about 10% slower per clock than Intel. However, since K8's are priced lower per clock than Intel this makes the AMD processors about the same except for FX-62 which is more expensive.

That AMD's processors are priced about the same as Intel's for the same power has again caused a lot of consternation from Intel fans. Some have even insisted that AMD should price its X2's below the lowest priced C2D, the E4200 but this has not happened. I still see the arguments popping up, groundless though they may be. It seems that some just cannot understand that the fact that E6700 is faster than FX-62 doesn't mean that all C2D's are faster than all X2's. The AMD X2's priced similarly to E6300, E6400, and E6600 have similar peformance for most tasks. And, there is the neverending misuse of benchmarks to prop up smoke and mirrors arguments about C2D. For example, it is understandable that video conversion benchmarks keep all of the data in memory to test the processor. However, this does not mean that users will see the same speed when the data is pulled off the harddrive under actual use. This is true of common user applications which spend most of their time waiting on the user. There are similar problems with tests that don't use all of the cores, benchmarks that show big gains because of cache, and benchmarks that are based mostly on the graphics processor. For example, a benchmark might be very speedy due to cache when it is the only thing running but this doesn't reflect a real computing environment at all where that same process would have to make do which much less cache.

Perhaps because of the lack of gains by Intel there has been a flurry of negative rumors about AMD lately. This includes ideas that AMD is behind in 65nm, is having problems with 65nm, and won't deliver K8L on the desktop until Q4 07. AMD has said that 65nm production is at 25% at the end of 2006. It is impossible that AMD could be at this level without having shipped something like 8 million processors. Some of these would have to have been delivered early enough to be out before year's end yet rumors persist that AMD has delivered no 65nm chips. The second idea comes from the fact that the 65nm chips are not half the size of the 90nm chips. This has led to some rather wild speculation that AMD has added new undocumented sections to the die or that AMD is having problems with 65nm. Actually, neither of these are true. Although 65nm is capable of producing die features that are half the area of 90nm you can't simply shrink 90nm transistors like you would a photograph. If you just shrink the 90nm transistors then at some size they will stop working properly. The first production with 65nm is only a partial shrink because this is the first round of 65nm transistors. These will get tweaked better for the 65nm process in revision H (K8L) and will be able to be reduced more. The 65nm dies that AMD produces at the end of 2007 should be shrunk more than what is being produced now. Again, this is not unusual as AMD did the same thing with 90nm.

As to the idea that AMD won't deliver K8L on the desktop until Q4 07, AMD has contracts with one supercomputer that will be running by end of Q2 07 so K8L production has to start in Q2. Desktop versions of K8L will come out in Q3. Apparently, the fact that the intitial supply of K8L is tight because of preordering has led some to believe that production won't start until Q3. Then apparently the logical assumption is to push back the desktop production to Q4. People have likewise pushed Intel's production of 45nm up from Q4 to Q3. This would then have 45nm being released from Intel before AMD has any desktop K8L's. However, not even Intel is saying this. Real production of 45nm by Intel in Q3 07 is unlikely.

And, what would rumors be without speculation of a point to point bus for Intel? Supposedly, Intel will release a point to point bus (CSI) in Q3 07. Actually there is no indication of this at all. The earliest that it looks like Intel could release CSI is early 2009. It looks like people are getting CSI confused with Intel's quad FSB chipset which will finally allow Woodcrest to be 4-way. The confusion seems to come because Intel insists on calling this a high speed bus rather than a Front Side Bus. Other than the name change it looks and works exactly the same as the current Intel FSB's that connect the processors to the northbridge. A true use of CSI would require an onboard memory controller like AMD has plus a new socket since you cannot run a memory controller on the current socket 775/771. There are other cards to fall yet. For example, AMD is releasing a faster Athlon 64 which should match E6700. This could be offset by a faster 3.2Ghz Conroe but only if Intel slides its prices down one notch. I'm sure Intel will do this eventually but perhaps not in Q1 07. There have been rumors that Intel will have 3.5Ghz chips by end of 2007 but then again the rumors insisted vehemently that Intel was going to have a 3.2Ghz Conroe at launch and this hasn't happened yet.

So, 2007 becomes a race. Intel will ramp C2D up from 25% while AMD continues to ramp 65nm on FAB36. Then once the new bump and test facility is completed AMD will begin taking FAB 30 down for conversion to 45nm. The best situation for Intel would be that the market expands rapidly which would allow Intel to grow faster than AMD since AMD will be capacity limited through most of 2007. With normal expansion, Intel is likely to continue to slowly lose share to AMD. This may seem unfair however AMD has cultivated better relationships with customers than Intel has in the past. By the time Intel has demonstrated any change AMD will have K8L. Further, Intel's gains have mostly been on the desktop which is the least profitable area. AMD has gained in mobile which is more profitable, has produced fewer low end Semprons, and is still holding onto the top 4-way server market. Intel won't be able to move into 4-way until Q3 07 at the earliest when it finally has a chipset with 4 FSB's. Further gains by Intel on the desktop will most likely be offset by gains from AMD in both mobile and corporate areas. I would expect AMD to end 2007 with a gain of about 2.4% in volume share.

The problem for Intel today is that this isn't like 2002 when Intel was able to grab back large amounts of share. Intel now has to deal with an AMD that has a credible name in servers, a growing presence in corporate, total system solutions from the ATI purchase, a growing presence in mobile, and both reduced costs and increased capacity with a second 300mm FAB. The situation would be completely different today if AMD still had only FAB 30 and if this were the case all my money would on Intel. However, with FAB 36 coming up in capacity and AMD's uncanny ability to do rennovation without taking a FAB completely offlline I just can't see anyway for Intel to take advantage. If the market expands quickly then AMD would gain less volume share but revenue would be better for both AMD and Intel. I'm still expecting Intel to have to do more divestment and probably another reorganization in 2008. Overall, however, 2007 should see Intel's and AMD's product lines and graphics support move much closer. By end of 2007 Intel should have something to compete in 4-way but it will have increased competition in 2-way and the desktop. 2008 will then bring new mobile competition from AMD as well as increased server and graphic competition. Intel will have to be on its toes in 2008 to stay competitive and 45nm by itself won't be enough. Looking further ahead Intel proponents are quick to point to CSI but Intel will then be facing an AMD with two functional 300mm FABs, tightly integrated graphics, and a modular core for easier upgrades so 2009 will still be highly competitive. Worse still for Intel, with 30% of the cpu market for AMD, Intel's monopoly will be broken. So, appearances in 2007 should be better for AMD and Intel should be able to climb back up to the 2005 processor revenues but mainly the advances by both Intel and AMD will be good for consumers.

139 comments:

Erlindo said...

Great forecast Scientia.

I do hope AMD to do well in all fronts (servers, mobile and desktop).
Once AMD captures 30% of overall market share, we won't need to worry about intel and their efforts to monopolize the entire market.


Happy new year to all. :D

gdp77 said...

Happy new year!

I generally agree with your sayings, but :

It seems that some just cannot understand that the fact that E6700 is faster than FX-62 doesn't mean that all C2D's are faster than all X2's.

Seems to me that 6600 is faster than anything AMD has to offer atm.

The AMD X2's priced similarly to E4200, E4400, and E4600 have similar peformance for most tasks.

You probably mean 6300,6400,6600 . And no u r definitely wrong. AMD x2 have much worst performance / dollar compared to conroes and they should be lowered prices. I guess AMD don't have to lower their prices, since they manage to sell all they can make. So it is fair by me. Just DON'T wait the informed (not average Joe) computer user to buy x2 over C2D. This won't happen.

The latest reviews that I've seen suggest that K8 is about 10% slower per clock than Intel.

What do you mean? That a 2.13 GHz C2D would have a 10% performance lead over a similarly clocked x2 ?

Greg said...

gdp77, I see nothing that completely justifies your assumptions on the performance deficits between AMD's and Intel's processors. I've seen reviews that at best showed a 15% clock per clock performance advantage, but those reviews were from anandtech and THG, and all others showed between 8% and 11% clock per clock performance advantages. So, the 6600 is not faster than the fx62 consistently enough to be faster overall.

As to the "informed" buyer, those are few and far between, which has always been part of Intel's advantage until late. Now it's part of their disadvantage, as too many people are still riding on the wave of good press AMD had back in the netburst days. Also, as an "informed" buyer, who needed to upgrade his father's computer to a dual core machine for $350, I went with an AMD build, because Intel's option's in that price range were not up to par.

Scientia from AMDZone said...

Seems to me that 6600 is faster than anything AMD has to offer atm.

Well, you need about 2.6Ghz from AMD to match E6600 and that is available. I'm not going to give any weight to most of the "testing" that has been done. The tests at Tech Report are the best ones I've seen so far.

You probably mean 6300,6400,6600 .

Yes, thanks for the correction. I fixed it in the article.

And no u r definitely wrong. AMD x2 have much worst performance / dollar

No. AMD is about the same up to E6600. Intel's current advantage is for E6700 and higher plus Kentsfield.

Just DON'T wait the informed (not average Joe) computer user to buy x2 over C2D. This won't happen.

It happens everyday.

What do you mean? That a 2.13 GHz C2D would have a 10% performance lead over a similarly clocked x2 ?

Yes. C2D is faster at the same clock. Rough matches:

E6300 1.86Ghz = 3800+ 2.0Ghz
E6400 2.13Ghz = 4600+ 2.4Ghz
E6600 2.4Ghz = 5000+ 2.6Ghz
E6700 2.66Ghz = 6000+ 3.0Ghz

gdp77 said...

No. AMD is about the same up to E6600. Intel's current advantage is for E6700 and higher plus Kentsfield.

Even if I accept your opinion (which I do not) , what's the price for buying a FX62 in comparison to a E6700? In Greece, where I come from, E6700 is about 200€ cheaper.

So, maybe u wanna change your comment about the "informed" people. U can't pay more, in order to buy less. U r not informed then, u r something else...

gdp77 said...

http://www.techreport.com/reviews/2006q3/core2/index.x?pg=1

In this review from tech-report (which u respect most) E6600 destroys in almost every test the competition form AMD.

Maybe u want to reconsider the performance / dollar statement again.

ashenman said...

gdp, we're not considering the fx model processors because those always carry an inconsisten price tag with retailers. However, you could look at the fx74, which definitely needs some benchmarking to be done on its own (I wish there were single socket socket F mobos widely available).

Looking at the 5000+ and down, the prices are exactly on par in terms of performance with Intel's lineup.

Jeach! said...

Hey Scientia, nice work!

I'd like to add a point in regards to production prices and margin numbers.

Many might say that Intel has a lot of low margin products in its lineup, such as chips, flash memory, etc.

But for one, how much of these products account for total revenues?

Second, when they say low margins on these Intel products, are we talking about 10%, 20%, 40% or 55%?

You've got to also remember that most of 2006 Intel produced at 65nm while most of AMD's production was at 90nm. All the while considering this fact AMD still managed to raise its margins to surpass Intel.

It will be very interesting to see AMD's margins for 2007 since AMD has only to gain with 65nm production maturing.

qurious69ss said...

"It will be very interesting to see AMD's margins for 2007 since AMD has only to gain with 65nm production maturing."

ASP are under pressure from the price war and the deal that Ruiz made with Dell was signed in blood not to mention that Ati now comes into the mix and their margins are anemic to say the least so I would say for 2007 margins for Amd will at best remain flat.

Dr. Yield, PhD, MBA said...

jeach! wrote:
Many might say that Intel has a lot of low margin products in its lineup, such as chips, flash memory, etc.

But for one, how much of these products account for total revenues?

Second, when they say low margins on these Intel products, are we talking about 10%, 20%, 40% or 55%?


C'mon jeach!, how hard is it to look up yourself? The numbers are right there for the taking on the Intel investor relations page. If you look at p.10 of their Q3 earnings release, you can see that even with chipsets rolled in, Q3 net margins were:

digital enterprise group 17%
mobility group 41%
flash memory group -23%
all others -264%

If you amortize R&D, SG&A and restructing costs by revenue %ages, you can estimate group gross margins:

digital enterprise group 51%
mobility group 75%
flash memory group 11%
all others -230%

So you can see, even with chipsets and mobos, Intel is still getting margins equivalent to or better than AMD in their processor groups as of Q3. You can also see that flash and others (some of which are now likely divested) are a HUGE drag on margins and profitability.

Mo said...

"Well, you need about 2.6Ghz from AMD to match E6600 and that is available. I'm not going to give any weight to most of the "testing" that has been done. The tests at Tech Report are the best ones I've seen so far."

Why is that E6600 edges out FX-62 (2.8Ghz) in most all benchmarks?

Wondering if it's a typo?

Greg said...

It doesn't edge out the fx62 in "most" benchmarks. The 6600 edges out the 62 in some benchmarks, but at most, 40% of them.

Ignoring gaming (because that is wholey dependent on the platform in all of the situations the tech report tested, and all the results were close enough to be too easily swayed by that) the 62 and 6600 are neck and neck. The 5200 is literally a 62 without 1M of cache, so AMD has an equally priced answer to the 6600(though only available from pc builders). Also, if you don't ignore the cost of the platform, Intel's systems become more expensive, and the price/performance advantage is much more justifiably equal.

gdp77 said...

Ι don't understand why we should ignore gaming. Wasn't gaming performance that "showed to the world" the superiority of K8 platform? Whenever reviewers bench games, they use same graphics cards for K8 and C2D. So why we should ignore gaming? Using the same graphics card, C2D is still superior.

enumae said...

Greg said...

"Also, if you don't ignore the cost of the platform, Intel's systems become more expensive, and the price/performance advantage is much more justifiably equal."

Example,

1. AMD Highend - Asus L1N64-SLI WS (4x4 capable) $430

2. Intel Highend - EVGA 680i $250 (Quadcore capable)

Not really, maybe lowend, but not at the highend.

Greg said...

Well, considering that's technically a platform Intel doesn't even have right now, I'd say that it's not a good comparison. In fact, that's a really bad comparison, and doesn't contribute at all to the argument, since that's the only AMD board that's meant for the consumer that actually costs that much. Thanks a lot enumae.

enumae said...

Greg said...

"Well, considering that's technically a platform Intel doesn't even have right now..."

They don't have quad core?

Or they don't have a native quad core?

What was it that was competeing pretty well against 4x4 then?

"...and doesn't contribute at all to the argument, since that's the only AMD board that's meant for the consumer that actually costs that much."

Is that Intels fault, or AMD's?

The fact is on a platform price/performance Intel is beating AMD on the high end.

Do you just not like this, or can you actually prove it wrong?

enumae said...

Scientia said...

I know you have stated in the past that you try and be unbiased, and to back up your claims with logic and facts, but as I read this post I am not seeing the unbiased journalist I have in the past.

"AMD has said that 65nm production is at 25% at the end of 2006."

25% of Wafer starts, or 25% of finished processors, there would be a large difference there so please elaborate, and a link would be nice?

"Then once the new bump and test facility is completed AMD will begin taking FAB 30 down for conversion to 45nm."

Could you link to AMD saying this?

Everything I had read I believe said 65nm.

"With normal expansion, Intel is likely to continue to slowly lose share to AMD. This may seem unfair however AMD has cultivated better relationships with customers than Intel has in the past. By the time Intel has demonstrated any change AMD will have K8L."

I am really going to try and put you on the spot, and I am looking for an objective answer here...

1. What happens if K8L does not match Core 2's performance, or is just on par?

2. What happens if AMD can not increase capacity as fast as planned?

I really do enjoy your post, but in this one there is alot of if's and assumptions.

Wise lnvestor said...

enumae I really don't like to argue... Cause arguing (even if done intelligently) will only make people look like a fanboy.

1. What happens if K8L does not match Core 2's performance, or is just on par?

A: MAKE K8L a sold out regardless of performance. Especially on the server front.
AMD will just have to find a good balance for server parts and desktop parts and how many to make.

2. What happens if AMD can not increase capacity as fast as planned?

A: Build another Fab. like 1 in New York with 650mil grant. If that's not enough there are numerous ways to raise capital. Like issuing more stocks. Just like all growing enterprises do.

Think like a chess player!

BTW a little about me. I am a investor. More to it: a neutral investor. I bough AMD I sold AMD, I bough INTC I sold INTC, and I bough
NVDA and sold NVDA.

Bottom line is I only increase the value of my portfolio. If a company does not contribute to it then it will not be in my portfolio.

Greg said...

My point about that board was that it's the only board out of hundreds of boards that is not a server board that costs even more than $200, and in fact, they probably all cost even less than that now.

The platform is essentially a consumer grade workstation board, which Intel doesn't have. Obviously a platform goes beyond just the number of cores, which I thought you would have realized.

AMD basically doesn't sell to the normal high end. They do have a problem with who actually needs or would want to use 4x4 as it is right now, but it is still pointless and inflammatory to consider it in normal AMD platform cost.

enumae said...

Greg said...

"...but it is still pointless and inflammatory to consider it in normal AMD platform cost."

I do not want to argue with you, so I will ask one simple question...

Do you condier the benchmarks that compare the QX6700 and FX 70 - FX 74 pointless and inflamatory as well?

gdp77 said...

He also forgets the fact that u can find 965 mobos for the same price u can find AM2 mobos. So the argument that K8 is a cheaper platform is simply not true.

My position still remains that "no informed buyer will buy a AM2 based computer until Q2 06".

enumae said...

*EDIT*

Do you consider...

Sorry about that :)

Dr. Yield, PhD, MBA said...

Wise investor wrote:
K8L might not be a great desktop chip. But just might be a great server chip. It score 3 pre-orders for HPC and that in itself is a accomplishment.

And naturally investors likes high margin segments.



But at what cost? I've posted on a lesser site an analysis that leads me to believe that for every 1 K8L produced, AMD loses capacity to produce 3+ 65nm X2s. Can they make this up in higher ASPs? Probably- but if they piss off OEMs and the channel by starving mainstream part supply to get the high end capacity, they cut off their nose to spite their face.

As to the nature of the high margin HPC market- think about it in gross numbers. Assume you sell a dozen 1024 CPU systems in a year (they are uncommon). Assume you can clear $1k GM per cpu. That a bit under 12.3M in gross profit, and substantially less in net profit. Even if you move that to a dozen systems a quarter, you haven't even cleared enough gross profit to buy 2 immersion scanners with tracks for your 45nm node fab conversion. Big whoop- dig deeper.

Wise lnvestor said...

Sorry I should not respond to enumae
in place for scientia. But sometime it just hits me that some arguments does not contribute to exchange of knowledge. At the end of day there is no benefit but seeing time wasted.

From my investor point of view:

There a big world out there, just beyond your living room window.
And AMD is aiming for that.

K8L might not be a great desktop chip. But just might be a great server chip. It score 3 pre-orders for HPC and that in itself is a accomplishment.

And naturally investors likes high margin server segment.

Wise lnvestor said...

Dr. Yield, PhD, MBA:...leads me to believe...

So I take it those aren't final specs...

Besides Dr,(if that's really who you are)specializations, you can put specialized chip in substitute for a Opteron chip.(IBM's new HPC)

As for pissing oem's. Well...you should really ask someone who care... I am just a investor. ^^

Wise lnvestor said...

Humm "Dr. Yield, PhD, MBA" you sound like AMD would actully loss production for 3 x2's for 1 K8L. But in reality, I think AMD will switch their Opteron alocation for new K8L Opteron production. I don't REALLY don't see a production loss here.

Greg said...

Dr. Yield, AMD, like Intel, will produce quad core products to meet that market's demand.

Unlike dual core, quad core comes at a time when people's performance impovement expctations are saturated. That's why people think computers are so cheap right now. This slows quad core adoption and allows AMD to more dynamically address that market while ramping dual core barcelona and selling older dual core to the larger markets. Ya, supply will be short compared to demand, but will be much larger than it ever has been.

Enumae, we're talking about in-general at them moment, which means AMD is just as expensive as Intel if you buy equally priced motherboards. If you buy equally capable motherboards, Intel's will be more expensive (and please don't bring performance into this, because motherboards shouldn't change performance).

No, I don't find the quad fx benchmarks inflammatory at all. However, using quad FX to describe the state of AMD's options in general or, even worse, for the mass market, is inflammatory and pointless, and you really should know why.

george said...

Core 2 duo is approximately equal to k8 when it is price moderated. and k8 will have an upgrade option to k8l with 4 cores or if you got an quad fx box then you can get an 8 core box.

enumae said...

Greg said...

"Enumae, we're talking about in-general at them moment, which means AMD is just as expensive as Intel if you buy equally priced motherboards."

Agreed, and being due to all other parts being the same (RAM, GPU...).

"If you buy equally capable motherboards, Intel's will be more expensive (and please don't bring performance into this, because motherboards shouldn't change performance)."

Please show me an example of this.

"No, I don't find the quad fx benchmarks inflammatory at all. However, using quad FX to describe the state of AMD's options in general or, even worse, for the mass market, is inflammatory and pointless, and you really should know why."

Is this going to be a debate where you don't read my post?

I specified high end and never made mention nor notion to "...the state of AMD's options in general or, even worse, for the mass market".

4x4 is an AMD platform, and as such should be factored in when comparing Intel and AMD systems.

gdp77 said...

There are bigger monopolies in the world. It's not like there's a shortage of AMD's at Best Buy. Intel can't help it if AMD wants to buy graphics companies instead of fabs:p

I agree with your comments, but not this one. AMD ones more wants to innovate, something Intel haven't done for many years. And buying ATI will help them to create the pc of tomorrow (fusion / modular / etc. ) . AMD's cpus suck now at the desktop compared to Intel, but let's not forget who brought into desktop the IMC, 64bit, multicore cpu, HT etc.

Red said...

Getting more in depth with the "bigger monopolies in the world" statement, there are other sectors that have more of an impact on us than CPU's, like Microsoft "tieing" IE into Windows, Apple and the iPod/iTunes lockup, Google having their products placed over others in AdSense, Wal-Mart just running over the boutiques, Sony using their previous dominance to try and push $600 crap on us, etc, etc, etc. I really don't see what evil monopoly Intel has now and how the market would be any different if AMD had over half the market. Are invisible pink unicorns finally going to become visible or something?:D

My AMD buying graphics companies as opposed to fabs was implying that if AMD tried to get more capacity, maybe they would be able to serve Dell and the channel and everyone would be happy and that Intel wouldn't have a "monopoly" since AMD would actually be able to sell to the % they couldn't sell to before. Besides, Fusion is aimed at the extremely low end, mobile, UMPCish market, right? What expertise does ATI have for implementing low end graphics on the CPU that they themselves can't do? Yeah, yeah, modular, 64 bit, "innovations", all of that is nice and all. I could care less who brings it on the market as long as it works and works well:) But I guess you are right in a way, no point in buying fabs to produce inferior chips so they have to find something in ATI..:D

Greg said...

Sorry I wasn't clearer about motherboards and performance, but as long as you're not buying a horribly manufactured chipset and motherboard, motherboards don't effect performance enough to warrant a price premium. From there, you simply look for features, which AMD has boards that are cheaper than Intel's that have just as many features. For example...
http://www.newegg.com/Product/Product.asp?Item=N82E16813131013
compared to
http://www.newegg.com/Product/Product.asp?Item=N82E16813131071

But we're not talking about only the high end. When you referred to the high end, I assumed you were making the assumption that AMD's price/performance scaled equally with quad fx. My point still stands though that quad fx is not a true quad core solution but a consumer grade workstation solution.

Though it is their platform, it is in such limited release that it should barely effect AMD's standing.

Red, the e6600 is not just $200 dollars, even if it did beat the fx62 most of the time (which it didn't). Most reviews show multiple instances of gaming (which will all perform the same way) and multiple instances of the same benchmarks both optimized and unoptimized (which only determine whether or not the optmization matters or not) which in most cases inflated the performance of the core2. As is, the 62 is on par with the 6600.

Though he has only specifically picked on THG and Anand, he has shown us a shining example (the one I was referring to) of unbias and effective testing methods. Thus, you can figure it out by comparison red.

Mid 2007 = q2.

No, he's not arguing against 1 person's opinion, especially since he was arguing the same point against you before, and I've heard multiple claims of the same thing from hoho on rubyworks and others on THG forums.

While it is stupid to believe AMD would only ever tell the truth, a number like 25% so late in the game, is just too much to lie about. Chances are Dell and the like are buying most of them and when they distribute them most people wont even notice if they have it.

Also, regardless of the fact that the core architecture scales better at higher clock speeds doesn't change the fact that the benchmarks we see agree with what we're saying about the values.

Nice job reading scientia's article where he perfectly explains the die size and the poorer cache performance.

Also, seeing how AMD will release a 3ghz x2 and have a 2.8 ghz x2 (so basically a 6700 and 6600, though roughly at best). Also, notice my previous points regarding price competition.

His point is not that because one rumor is wrong others will be, but that they are only rumors, and that they should be given only the worth of rumors, and nothing more.

enumae said...

Greg said...

"...From there, you simply look for features, which AMD has boards that are cheaper than Intel's that have just as many features...."

I am not trying to nit pick here, but although the boards are close they are not exactly comparable...

Intel has three PCI-E slots (2x16,1x8), Wi-Fi, an additional Raid controller, 2 external Sata ports, and is using the 590 chipset.

I do not know if that would be enough to account for the $60.

For an equivelant AMD board look here. The price seems more in line with what you are getting, less the 3rd PCI-E slot (1x8), but does have Wi-Fi, an additional Raid controller, 1 external Sata ports, and is using the 590 chipset.

Now from what I have looked at the advantage for AMD begins at or around the mid level.

"...My point still stands though that quad fx is not a true quad core solution but a consumer grade workstation solution."

I can see your point and it makes sense.

"Though it is their platform, it is in such limited release that it should barely effect AMD's standing."

I agree it won't change there overall advantage on platform price/performance on the mid to low end, like I had said, my reference was strictly high end.

Wallachian said...

K8L quad core will definitely outperform Clovertown by upto 5-8% in most benchmarks, due to its more robust design. But it will be atmost about 5-10% of AMD shipments in Q2-Q3. The max clock-speed of this quad-core will be in the 2.5-2.6Ghz range.

Though AMD uses SOI, and Intel uses strained-Si, the process performance of both is not unlike each other. At 65nm, both quad-cores run at 2.5-2.6Ghz at around 120-125W TDP. Only AMD quad-cores will have slightly better overall performance due to their robust system design. This will be the case in Q3.

In Q4, we will start seeing some launches of Intel's highend quad-cores based on the Penryn 45nm design. The first clock speed will probably be 3.0Ghz to probably 3.33Ghz in Q1'08 to Q2'08. At 3.0Ghz Intel's quad-core will again be difficult to match by AMD, as a 3.0Ghz quad-core by AMD at 65nm will be around 150W TDP (loose guess).

AMD's dual core K8L's will probably be their mobile core, clocked upwards into the desktop realm. And this is where they can start to look good. But other than this, it doesnt look like AMD can do much to create new momentum, until they get to 45nm themselves with new quad-cores, 8-cores and their fusion plans. They will probably have to settle to being no.2 for the next year as well.

gdp77 said...

it doesnt look like AMD can do much to create new momentum, until they get to 45nm themselves with new quad-cores, 8-cores and their fusion plans. They will probably have to settle to being no.2 for the next year as well.

Well, they could introduce the HTX slot with RD790 and enchance performance by using HTX add ons (possibly a HTX gpu). Intel then would have the fastest cpu, but AMD could have the fastest platform. Moving the game into a platform one , would be the smartest move from AMD, since they can't match Intel's process lead.

Kalle said...

"let's not forget who brought into desktop the IMC, 64bit, multicore cpu, HT etc"

They brought it to desktop but certainly didn't make all those things from scrach themselves.


"Well, they could introduce the HTX slot with RD790 and enchance performance by using HTX add ons (possibly a HTX gpu)."

Since when has PCIe been too slow for connecting GPU's?


Also please keep in mind that when you are talking about high-end or enthusiast market then with those people OC'ing their stuff is almost mandatory. I don't think anyone questions who has the lead in that area, even with 4-core boxes.

Greg said...

Nice point red, unless you consider Vista, which actually lessens the margins of AMD's performance deficit, and is an operating system any serious computer user should buy (even though it sucks to have to buy it) due to its actual performance improvements.

Too bad he doesn't like those benchmarks, but that doesn't make my point moot about benchmark inflation.

Red your whole "paper-launch" argument exists because you don't consider pc makers as customers (even though they are much larger customers than us), which means that the 5000+ was not a paper launch, but a launch that was completely bought up by pc makers (being that it was available from Dell about 1 month after launch (sorry there is no real way of effectively proving this).

This is definitely why we don't see any 65nm parts available from AMD at the moment, because pc makers have bought up enough to make the amount that reaches the channel too hard to sell effectively.

Again, I'm sure we've mentioned this many times before, but not all transistor technology can immediately make the process shrink and work. Really, what their transition to 65nm has shown is how good their 90nm process has become with all these improvements, and how much we have to look forward to, which is why AMD wont be running 150 watt TDP when they do quad core at 3ghz.

Also, we have no example of Intel transitioning a core over to 65nm except for netburst, which did not gain significantly. Chances are, AMD is increasing their stepping in K8L (since that's what every company does with a new process) so I could easily see them making much high clock speeds on that process.

Actually, AMD has a perfect way to improve momentum with fusion, if they can hold onto their viability till 2008 (which doesn't look like it'll be a problem). Then Intel will have nothing to counter this approach from AMD unless larrabee just comes out of nowhere (not likely, as no matter how much money you throw at research, it will still take a large amount of time to get a new graphics company off the ground).

Chances are, at the worst we'll end up in a situation like we were in just before AMD had A64 in terms of performance, but this will still be completely different because Intel no longer has a stronghold like Dell, and no longer has flexible pricing, nor is it the only processor maker with a platform anymore. So yes, things are looking very good for AMD.

By the way enumae, very good points, and thank you for a better comparison on the boards.

gdp77 said...

"let's not forget who brought into desktop the IMC, 64bit, multicore cpu, HT etc"

They brought it to desktop but certainly didn't make all those things from scrach themselves.


Oh, come on... What Intel managed to do with 10 (or even more) times more R&D money?

enumae said...

gdp77 said...

"Oh, come on... What Intel managed to do with 10 (or even more) times more R&D money?"

One of the things most people overlook is how diverse Intel is with it's R&D, they do not only focus on CPU's, that may be what they are known for, but thats not all they do.

Intel's website, just take a look.

Scientia from AMDZone said...

I guess after the low number of comments on the last article I wasn't quite expecting this many on this one. Presumably being in Sharikou's blog roll makes a difference. According to Analytics Sharikou's blog is now referring 4X as many people as AMDZone which used to be the biggest. I guess I'll have to rake the comments and catch up.

25% of Wafer starts, or 25% of finished processors

Since it wouldn't make any sense to compare 300mm with 200mm and 90nm with 65nm wafers I assume this is processors. I don't see any reason to link this since this projection has been around for more than six months.

Everything I had read I believe said 65nm.

?? I'm sorry but this assertion makes no sense. FAB 38 won't be running until 2008. It would be insane for AMD to fill it with 65nm tooling. The tooling will be 45nm. AMD will start it at 65nm and then ramp to 45nm. It will probably follow FAB 36 but not by much.

1. What happens if K8L does not match Core 2's performance, or is just on par?

That was my assumption. I don't think K8l will be any faster on the desktop or for 2-way servers.

2. What happens if AMD can not increase capacity as fast as planned?

That was my assumption too, that production will be tight for AMD all through 2007. That is why if the market were to expand rapidly, AMD would definitely lose volume share to Intel.

You need to understand that the desktop is now the least profitable area. The fact that Intel leads here doesn't help as much as it used to.

Scientia from AMDZone said...

Dr. Yield
But at what cost? I've posted on a lesser site an analysis that leads me to believe that for every 1 K8L produced, AMD loses capacity to produce 3+ 65nm X2s.


You need to look at history. When AMD started making K8 these chips were twice the size of K7; the same argument applied. AMD will also produce dual core versions of K8L which will be smaller.

Probably- but if they piss off OEMs and the channel by starving mainstream part supply

Sounds like you've been reading all those negative supply channel stories about AMD. The channel should be back in shape soon assuming it isn't already.

As to the nature of the high margin HPC market- think about it in gross numbers. Assume you sell a dozen 1024 CPU systems in a year (they are uncommon).

I think AMD sold that much last year for just one system. You are forgetting that AMD is 2nd in HPC behind IBM's Power. However, AMD is rising the fastest while Power is falling off. AMD is in a very good position in HPC and should climb into the #1 spot by the end of 2007.

Scientia from AMDZone said...

qurious
ASP are under pressure from the price war and the deal that Ruiz made with Dell was signed in blood not to mention that Ati now comes into the mix and their margins are anemic to say the least so I would say for 2007 margins for Amd will at best remain flat.


I'm sorry but this is incorrect. AMD's costs in 2007 will drop twice as fast as Intel's unless Intel begins another round of divestment. AMD's margins should slowly rise as 65nm 300mm production becomes dominant.

Scientia from AMDZone said...

enumae
4x4 is an AMD platform, and as such should be factored in when comparing Intel and AMD systems.


I don't think that 4x4 is putting much pressure on Intel right now. Secondly, this is such a small fraction of AMD's revenue and volume that you can pretty much ignore it. However, AMD has stated plans to upgrade 4x4 twice. This system may actually force Intel to release a similar platform after the first upgrade.

enumae said...

Scientia said...

"I'm sorry but this assertion makes no sense."

I made no assertion. This is what AMD is saying...

-----------------------------

AMD will ramp down 200mm manufacturing in the second half of 2007, with preparation already underway for the ramp of 300mm wafers on 65nm process technology at Fab 38 by the end of 2007. Through the combination of leading-edge equipment, Automated Precision Manufacturing (APM) and the great people of Dresden, the plant will produce the latest generations of AMD microprocessors, reaching full capacity by the end of 2008. The majority of the investment will go into new equipment in the Fab 38 facility. AMD will also build a new clean room facility on its Dresden campus for Bump and Test requirements which will support both fabrication facilities. Previously the clean room facilities for Bump and Test activities were located within Fab 30 and Fab 36. By moving them into a new facility in 2007, AMD has the ability to maximize production space at both Fab 36 and Fab 38, providing increased output and capacity. These three projects have the potential to increase Dresden-based manufacturing to a full capacity of 45,000 300mm wafer starts per month by the end of 2008.

-----------------------------

"FAB 38 won't be running until 2008. It would be insane for AMD to fill it with 65nm tooling. The tooling will be 45nm."

Your comments would suggest that it will be 45nm from the satrt, and that would be incorrect, and is why I am asking for a link to clarify your statement.

"AMD will start it at 65nm and then ramp to 45nm."

Well that contadicts your previos statements, so please clarify.

Scientia from AMDZone said...

Red
Blah, blah, blah, benchmarks are irrelevant or rigged. Care to show us some examples?

Please cite your source when you reference the obscure.

All you are doing is just blindly believing anything AMD feeds you without actual physical evidence.

If that isn't silly, I don't know what is.

Wow, thanks, and this whole article is based on rumor and speculation.


Post deleted. You are welcome.

Scientia from AMDZone said...

Red
Scientia does not like gaming benches because it's supposedly GPU limited, encoding benches because they are supposedly hard drive limited, apps that don't scale to his typical 95% because they are "bad".


Hmmmm.

mid07=probably paper launch late Q2, early Q3 at best.

What is your basis for this?

Again, link to the K8L supercomputer please.

Why bother? At last count AMD had four supercomputer contracts.

That 25% would mean something if it actually existed. I think it's Scientia's wishful thinking:P

Really?

He basically said that AMD takes baby steps with their process.

AMD did the same thing with 90nm.

I can't find where you talk about AMD's E6700 competitor:(

One doesn't currently exist.

Rather silly to try and dismiss 1 rumor by showing that rumors have been wrong in the past when once again, this whole entry is speculation.

Second post deleted.

Scientia from AMDZone said...

Wallachian
AMD's dual core K8L's will probably be their mobile core, clocked upwards into the desktop realm.


No. The mobile core will be different in 2007.

But other than this, it doesnt look like AMD can do much to create new momentum,

Momentum is a myth. People keep trying to suggest that Intel is gaining momentum but there is no evidence of this. Also, my estimate is that AMD will gain at the rate of 0.6% per quarter. This isn't exactly momentum but it would add up after awhile.

until they get to 45nm themselves

No. I expect AMD to keep gaining during 2007.

with new quad-cores, 8-cores and their fusion plans.

The quad core is the same on 45nm; the main difference is DC 2.0. I'm not sure about any 8 core plans.

They will probably have to settle to being no.2 for the next year as well.

AMD will be doing well if they can get up to 3rd or 4th on the top 20 IC manufacturers list by end of 2008.

Scientia from AMDZone said...

enumae

What reason would AMD have to buy new 65nm equipment that will begin operating in Q1 08 and be obsolete by the end of the year? AMD first started producing 90nm on FAB 36 and then later switched to 65nm. All of the equipment that goes into FAB 36 after mid 2007 will be 45nm. Why wouldn't AMD do this for FAB 38 as well?

enumae said...

Scientia said...

"...What reason would AMD have to buy new 65nm equipment that will begin operating in Q1 08 and be obsolete by the end of the year?"

Thats why I am asking for a link, I completely understand your point, but... are you speculating or do you have a source?

If you have one please link it.

"...All of the equipment that goes into FAB 36 after mid 2007 will be 45nm."

That I do not doubt, but again please show a link depicting AMD's 45nm capital spending.

"Why wouldn't AMD do this for FAB 38 as well?"

Well off the top of my head, they could take quite a bit of the old 65nm equipment from Fab 36 and continue to use it in Fab 38.

----------------------------------

Your logic is sound, but without a source it is just speculation.

Red said...

Scientia, why are you deleting my posts? I mean, I can't help but jab when one says the obvious XD


Still waiting on the link for 25% and K8L supercomputer in Q207.

Greg, what makes you think Vista lessens the margin? From that 1 test, I would say that Vista fixed the Numa issues, but we need more reviews on Vista before we can say that Vista helps K8.

All this complaining about the same apps over and over again, any suggestions onto what they should bench then?

I'd like evidence that Dell is stuffing their systems with Brisbanes rather than assumptions. By now, one should be able to find systems with the 65nm exclusive X2 4400+/4800+, no?

Just wondering, how bad would Brisbane have to be for you consider it to have problems and the sort?

That's nice that AMD is having their best year. Not hard to beat when they were constantly bleeding and eeking out profits and now just eeking:p I'm still confused as to why you think a cumulative end of year report has any significance.

1% better, 10% better, maybe I'm blind but I see nothing competitive from AMD above the X2 4200+.

Scientia from AMDZone said...

Red.

Supercomputers:

1. TACC for 2007.

2. Roadrunner for 2007.

3. NERSC for first half 2007.

4. Baker in 2008.

Scientia from AMDZone said...

Red
Scientia, why are you deleting my posts? I mean, I can't help but jab when one says the obvious XD


You can jab; just be civil about it.

and K8L supercomputer in Q207.

Done; see last post. Maybe I'll look up the 25% reference although this should be fairly easy for you to google.

Just wondering, how bad would Brisbane have to be for you consider it to have problems and the sort?

Well, if AMD's volume production for 65nm were at, say, 15% that would be bad.

That's nice that AMD is having their best year.

And, Intel has slipped to 2004 levels.

Not hard to beat when they were constantly bleeding and eeking out profits

Sorry, but this is incorrect. Since late 2003, AMD has only had one quarter when they didn't earn a profit.

and now just eeking:p I'm still confused as to why you think a cumulative end of year report has any significance.

AMD's cpu revenue will have doubled since 2004 while Intel's will roughly equal 2004. That seems significant to me.

1% better, 10% better, maybe I'm blind but I see nothing competitive from AMD above the X2 4200+.

4600+ is still competitive. 5000+ could be if the price comes down.

Red said...

As to the idea that AMD won't deliver K8L on the desktop until Q4 07, AMD has contracts with one supercomputer that will be running by end of Q2 07 so K8L production has to start in Q2. Desktop versions of K8L will come out in Q3.

Ehh? End of Q2 is basically Q3. You are using supercomputer prodution dates to say that desktop K8L will come like the rumors suggest. But remember, the rumors have been wrong XD

1. TACC for 2007.

2. Roadrunner for 2007.
According to IBM, 2008.

3. NERSC for first half 2007.
"..with two cores per socket.."

4. Baker in 2008.
Your own date speaks for itself. I did not ask for AMD hype, I asked for K8L in Q207.

TACC is the only solid one that was actually a response to my question. With initial production on it starting in June (and them plugging the quads probably in Q3), K8L production on AMD's side in Q2 doesn't seem unreasonable. They are going to have to do better than run task manager though:D I wouldn't be surprised either way if Kuma comes in July or October, a rather moot point to defend anyhow.

Red said...

What does % volume have any significance to the technology when it's 75nm, slower cache, slower clocks? Not hard to have decent yields when your new stuff isn't better than your old stuff.

AMD's best year ever, I thought that pointing out their previous bleeding as well as their consistently modest earnings made it easy to have a good year.

AMD's growth is significant, but AMD's stock isn't going to increase 85% back to year ago levels just because their report card is going to look cute, nor is Intel's going to tank because it doesn't look as pretty as 2005's.

The X2 3800+ is a steal. Unless you want raw perf:p. The X2 4600+ is not so much. Please show me a review where it makes itself worthy. It's not that bad, but if you're going to spend $200+, might as well go Conroe.

Wallachian said...

greg
Again, I'm sure we've mentioned this many times before, but not all transistor technology can immediately make the process shrink and work. Really, what their transition to 65nm has shown is how good their 90nm process has become with all these improvements, and how much we have to look forward to, which is why AMD wont be running 150 watt TDP when they do quad core at 3ghz.

Also, we have no example of Intel transitioning a core over to 65nm except for netburst, which did not gain significantly. Chances are, AMD is increasing their stepping in K8L (since that's what every company does with a new process) so I could easily see them making much high clock speeds on that process.

Netburst did gain when it moved over from 90nm to 65nm. A top bin smithfield dual-core was 3.2Ghz HT, 800Mhz FSB at 130W, while a 65nm Presler was 3.73Ghz HT, 1066Mhz FSB at 130W. A 3.6Ghz Presler with 800Mhz FSB was rated at around 95W. See, when you stick to a TDP, your clock-speeds get limited.

The reason Intel does not want to deliver higher clocked versions of Conroe is they want to stick to a 65W-75W TDP. If they decided that like AMD, they would like to have a 125W TDP like the FX-62, Conroe would surely be at 3.46Ghz now, and then the comparisons wont look nice.

If AMD's 90nm looks so good, its only because AMD is stretching it to its absolute limits. AMD can definitely deliver a 3.2Ghz 90nm today at 125W, this is not because of some continuous transistor improvement, its simply because of "cherry-picking".

Though AMD may use SOI, while Intel uses strained-silicon, the only differences there is between the two is subtle power-reductions, current-reductions, improvements in leakage and drive current (the odd optimization here and there in timing). Subtle optimizations, but by and large, the laws of physics are the same between the two, and two similar architectures like Core and K8 will have nearly (read exactly the same) clock speeds at the same TDP. AMD will probably get much better idle power, and power optimization due to their single-die nature, where all cores are controlled. Idle power gets much better with better processes, but the max clock speed at a particular TDP remains the same. That is why AMD's quad-core and Intel's quad-core will have nearly the same clock-speed at the same TDP.

AMD does not look better at 65nm than at 90nm because ideally they should have made it to 65nm the moment TDPs slipped past 89W, that was around 2.4Ghz. At 65nm you are now getting 2.6Ghz at 76W (of course they need to optimize). It doesnt mean their 90nm was so great. It was just that Intel was so bad with all the Netburst stuff, that AMD just sat back and kept milking a process for all its hassle value. And now they are scampering to get 65nm off the ground.

Scientia from AMDZone said...

Red

One of the contracted supercomputers is due to be online at the end of Q2. It may not be one I've listed because there have been several. I may or may not be able to find a reference.

AMD's best year ever, I thought that pointing out their previous bleeding as well as their consistently modest earnings made it easy to have a good year.

Yes, and your point is still incorrect after repeating it. Maybe you'll be able to understand it this time. AMD has made a profit every quarter except one since Q3 03. So, where do you get the bleeding idea from? Do you have a link or reference?

Scientia from AMDZone said...

Wallachian

that AMD just sat back and kept milking a process for all its hassle value. And now they are scampering to get 65nm off the ground.


Not true at all. There was never a time when AMD sat back milking a process. Where do you get such ridiculous ideas from? AMD actually spends proportionately more on R&D than Intel and much more on capital expenses.

enumae said...

Hey Scientia just wondering if you missed my earlier post...

Scientia said...

"...What reason would AMD have to buy new 65nm equipment that will begin operating in Q1 08 and be obsolete by the end of the year?"

Thats why I am asking for a link, I completely understand your point, but... are you speculating or do you have a source?

If you have one please link it.

"...All of the equipment that goes into FAB 36 after mid 2007 will be 45nm."

That I do not doubt, but again please show a link depicting AMD's 45nm capital spending.

"Why wouldn't AMD do this for FAB 38 as well?"

Well off the top of my head, they could take quite a bit of the old 65nm equipment from Fab 36 and continue to use it in Fab 38.

----------------------------------

Your logic is sound, but without a source it is just speculation.

Greg said...

To one of Red's old points, you don't just get to pick (eeny-meeny-miney-moe) whether or not you want to invest in a new fab or if you want to throw money at a die shrink to try to (fruitlessly) make the transition to it go faster, or if you want to buy the second largest discrete graphics card company. Either way, for AMD to invest more heavily in product research than diversity would be highly idiotic. While AMD devotes slightly more than 10x its percentage of income to R&D compared to Intel, it's still far less than 10% of the amount of money Intel pours into R&D.

AMD will generally trail in terms of raw transistor, processor, and architectural performance unless they get lucky again (like with netburst), and even then, that can only last so long, and only give them so much ground. This is why AMD has concentrated so much on platform innovations like IMC and HTX, so it can make partners far more exclusive and avoid having to rush development to overcome forseen prolems like bottlenecking, cache size increases, power consumption, and stingy partners, so that they don't have to pump any more money into their development in order to rush them, and so that if Intel does run into an unforseen hitch they'll be ahead for a short while yet again. It also needs to make sure it's not playing Intel's game, so that Intel doesn't control where AMD is forced to go, and instead, AMD can start controlling where Intel is forced to go.

Walachian, AMD has a very robust and extremely active transistor improvement program. and if it didn't yield what I described, then I'd see no reason for AMD to implement it at all as I'm sure it's a somewhat expensive process compared to not using it at all.

Also, seeing where AMD's percentages of production are would easily show whether or not it's cherry picking at all (which is not likely due to the massive number of 5000+ and or 5200+ processor being bought up by Dell and others like it (which is why they were and are so hard to find).

Wallachian, my point is as above, pouring money into process development plans very rarely makes the transition to them any faster, and even when it does, the gains are minute.

Red said...

If you can't find the supercomputer that's going to be operational at the end of Q2 (Q3), that's about as useful as referencing your 25% 65nm.

If you're going to say that it's their best year ever..relative to all years, rather than just going back 3 years, they weren't exactly financially stable before (Google: AMD earnings loss *various years*), therefore, not impressive that they are having their best year ever. They were disappointed in last quarter's results (could have even better 2006) and are expecting a better 2007, right?

I don't see the point of AMD accusing Intel of monopolizing (not that I don't believe they don't) when they're basically selling all that they can make. Buying ATI is not going to give them capacity anymore than having a superior architecture.

% volume isn't indicative of process performance when the 2.6GHz X2 5200+ isn't exactly pushing the borders thermal/clock wise. Also, you don't know how much of those the OEM's are buying, they could be scraping the few that are made or just hoarding them either way.

I'd have to say that AMD is doing some serious cherry picking with the FX74 running at 1.4V+ yet normal chips can't get that far. It befuddles me how they're going to make 3.2GHz..unless they all magically sell out to OEM's like the 5000+ and sell for $1200 retail:p

Wallachian said...

Another curious move last year by AMD, was the AM2 revision. A good solid business move would probably have been, continue to release 2.8Ghz, 3.0Ghz products on socket-939, while developing a 65nm version that has lower power consumption and a ddr2 memory controller. While what AMD did makes sense to people (partly because its already done and over with), it really gives the impression that they took 6 months longer than normal to begin the transition to 65nm. (I believe sometime in the latter half of 2005, people were forecasting AMD's move to 65nm at around Q3'06).

The quad core is the same on 45nm; the main difference is DC 2.0. I'm not sure about any 8 core plans.

Not only DC 2.0, but also integrated PCIe. I believe there was a slide that mentioned they were developing a 8-core, and after that they were focussing only on accelerated processors, therefore no longer engaging in a core-war.

Greg said...

My point is, Red, that AMD either had to choose to finally aquire a way to make a mobile platform so Dell and the like will finally want to buy entire systems from AMD and be able to rely on them, and actually diversify their product lineup so that they no longer have said financial dips, or to try to spend money on something that will allow them to meet demand that they actually may not have and that wont actually let them address that for the better part of a year if not more. So ya, buying ATI was far smarter than trying to spend all the money they spent on ATI in fabs.

Greg said...

Also, AMD isn't saying Intel is monopolizing now (any idiot knows that). AMD is saying that Intel's anti-competitive practices such as planning to pull product lines from distributors that didn't only sell Intel or specifically exclude AMD or to increase prices to ridiculous amounts for the same reasons.

It's widely known that Intel used these practices too, the only question, is whether or not they're actually illegal in the united states. It doesn't matter, if AMD is doing "better" now, they're still a far cry from a highly-competitive threat to Intel's dominance. The fact we're even impressed with AMD nearly having 1/4 of the entire x86 market, and Intel still owns the entirety of the rest of that is pretty impressive in and of itself. Also, the fact that AMD could be doing substantially better now if it weren't for such busines practices is far more important than how they are doing right now (if that makes any sense).

Greg said...

Oh, and Red, sorry, if it sounds like I'm calling you an idiot, I'm saying that any idiot knows that Intel is no longer monopolizing, or at least not monopolizing to the extent it once did.

Kalle said...

"Also, we have no example of Intel transitioning a core over to 65nm except for netburst, which did not gain significantly"

65nm Netburst had vastly reduced power usage and it OC'd a lot better. 2.8@4.5GHGz 24/7 stable with only +0.1V, anyone?

"Chances are, AMD is increasing their stepping in K8L (since that's what every company does with a new process) so I could easily see them making much high clock speeds on that process."

Their own roadmaps show 2.5GHz as maximum at release.

"Oh, come on... What Intel managed to do with 10 (or even more) times more R&D money?"

They earned even more money? Every commercial company has only one goal: make their owners rich. While Intel had almost no meaningful competition they just stuffed the money to their pockets. AMD has to innovate and bring lots of new stuff to table to be able to compete.

"However, AMD is rising the fastest while Power is falling off."

Are you sure? top500 list shows that A64 and Power are both increasing their share mostly from taking it away from IA-32. With Power6 and Cell2 coming I'd expect Power to gain additional share of that pie.

"That is why AMD's quad-core and Intel's quad-core will have nearly the same clock-speed at the same TDP."

As I've said before, K8L will clock lower at release than first C2Q clocked about two months ago.

Scientia from AMDZone said...

Red
I don't see the point of AMD accusing Intel of monopolizing


If you've read my article on why this is not like 2002 then you should know that I disputed the notion that AMD's drop was due to monopoly practices by Intel. When I mention monopoly in my recent article this refers to Intel's ability to create standards as it did when it crushed AMD's 3DNow extensions by creating SSE. Without monopoly the earlier 3DNow would have become the standard. The one example we have is Intel's following of AMD64. If JEDEC does indeed extend the DDR2 standard to 1066Mhz then this would also be evidence that AMD is now able to create standards of its own.

BTW, you probably are not familiar with American trade law but having exclusive contracts is a violation of the Sherman Anti-Trust Act which was created to break previous monopolies like Standard Oil. Since Intel did have exclusive contracts it could reasonably be said to be a monopoly. These exclusive contracts were only broken after AMD began the lawsuit.

Scientia from AMDZone said...

Kalle

I would have to agree about 65nm netburst. Presler was much better than Smithfield in terms of heat of power draw. Tulsa does gain some additional power due to the use of a hybrid bus.

Are you sure? top500 list shows that A64 and Power are both increasing their share mostly from taking it away from IA-32. With Power6 and Cell2 coming I'd expect Power to gain additional share of that pie.

Yes, I'm absolutely certain. The top500 list shows that Opteron is taking share in HPC and so are Intel's 64 bit processors. It does not show that Power is taking share. In fact, Power has dropped for the last year (two lists).

I have no idea about how clock speeds will compare with AMD and Intel quad cores. AMD should be competitive at the same clock; it may not be as competitive if Intel is clocking higher.

Scientia from AMDZone said...

Red

You are arguing that I am engaging in rhetoric without proof. However, you then give casual counterarguments of your own without any proof. You don't get to have two standards. You have made several assertions without references. If you continue to criticize me for rhetoric while giving rhetoric of your own, your comments will become irrelevant to me.

If you can't find the supercomputer that's going to be operational at the end of Q2 (Q3), that's about as useful as referencing your 25% 65nm.

I see. Well, it will be known in just two quarters whether or not a supercomputer is taking delivery of K8L's from AMD.

therefore, not impressive that they are having their best year ever.

The reality is that AMD is one of the largest electronics companies in the world. I just don't see the logic of excusing Intel's poor performance by suggesting that they were so big that they had no room to grow and then trying to diminish AMD's impressive growth by suggesting that they were doing so poorly that doing better was nothing.

The truth is that AMD has outperformed every other electronics company of similar size. So, I guess the only way that your argument would work would be to insist that every other company was even more financially unstable.

enumae said...

Scientia, I am still waiting for a respomse/link/source from you...

Scientia said...

"...What reason would AMD have to buy new 65nm equipment that will begin operating in Q1 08 and be obsolete by the end of the year?"

Thats why I am asking for a link, I completely understand your point, but... are you speculating or do you have a source?

If you have one please link it.

"...All of the equipment that goes into FAB 36 after mid 2007 will be 45nm."

That I do not doubt, but again please show a link depicting AMD's 45nm capital spending.

"Why wouldn't AMD do this for FAB 38 as well?"

Well off the top of my head, they could take quite a bit of the old 65nm equipment from Fab 36 and continue to use it in Fab 38.

----------------------------------

Your logic is sound, but without a source it is just speculation.

Roborat, Ph. D. said...

From a survivability stand point AMD should be fine for another 3-4 years. But from a profitability point of view, they’re in a lot of danger. They mentioned themselves that the ATI merger is dilutive to AMD’s EPS. ATI will be bringing in a huge loss this Q4’06 and will continue to be a drag until AMD completes its “synergy” restructuring. Without the Intel license ATI really should be cut down to ¼ of its size to be profitable.

AMD’s 2007 gross margin will be in the 40’s again because of this merger without any concrete plans of getting back to 50s in the short term. This just shows you Intel’s ability to maintain margins. It’s probably the reason why Intel refuses to play the discrete graphics game. With the current investigation on price fixing between ATI and NVDA, expect their margins to go down even further.

It is very much likely that AMD can go back into the red in Q1’07. Bearing in mind Fab36 is now year+1 with depreciation cost for AMD going to $1.5B up from $700M in 2006. And with current OEM inventories piling up, it’s almost guaranteed that Intel will pick up market share again.

Your analysis is flawed simply because you’re only looking at the small piece of the revenue pie.
25% of Intel’s production which is Core2Duo is enough to suffocate AMD from the high ASP segments. Both Intel and AMD pump out a significant volume of mainstream CPUs for the commercial segment. While this provides the market share numbers which will most likely be flat for both, expect a big shift in revenue share as all of AMDs products are driven down 3 steps from the high-end segment. How come you never mentioned this as a fact? Unlike 2006, AMD is going into 2007 with an anaemic product price points.

Looking farther ahead, I wouldn’t dismiss Intel’s coming 45nm and 32nm nodes. This time it is different, Intel has the performance crown and any improvement Intel does just adds to its performance lead.

Kalle said...

"In fact, Power has dropped for the last year (two lists)."

Fact?
http://top500.org/stats -> processor family

Last three years/6 lists. All numbers are in percentages:
AMD64: 6.00, 5.80, 5.00, 11.20, 16.00, 22.60
Power: 15.60, 12.40, 15.40, 14.40, 16.60, 18.20

Where did you get your facts?

I don't argue that AMD has had massive growth but Power doesn't show much sings of getting weaker either. With Power6 I only expect them to grow their share, though perhaps not as fast as AMD has done it.

Roborat, Ph. D. said...

From Wallachan:
... Though AMD may use SOI, while Intel uses strained-silicon, the only differences there is between the two is subtle power-reductions...

you're forgetting SOI cost more than twice the price of bulk silicon. Not helping when there's currently wafer shortages as well.

From Wise Investor:
A: Build another Fab. like 1 in New York with 650mil grant. If that's not enough there are numerous ways to raise capital. Like issuing more stocks. Just like all growing enterprises do.
It takes about 5 years to build a fab and cost at least $5B. I believe AMD will more likely lean on the subcons for more output.
But I wouldn't want to imagine subcontracting 90nm with 9 interconnet metal layers on SOI.

Greg said...

Actually roborat, the larrabee group shows Intel is being forced to play the discrete graphics game, and the fact that you do not know what AMD's plans are to get back to the 50s, does not mean that they do not have any, as you obviously aren't on the inside of AMD's information chain.

While Intel does hold a good portion of the market with high ASP's, the only portion of the market with high ASP's and high sales numbers are servers, which AMD will not lose very quickly due to the extremely slow nature of the server market and Intel's constant changing of platforms. Core2 is no longer in the high ASP portion of the market except for the x6800 and the qx6700, which are both very small portions of the market. As such, and because AMD still is the strongest option in the 4p market, AMD will not suffer massive changes in their revenues (which is odd, as it seems you were the one accusing people of only looking at small portions of the revenue pie).

I would dismiss their 45nm node and especially 32nm node for now, because we definitely don't know what arch they'll have at 32nm, and we don't know exactly how AMD will be executing its modular core architectures by the time 45nm comes into its own. On top of all of that, we don't know exactly how barcelona will fair against core2, much less penryn. Also, with AMD apparently finding ways to accelerate the slow process of churning out new process sizes (probably by cooperating much more closely with IBM) we definitely have to reconsider how we analyze Intel's process advantage.

I think Power is having gains do to interoperability with opteron and the fact that Cell is probably considered Power.

Roborat, give us a link showing the cost of Strained Bulk SI and the cost of SOI. Also, make sure it's one that actually shows AMD's cost in implementing SOI and actually shows Intel's cost of implementing Bulk strained. If you can't find that, then find something that shows estimated costs, so we can at least analyze that.

Roborat, building a fab is not something that you can just forego by leaning on subcontractors. Subcontractors are much more expensive in terms of producing processors, and normally as long as they're getting money, subcontractors don't care what they have to produce, especially if it's a big company like AMD (since not anybody can be as big as Intel, as long as Intel has something they can say about it). Point being, subcontractors are only a temporary solution when you cannot meet all the demand for your product.

Red said...

Thanks for reminding me that even if AMD can't sell anymore chips if Intel wasn't evil, evil is still evil..:)

Scientia, I've only asked for obscure references which you obviously have a hard time finding because they don't exist.. The only obscure thing I have referenced is my 90nm/65nm/45nm chart, the others are easily searchable and if you can't find them, ask me.
http://www.amd.com/us-en/assets/content_type/DownloadableAssets/Dec-06A-Day20BobRivet.pdf
Slide 16 is the chart btw.

Greg said...

Though it's really hard to figure out what your first paragraph means red, I think it's a good point (though a highly unlikely one).

enumae said...

Scientia, I am just wondering if you are going to answer my previous post?

Thanks

Scientia from AMDZone said...

Roborat
But from a profitability point of view, they’re in a lot of danger.


Actually, they are not.

ATI will be bringing in a huge loss this Q4’06

Where did you get this from?

Without the Intel license ATI really should be cut down to ¼ of its size to be profitable.

No. You don't seem to understand what percentage Intel was of ATI's business; it was not 75%.

It’s probably the reason why Intel refuses to play the discrete graphics game.

You mean Intel disbanded its huge discrete graphics business? When did this happen?

It is very much likely that AMD can go back into the red in Q1’07.

Why?

Bearing in mind Fab36 is now year+1 with depreciation cost for AMD going to $1.5B up from $700M in 2006.

What are you talking about? 1.5 bilion in depreciation? That is truly absurd.

And with current OEM inventories piling up

Where did you get this from?

it’s almost guaranteed that Intel will pick up market share again.

Or not.

25% of Intel’s production which is Core2Duo is enough to suffocate AMD from the high ASP segments.

Incorrect. AMD only makes a tiny bit of revenue from the highest speed grade plus FX and this is all that C2D is affecting.

expect a big shift in revenue share as all of AMDs products are driven down 3 steps

Nonsense. You are drastically overestimating C2D's importance.

Scientia from AMDZone said...

Kalle
Where did you get your facts?


The Top500 list.

It hadn't occured to me that you were only looking at the percentage of systems (which is largely irrelevant). Let's look at all of the statistics:

For the last year (two lists).

% Total systems
This is the only area where Power has had an increase. They went up less than 4% while AMD increased more than 11%.

Computing Power/system
Power - sharp drop
AMD - slight increase

Processor/system
Power - sharp drop
AMD - up slightly

% Total Processors
Power - down
AMD - up sharply

% Total Computing Power
Power - down and down faster on the last list
AMD - up sharply

Power doesn't show much sings of getting weaker either.

It does in every other way. There were a few more Power based systems but they contained far fewer processors so Power is being pushed down in the list.

With Power6 I only expect them to grow their share, though perhaps not as fast as AMD has done it.

Power would need a significant turnaround just to stop losing ground. Gaining is unlikely.

Scientia from AMDZone said...

Enumae
Scientia, I am just wondering if you are going to answer my previous post?


65nm production does continue until Q4 09. So, I suppose it is possible that AMD could move equipment from FAB 36 to FAB 38 and not bother buying new 45nm equipment for FAB 38 until FAB 36 is fully upgraded.

However, the graphs show that both FABs will be producing 45nm on 300mm by Q4 08. So, I'm not sure how this would work. It is possible though.

Scientia from AMDZone said...

Red
The only obscure thing I have referenced is my 90nm/65nm/45nm chart


Yes, that chart is different from the chart from June. So, either one of the charts is wrong or AMD altered their projections since June and pushed the schedule back by 1 1/2 quarters. Presumably this is your assumption.

enumae said...

Thanks you for responding :)

Red said...

Greg, what I was trying to get at was that regardless of whether AMD could sell more or not (barely juggling Dell+channel, now instant rebating back their support), behavior like undercutting solely for the purpose of running out others, etc. should not be tolerated.

Obviously the more recent one ire more accurate. One can hope though ;) I've gone through all of the June slides. Which presentation and which slide # are you talking about that shows a better 65nm schedule in the first place?

Scientia from AMDZone said...

Red

This should be it. Daryl Ostrander June Analyst Day. Page 5.

Scientia from AMDZone said...

In this slide it shows 200mm production ramping down in Q2 07 whereas the newer slide shows 200mm ramping in Q4 07.

I've been trying to find a transcript of the December Analyst Day Q&A to see if there was clarification for this but I haven't found one yet.

Greg said...

Actually, I have no problem with undercutting. In fact, undercutting is a really hard sell in terms of suing people for it. The problem, is that Intel signed agreements with companies that were basically death threats in terms of those companies viability if they ever dared to sell AMD processors. AMD couldn't do anything about it, because they couldn't produce enough volume because they'd never had enough capitol to buy and start that many fabs, because they were basically just getting into the business (they started out because IBM needed multiple manufacturing supplies for Intel processors, which is pretty funny).

Also, could you be more clear about the whole instant rebate thing, because I haven't heard anything about it.

Scientia from AMDZone said...

Red

BTW, you do have a point about capacity. As far as I can tell AMD's capacity will top out in Q4 09 when FAB 38 is fully tooled. Even if AMD starts a new FAB in 2007 it won't be finished for several years. I'm not quite certain why it would take until 2012 since none of the other FABs did but this would leave a gap of 3 years with no growth.

It seems to me that if AMD wants to continue growing then it should do something to add capacity sooner like building another FAB in Dresden. Presumably, this could be online in early 2010.

Red said...

Sorry Sci, I was hoping for a hard quote of "25% of our chips will be 65nm by the end of the year" or something like that. They have 300mm 90nm already anyways, no?

I also have no problem with AMD undercutting Opterons with Quad FX, but undercutting yourself solely to destroy your competition is what I'm talking about. Of course there are other supposed Intel monopolistic acts, like your aforementioned death threats, etc.

http://techreport.com/onearticle.x/11498
http://www.theinquirer.net/default.aspx?article=36129
Pre-December, X2 3800+ was $150 or so. It dipped below $130 on Newegg during December and it seems the "rebates" (price cuts) are still basically in effect. Funny, a couple months ago, some said it was great that AMD was selling everything it could, channel or no channel. Now they are trying to lure them back with lowered prices, these same chips that were such hotcakes when Dell was hoarding.

They say that they will have 4x capacity of today or something by 2009. Offset by multi-core and Intel increasing its capcity also. Do you have AMD's percentage of dual core production for comparison btw?

Greg said...

I think AMD is trying to tier its growth into secions of years, so it doesn't have to bet the farm on its growth, and doesn't end up with fabs it can't use if something bad happens and they lose a substantial amount of demand.

Wise lnvestor said...

Roborat: It takes about 5 years to build a fab.

Wrong! That was the case...When it was 5 years before.

AMD's New York fab should be complete
by the end of 2009(or even before that). I have 2 sources. 1 from eetimes.com and another from fabtech.org

www.eetimes.com

www.fabtech.org

Both state groundbreaking in June-July 2007. Cost about 3.2-3.5 to build while fabtech states:"Over US$5 billion could be spent"

It may sound like your suggestion of 5 bil at least is on target. But
details revealed that AMD actually gotten 1bil-1.2bil incentives!

I did some research on those incentives. Here they are, look in the 'Site Attributes' section.
LUTHER FOREST TECHNOLOGY CAMPUS

BTW scientia the 2012 date was wrong. The arthur of that article(Wolfgang Gruener I believe)had made a numerical misteke, but never publish a correction.
Fab in 2012

Here I found another article. This time it is AMD's press release.
AMD's press release
"Similar to AMD's other fabrication facilities, construction of the building is expected to take about a year with another year needed for the installation of equipment and tools."

I pay special attention to these details. After all, I'm the guy who sold AMD @ 25.81

Roborat, Ph. D. said...

Roborat: But from a profitability point of view, they’re in a lot of danger.
Scientia: Actually, they are not.

AMD held an Analyst day last December. You need to read the foil presented by AMD. It’s in their website. The merger is described as dilutive – i.e., reduced earnings per share. Add higher operating costs to ATI’s losses and price pressure from Intel, that basically spells AMD not reporting a profit to me.
http://www.amd.com/us-en/assets/content_type/DownloadableAssets/Dec-06A-Day20BobRivet.pdf

Roborat: ATI will be bringing in a huge loss this Q4’06
Scientia: Where did you get this from?

From AMD. ATI reported a loss of ~$50M last Q3. It’s gone bigger in Q4. How much profit does AMD make every quarter? You do the math.

Roborat: Without the Intel license ATI really should be cut down to ¼ of its size to be profitable.
Scientia: No. You don't seem to understand what percentage Intel was of ATI's business; it was not 75%.

Some investor made a comment at 60% coming from all intel platforms. ATI margins are at ~24%. Again, you do the math to target it at AMD’s goal of 50%.

Roborat: It’s probably the reason why Intel refuses to play the discrete graphics game.
Scientia: You mean Intel disbanded its huge discrete graphics business? When did this happen?

No, I meant Intel’s refusal to enter and make high-end discrete graphics. Margins aren’t healthy much like Xscale. AMD on the other hand now targets only above 40% margins for 2007 by taking in ATI (as per AMD).

Roborat: It is very much likely that AMD can go back into the red in Q1’07.
Scientia: Why?

Again, due to ATI’s losses plus AMD’s slow revenue growth compared to Operating costs for 2007. It’s a prediction which I think can happen. No one seems to think AMD can increase revenue as quickly as operating costs. See AMD’s foil or any financial analyst out there:

Roborat: Bearing in mind Fab36 is now year+1 with depreciation cost for AMD going to $1.5B up from $700M in 2006.
Scientia: What are you talking about? 1.5 bilion in depreciation? That is truly absurd.
Again, the numbers are from AMD. It’s actually 1.3B, my mistake. I don’t know why it shouldn’t when they have a new fab & capital equipment depreciating at 20% per year. Starting at year 1 until year 5. Accounting 101.

Roborat: it’s almost guaranteed that Intel will pick up market share again.
Scientia: Or not.

You’re right, they may not with the way OEMs are sucking AMD dry. But I would be interested in the changes to revenue share.

Roborat: expect a big shift in revenue share as all of AMDs products are driven down 3 steps
Scientia: Nonsense. You are drastically overestimating C2D's importance.

You seem to be underestimating it. It affected Intel when AMD was in the lead, why should AMD?

Roborat, Ph. D. said...

To wise investor:

I have better source: AMD
page 17.
http://www.amd.com/us-en/assets/content_type/DownloadableAssets/Dec-06A-Day20BobRivet.pdf

"Two year option to commit to constructing 300mm wafer Fab

Fab construction could begin between July 2007 and July 2009

If committed to construction, could be fully operational betweenDecember 31, 2012 and December 31, 2014"

Building a Fab takes 5 years from approval to full site certification.

My point about the $5B fab was to highlight the enormous investment to start one and the decision to start one has to be backed by something lucrative and likely to happen. Not simple because you're under capacity.

Scientia from AMDZone said...

Roborat

The merger is described as dilutive – i.e., reduced earnings per share.


Yes, I know all about the dillution and it is of no consequence.

Add higher operating costs

Sorry, you've got this backwards; the costs go down, not up.

to ATI’s losses and price pressure from Intel, that basically spells AMD not reporting a profit to me.

No. The effect will be small.

From AMD. ATI reported a loss of ~$50M last Q3.

You have the wrong scale. $50M is not a huge loss; that is a small loss. Could ATI have another small loss? Yes.

And, yes, there will be some loss of revenue from Intel. However, you are confusing Intel with Intel platform which are not the same thing.

No, I meant Intel’s refusal to enter and make high-end discrete graphics.

Your statment is contradictory. Low end graphics are integrated; this has nothing to do with high end graphics. The main reason why Intel is not working on high end graphics is simply a lack of expertise.

AMD on the other hand now targets only above 40% margins for 2007 by taking in ATI (as per AMD).

And, again, this is of no consequence.

Again, due to ATI’s losses plus AMD’s slow revenue growth compared to Operating costs for 2007. It’s a prediction which I think can happen.

I don't play games with numbers. You take the gross revenue and gross margin and that is income. When AMD spends a huge chunk of its money on R&D and capital expenditures and the reported earnings fall, that is of no consequence. If AMD reported a reduction in earnings and had to cut its R&D budget and reduce capital purchases that would be significant.

In a similar vein if Intel keeps its reported earnings the same while slashing stock buyback that too is significant. Reported earnings is not the proper yardstick.

No one seems to think AMD can increase revenue as quickly as operating costs.

Okay, what are you talking about? What increase in operating costs? Are you still talking about ATI or are you talking about something real?

Again, the numbers are from AMD. It’s actually 1.3B, my mistake. I don’t know why it shouldn’t when they have a new fab & capital equipment depreciating at 20% per year. Starting at year 1 until year 5.

This isn't a cost; it's an asset loss, Accounting 101. Why did you erroneously call it a cost?

You seem to be underestimating it. It affected Intel when AMD was in the lead, why should AMD?

Affects? Yes, but not three steps.

Greg said...

Also, Intel is investing in, and starting a segment dedicated to high-end discrete graphics that will begin to operate about the same time Nvidia is rumored to come into the processor market. It's called the larrabee group, you can look it up.

Scientia from AMDZone said...

Roborat

I started writing an explanation of how you are looking at this the wrong way but the details get long; I may have to write an article on this.

Red said...

http://www.legitreviews.com/news/3060/
Excluding the money losing ATI..
*Revenue has increased 3% from Q306.
*Operating income is "positive" but significantly lower than in Q306.
*Gross margin and operating income were impacted by significantly lower CPU ASP, which largely offset a significant increase in unit sales.

Thoughts? At the very worst, I thought that seasonalities would make their Q4 at least on par with Q3. Not that much surprise if one were to believe stories like..
http://chip.seekingalpha.com/article/23367

Roborat, Ph. D. said...

Scientia,
I thought you might be interested to see this:

http://tinyurl.com/y26qyg

Grats :)
I'm suprised though why he thinks its a "new blog" hehe.

BTW, AMD pre-warned so I guess we now know who's 2007 prediction is more in line.

Scientia from AMDZone said...

Red
At the very worst, I thought that seasonalities would make their Q4 at least on par with Q3.


It is; it's actually 3% better than Q3 06. Operating income is not important. I have no idea how good this is until I have Intel information to compare with.

Red said...

Revenue is 3% better. Factoring in seasonalities, on top of increased volume, and making less $$$ on that revenue, I'd say that's pathetic and hardly on par. I'd like to add that bad AMD earnings could also signify bad Intel earnings, but it's looking more like AMD slipped rather than the market. Just wondering, how terrible would their results have to be for you to consider them terrible? Then again, AMD bombing earnings is not a surprise so I guess you'll have to set the bar really low:D

Fujiyama said...

I think that Q4 2006 results are the sum of mistakes AMD did this year.
1. They focused on AM2 migration instead of capacity increase and die shrink
2. Production shift to X2 should follow after great Q4 2005 - Q1 and Q2 should follow by maintaining Q4 output, building inventory and falling X2 prices faster.
3. AM2 shift without performance gains is questionable - big effort - small effect I would say.

Scientia from AMDZone said...

I have no idea where you are getting your certainty from about AMD's financial situation. However, this wouldn't be the first time someone rushed to make a negative assessment of AMD.

I need to have something to compare this with and I need more detailed information. When I get that information I'll give my assessment.

Scientia from AMDZone said...

Fujiyama
I think that Q4 2006 results are the sum of mistakes AMD did this year.


Are AMD's and Intel's actual financial information available yet? If they aren't then what results are you talking about?

1. They focused on AM2 migration instead of capacity increase and die shrink

Well, and K8L as well as 4x4 and Torrenza and Pacifica. Your statement about capacity increase is incorrect; FAB 36 is ramping normally. They'll improve the 65nm process later in 2007.

3. AM2 shift without performance gains is questionable - big effort - small effect I would say.

Not really; this matches AMD's historical pattern.

Fujiyama said...

Are AMD's and Intel's actual financial information available yet? If they aren't then what results are you talking about?

Scienta, regarding Q4 profit/revenue warning it is clear that AMD had a much worse quarter comparing to Q4 2005. Did they learn that capacity should ramped faster or doubled? No.

Well, and K8L as well as 4x4 and Torrenza and Pacifica. Your statement about capacity increase is incorrect; FAB 36 is ramping normally. They'll improve the 65nm process later in 2007.

I agree but it happens to slow. Torrenza as an idea but where are these Java/XML accelerators? This concept has no influence on 2006 revenues. No.

Regarding AM2 shift what you mean historical pattern - the consumer should benefit if you change technology. AM2 didn't prove as a better than S939/754. It was (I assume) requested by big vendors to simplify memory orders but has no effect on revenues also.

I would expect another technology way for 2006 which drive sales: QuadFX introduced early 2006, faster 65nm switch, more agressive roadmap for K8 core - all these 5600+,5800+,6000+ introduced along with C2D launch.

This is only assumption but Hector is going to believe or he knows that history likes to repeat and K8L (or else) design are going to give another 3 year of advantage. This also means that 2006 is the survival game - so maybe the plan is simple - "we need to stay alive and focus all efforts on H2 2007 H2 when new era begins..."

gdp77 said...

What will "the new era" be? If u mean K8L, I suppose u r wrong, since this chip will fight very hard to keep up with C2D @ 45nm. According to the data we got up to now K8L will be far less than a "new era". So the survivability game will have to go on up to late 2008 (fusion @ 45nm).

Fujiyama said...

gdp77 - We don't know.
There are two ends of this game - I think.
First is that AMD has no idea how to fight with Intel except Fusion - more functionality for less money. In this case ATI people should help in a new CPU design.

Second that they spend almost two years after X2 introduction for major redesign of the chip.
27 months is a long time - long enough...

Scientia from AMDZone said...

Fujiyama
it is clear that AMD had a much worse quarter comparing to Q4 2005.


Compared to 2005 the rough estimated figure is 1.5% growth. I'm sure how this would be worse. What other information are you going by?

AM2 didn't prove as a better than S939/754

Not true at all. DDR2 uses less power than DDR. Also, it appears taht DDR2 speeds will exceed DDR.

K8L (or else) design are going to give another 3 year of advantage.

What are you trying to say?? Are you saying that K8L isn't out yet and you've already decided that AMD won't be competitive for 3 years?

This also means that 2006 is the survival game - so maybe the plan is simple - "we need to stay alive and focus all efforts on H2 2007 H2 when new era begins..."

No. 2006 is the year that Intel lost most of its FAB advantage when FAB 36 came online. AMD's cost will steadily decrease all during 2007 as 300mm production increases.

gdp77 said...

Not true at all. DDR2 uses less power than DDR. Also, it appears taht DDR2 speeds will exceed DDR.

U must be kidding me... AM2 offered nothing to the end user. Except the money that users paid in order to upgrade their mobos... We would be just fine if we used 939 + DDR up to now. DDR2 offered more bandwidth and more latencies as well. And we all know the performance impact that big latenvies have on K8 chips. Maybe DDR2 could help if K8 could overclock using big FSB's and multiplayers. (like C2D up to 500FSB)

DDR2 uses less power... LOL. how much less? 1 W ? Scientia when u want to backup AMD's choices , u really are imaginative...

Red said...

1.5% growth for what?
Revenue is down compared to last year's.
Margins are down.
Income is down.

Having more capacity while being the smaller company doesn't help when your products are worse. If you are talking about costs, I'd hate to know how their margins would've been without it, seeing as how margins tanked last quarter.

AM2 was necessary because the market was moving in that direction. The differences are too negligible to even tout though.

I think fuji was trying to say that Hector hopes K8L starts another 3 year AMD superiority cycle..

sharikouisallwaysright said...

AMD had to follow the prices that Intel sets becuase they had no better high-end product.
Intel could not drive AMD into the loosing zone so this quarter counts as a AMD-succes.
If AMD is still selling all and every CPU they will have stable income till their better nextgen CPU is on the narket and ASP goes up.
Then Intel with its pricewar has burned money for nothing and scared itself.
Time will show...

Wallachian said...

AMD has had a average quarter. This is probably a crap job done by their sales department. Digitimes reports their inventory went from zero to 1 month, now this is mainly a sales issue and not a product performance issue. I think their sales and supply chain could have handled the OEMs (big fat Dell) and channel in a better way. You can always sell lesser-than-great products if you have a clever sales team.

AMD would do well to acknowledge Intel as a serious threat than just trash-talking and being arrogant about it. Intel has acknowledged AMD as a serious threat, and didnt make much of a noise when Dell went to AMD, or when AMD went to ATI. I mean look at AMDs campaigns "Smarter Choice", "Better by Design", "Amount world spends in energy by NOT using AMD servers", they are not looking to take the fight to Intel, they are just simply trash-talking.

Kind of like how ATI started trash-talking Nvidia, when NV came out with their 7950GX2, ATI simply did not have a product to counter it, so they laid out 5-7 points on how NVs solution sucks.

Fujiyama said...

Scienta,
Compared to 2005 the rough estimated figure is 1.5% growth. I'm sure how this would be worse. What other information are you going by?

I would say that 1.5% increase when FAB36/Chartered runs 90nm/65nm is disappointing. Even larger X2 die doesn't explain that.
Regarding DDR2 - I agree partially that Opteron runs faster but an average consumer like me is locked with S939 single-core. Upgrades? No more.
K8L - I said that maybe Hector knows that next design is superior and outperforms Conroe or even 45nm derivatives.
I suspect 30%-70% performance increase per clock comparing to K8. Why? Answer is simple - it took 24 months of design, also HT3 is just...hmmm great.
Sorry for my english - but it is far not native.

Greg said...

Fuji, you could always go with a x2 3800+ and overclock to have a good and very cheap upgrade.

AM2 was not a necessity, if you don't think getting any major support from Dell or HP is important (cough* sarcasm). While AMD might have been able to negotiate a way for such retailers to use s939, they wouldn't get as high of an ASP from them as they're getting now, since it simplifies orders and saves the retailers boatloads of money when they have a simplified component lineup.

Wallachian, what are they supposed to do? Say, "oh ya, that's a really good solution, much better than our lack of one"? That wouldn't make any sense. It's a necessary step in order to stop consumer migration, as people will get confused over which facts actually matter.

AMD has always done a poor job focusing on sales, though Intel focuses to such an extent that it can become problematic for them.

Red, please pay attention to the whole, throwing money at your problems doesn't immediately fix them argument I've repeated multiple times. AMD is increasing capacity, because their cost analysts have judged that placing more money into K8L development will not speed its time to market enough to justify such expenditures, and that increased expenditures into 65nm development will not speed overall transistor performance increases or that such performance increases do not matter long term compared to capacity increase. And that's what this is all about for AMD, long term viability and survivability, which they are wisely considering more important that short term viability.

Again, AMD is making sacrifices now in terms of Margins, Income, and Revenue in order to have gains later. These people aren't stupid, and thinking they are shows that you are.

GDP, I'm really surprised your post hasn't been removed yet, but considering its arguments, I'd have to say that AM2 was obviously not aimed at the general consumer, and yielded most benefits in terms of servers and like I said above retailers. You're just being self centered, and not considering the bigger picture. Like I said, AM2 yields price cuts for the retailer, both because it simplifies memory selection for retailers, minimized power requirements for builders, which to retailers is huge, because they can, en mass, skimp much more on power supplies when you add a sff or ee amd cpu and ddr2 memory, and higher density memory into the picture (the other less mentioned benefit of ddr2). This lowers the price to the consumer (through the trickle down system) and allows greater competition to Intel through greater market penetration and thus consumer demand takes over. Though this is all very hard to see right now, in the long term, it will eventually be huge if AMD can keep it up. Even if they can't, Intel will have a lot of its Titanic appearance to retailers, which will still hurt a lot in the long run.

Red said...

Greg, wordy, much? I was responding to Scientia..
2006 is the year that Intel lost most of its FAB advantage when FAB 36 came online.
He says it as if having the fab boosted AMD and suffocated Intel. From their tanking margins, I'd say that it was a necessity, and rather than the "positive" earnings that they'll report, it is likely that they would've been in the red without the cost reduction. I didn't say anything about them throwing money to fix anything.

Again, AMD is making sacrifices now in terms of Margins, Income, and Revenue in order to have gains later. These people aren't stupid, and thinking they are shows that you are.
I'd like for you to elaborate on the gains to come. Your 2nd argument is pointless, not to mention insulting to stupid people everywhere.

Greg said...

I wasn't saying it wasn't a necessity red, it just sounded like you were saying making the fab 36 was stupid compared to their other options.

Having a 3rd fab, having integrated video cores, having a modular architecture, having a video company to use for their integrated video cores, having a workstation-gaming hybrid platform (which Intel's V8 has yet to accomplish)... I mean, obviously Intel is able to much more quickly play catch up because dumping money into R&D suddenly and in a way that to any other company would be reckless is able to actually yield results. However, AMD is now able to not have to worry about playing catch-up. Thus, the advantages they'll have in the future, that they would likely have lost if they did a massive R&D funding dump on a new architecture just so that they could have k8l out earlier.

Well, stupid people should be ashamed if they think they're smarter than an electrical engineer who runs one of the world's largest semi-conductor companies, because they aren't, and we don't get to know the same things they do until later, and then that's called hindsight, and again, you're stupid if you think that's as good as foresight (note, when I say you, it's a general you, and is not pointed at any specific contributor to this blog- just in case I offended anyone).

Also, I made a long post because there was a lot to address, and yes, I am very wordy, because I have a hard time knowing how to speak directly.

enumae said...

Hey Scientia, I do not mean to bring another blogs comments here, but in recent days the other blogger is very slow to post comments.

I am curious about your comment...

"Those numbers you gave weren't just cooked; they were carbonized. Please stop trying to dredge even deeper into the slime."

Please explain in minor detail how I have cooked the numbers.

All information was taken from their respective sites.

Thanks.

Red said...

Intellectual superiority, much? Your essentially saying that AMD deserves no criticism because they are "educated". The same thing can be said about other companies. Just wondering, what was the comment that led you to defend AMD on the basis that one can't criticize the executives?

Roborat, Ph. D. said...

Sharikouisalwaysdreaming Wrote:
Intel could not drive AMD into the loosing zone so this quarter counts as a AMD-succes.

It’s very interesting how your expectations for AMD are getting lower and lower. Forget the targeted 30% market share that AMD promised in when Fab36 came online. Forget the performance crown and the performance per watt crown. Forget about meeting schedules and commitments. Forget about beating Intel. Sure, as long as AMD’s turning a profit it is a success!
In business everything is relative and nothing is permanent. Watch the competition and watch the trend. Tell me again if you think we should congratulate AMD for this “successful” quarter.

Greg said...

I'm saying it's foolish for us to criticize those in places of power that we cannot control. The Presidents of our respective countries, obviously, are elected officials and thus we have every right (though still technically no intellectual high ground) to criticize them about it.

Ya, you can point out that, in hindsight, they had these options, and the one they took was the wrong one, but that does not mean they were stupid or that you are smart. And the people running these companies aren't people we get to choose for good reason, and as such, it's the companie's business to know when they're doing a good or bad job. We can say someone was better, and we're glad that the previous person was replaced, but saying that person was an idiot goes too far.

As for how this got started? I'm not exactly sure anymore, but I continued because you again questioned my statement.

Roborat, Ph. D. said...

Intel will report very strong earnings tomorrow.
Quick note:
Margins significantly up 52% while AMD is significantly down (below 50s).
Q/Q 10% revenue for Intel, AMD at 3%. (Q4 average is 6% from Q3)
2007 guidance for Intel to be increased by ~$1B.
On the 26th AMD will reduce revenue guidance by $1B! It’s very likely AMD will push out decision for NY fab by 2008/2009.
It just gets scarier for AMD in 2007. And we haven’t even factored in ATI’s loses.

Red said...

Why is it foolish to criticize when you have no power? That's like saying that the civil rights activists should have given up. Besides, investors are already "voting" by dumping the stock. Also, who criticized the executives? The closest thing I can think of is fuji saying that Hector hopes K8L repeats history. Seems like you're creating an argument to defend something that no one argued about.

Scientia from AMDZone said...

gdp77
U must be kidding me... AM2 offered nothing to the end user.

DDR2 uses less power... LOL. how much less? 1 W ? Scientia when u want to backup AMD's choices , u really are imaginative...


You are misunderstanding. There is not enough difference to make upgrading necessary. I'm saying that AMD had to change to DDR2 because it draws less power and is faster for future demands. If you already have a good 939 system there is no reason to upgrade.

Scientia from AMDZone said...

enumae

Q4 05 = 1.838 vs (Estimated) Q4 06 = 1.360

/-35% YOY/ -- /+3% PQ/


Your figure for year over year is completely wrong. AMD gained from 2005; it didn't lose. There was no 35% loss. Your 1.838 figure includes Flash memory but your 1.36 number does not include ATI. That is about as distorted as you can get.

enumae said...

Thank you for clarifying my mistake...

"That is about as distorted as you can get."

I take it you believe I have knowingly posted inaccurate numbers?

I have no intention to distort the facts, and I had not taken into account Spansion.

I will look up Spansion and adjust the numbers.

enumae said...

I found what looks like the numbers you are referring to, and as such I have revised the 3rd and 4th quarters.

Does this look more in line with what you thought would have been correct?

---------------------

Q3 05 = 1.010 vs Q3 06 = 1.327

/+24% YOY/ -- /+9% PQ/

Q4 05 = 1.350 vs (Estimated) Q4 06 = 1.360

/+1% YOY/ -- /+3% PQ/

----------------------

Thanks

Scientia from AMDZone said...

Red
He says it as if having the fab boosted AMD and suffocated Intel.


Actually, I wasn't. I'm not sure if you think that because you aren't familiar with history or because my English is a bit too subtle. It wouldn't suffocate Intel but it is yet another in a long line of improvements that bring AMD steadily closer to Intel.

From their tanking margins,

Their margins aren't tanking so I guess I can ignore what you said after that.

Scientia from AMDZone said...

enumae
Does this look more in line with what you thought would have been correct?


Yes, roughly 1.5% based on the estimated numbers. Not stellar but then not a loss either and certainly not the huge loss that you initially showed. However, I suppose we could point out that even though Intel's numbers are worse it will still be reported as a big victory for Intel.

enumae said...

Scientia
"However, I suppose we could point out that even though Intel's numbers are worse it will still be reported as a big victory for Intel."

Investors will see it simply as AMD missed there projected revenue and Intel did not, assuming Intel doesn't, thus a victory for Intel.

Scientia from AMDZone said...

Greg
GDP, I'm really surprised your post hasn't been removed yet,


Red and gdp were both banned from AMDZone. Red and gdp both place great faith in the Toms Hardware Guide test results. This is one of the things that leads them to have a much worse assessment of AMD than I do. Red has some difficulty with the subtleties of english and hangs around ForumZ and XtremeTech where there are very strong anti-AMD biases. Gdp lately just seems to want to rail about what he sees as the lack of performance by AMD.

However, I am not so naiive as to see a company's fortunes as being based only on the performance of the top desktop processor. AMD has been slowly gaining on Intel and Intel has been unable to stop this. I'm familiar with AMD's history so often something seems much different to me than it does to someone who is making an assessment based on the last quarter. Differences of opinion are okay; we'll just try to keep things civil. I definitely do not want it to be like Sharikou's blog where half the posts are personal insults.

Scientia from AMDZone said...

enumae
I take it you believe I have knowingly posted inaccurate numbers?

I have no intention to distort the facts, and I had not taken into account Spansion.


No, I'm not saying you were distorting. Those erroneous numbers were posted in article after article all over the web. You may not have known about the error but the authors who included those numbers in their articles were either incompetent or were knowingly biased. It just seems strange to me that when I point out things like this I'm labeled as an AMD fan. Doesn't it seem curious though that the errors are never against Intel?

Scientia from AMDZone said...

roborat
It just gets scarier for AMD in 2007. And we haven’t even factored in ATI’s loses.


You have this backwards. There is nothing scary. There could be a fall off in Q4 but don't take that as the pattern for 2007.

enumae said...

Scientia
"...It just seems strange to me that when I point out things like this I'm labeled as an AMD fan."

The simple fact that you can acknowledge certain aspects of each companies advantages eliminates you as an AMD fan, my only problem was that it seemed you were implying that it was intentional on my part.

I would say your opinionated, but that is a good thing as it shows another light on the subject you write about.

"Doesn't it seem curious though that the errors are never against Intel?"

As far as that goes, I haven't followed the processor segment long enough to have an objective view in regards to errors against AMD or Intel, but I try and only look at facts.

Keep up the great post.

Greg said...

But AMD is the victim of years of insane market domination by Intel for no reason other than the fact that Intel was there first. Then there's what is being discussed in the court case, but that's only a plausibility.

Again red, AMD is sacrificing profits now for marketability and marketplace acceptance later, when it will actually matter, so thanks for ignoring my point.

Naiive is a word, especially if you consider the fact that you can't place the required character in for "ii".

Intel has lost its ridiculously massive marketplace dominance, the ability to kill pc companies by threatening to deny dynamic or bulk pricing, and the fact that they were the only company people in the mainstream recognized as a viable processor source.

No red, I'm not making up an argument, as I said in my previous statement (again, thanks for reading them) I'm not exactly sure why I felt it necessary to specifically bring executives into it. However, when you talk about the decisions a company makes in terms of R&D and development costs, you're talking about the choices the executives are making, and your rhetoric concerning those choices had a very hostile and critical tone (I'm sorry if that seems wordy, but there's no other way to put it).

Red, your point is highly offensive to oppressed peoples the world over, as civil rights activists fought for rights they already had by divine right, but that others were denying them the fair use of for political and ideological reasons.

My point was red, that executives jobs are to use people in both their ignorance and intelligence in order to get more money, and if they're doing a good job, people will buy their product, and if they're doing a bad job, people will not buy their product. Their stock has little to do with their actual strength as a company, and any intelligent company knows to ignore their stock prices, unless they happen to depend heavily on them for income.

Scientia from AMDZone said...

The word was naive.

Red
you can whine on and on about how AMD is faster at Sciencemark

with Lady Logic on my side

Please stop playing victim, your constant defense of AMD, etc. speaks for itself. Fellow AMD fan, Fomitchev, even recognizes it.


If that is your attitude then perhaps you should avoid commenting on my next article. Consider this a warning.

Red said...

"Mean" parts omitted. Jeez

Explain to me how Intel "lost" anything then. Bad choice of words on your part, not a misunderstanding on mine. Margins tanked 10% last quarter and still slipping(do I need to do the math for you?)

Since when have I had "great faith" in Tom's? Not anymore so than other sites, they all say the same thing, Intel is faster for the moment, and even until recently, AMD was for the most part uncompetitive price wise.

I'm not predicting AMD's woes in the immediate future based on the QX6700, I'm predicting it on the fact that AMD has cut prices 3x this quarter, they sold a ton for cheap to Dell, while missing the opportunity to sell them at better prices to the channel, Intel has better stuff, etc.

The word was naiive, I corrected you, now you know.

Scientia from AMDZone said...

Red

Intel lost revenue versus 2005. Lost is an appropriate word. Intel lost revenue versus 2005 while AMD gained. It isn't that difficult to understand.

As far as THG goes, I place no faith in their testing; you obviously do.

Your comment on my misspelling of "naive" is called being a spellcheck troll. Since spelling errors are common in posting most posters are aware that you don't mention spelling unless you can't tell what the word was supposed to be. Apparently you didn't know that.

Nads said...

Hi,

this is my first time here,

Nice analysis i never thougt any1 can talk about INTEL/AMD war without being bias.

Anyway heres my suggestion,
I really hate blogging site like this, it 'd be cool if its like FORUM/Discussion board.I even created one here for free

http://myfastforum.org/

plz take a look,its PHPBB script totally free hosting.

Hope u switch to Forum/D. Board!

Erlindo said...

After all, Scientia was right:

Intel profits falls 39 percent

Erlindo said...

Can this be a consequence of the above posted? ;)

Intel to sell off Israeli flash fab

Roborat, Ph. D. said...

"After all, Scientia was right:"
Intel's 39% profit fall was announce by intel 9 months ago when they adjusted their earnings forcast. I don't know why you're giving credit to anyone.

"Can this be a consequence of the above posted? ;)
Intel to sell off Israeli flash fab"

Yes, they need the money for the biggest fab they're building just close by.

David said...

It's quite amusing to see how "forecasts" of this sort pans out, especially in the computer industry.