Monday, August 07, 2006

AMD heading to 40% market share

I projected that AMD will exit 2006 with 40% of the x64 market share (run rate). Now it seems this target is very realistic. It's simple math, starting with 22%, to end with 32% average, assuming a linear acceleration, the end rate is 42%. AMD plans to ship 26 million desktop CPUs in 2H06. Lenovo and DELL will both expand their AMD64 offerings.

I found this interesting piece of old news.

This report from HKEPC is very very revealing. As you can see, as of Q2, only 10% of AMD's desktop CPUs are x2 dual core (1 million), about 50% are Semprons (4.95 million), there were only 10,000 FX CPUs, or 0.1% (I guess AMD sold five FX CPUs in the whole 3rd world). Now, you understand why AMD's ASP was less than $100. Moving to Q4, AMD will produce 6.58 million x2 CPUs, 2.810 million single core Athlon 64, and 4.985 million Semprons. Now, you can also understand why AMD will see revenue growth. With 600% increase in x2 volume, AMD may even see ASP increase.

Now, go back and read my analysis on AMD's chip output on July 19, 2006. I estimated that AMD's Q3 chip unit to be 118.5% of its Q1 level. Now, look at this table, which shows AMD's Q3 desktop CPU unit (11.815 million) exactly 118.5% of its 1Q06 level (9.965 million). This was an exact match. Did I have AMD inside information? No. It's just a matter of correctly using the available data and construct the right formulation of the problem to be solved.

90 Comments:

Anonymous Anonymous said...

Now it seems that your target is realistic...? As in... you thought it was less realistic before...?

But I agree. AMD's marketshare is steadily increasing, capacity is steadily ramping up, and CONroe with all its devious marketing schemes has made NO impact whatsoever on this.

12:20 PM, August 07, 2006  
Anonymous Anonymous said...

First, I am an AMD fan. That being said, the numbers I read said if AMD shipped 26 million chips like they want to, they would be up to ~26% market share, although here it states 32%. Either way, it isn't 40%, but a HUGE victory for AMD, and more proof that Intel, with all their lies, can't compete with AMD in a fair market. I guess there are only so many idiot fanboys to go around for Intel. Yes, Conroe is great, but Intel was 3 years late bringing it out, now AMD is getting rewarded for what they accomplished in those 3+ years while Intel had no clue how to be competitive.

12:24 PM, August 07, 2006  
Anonymous Anonymous said...

Dell has ZERO AMD offerings today.

So, yes, anything past ZERO would be some sort of "expansion".

As we are starting from zero, maybe you can claim a 10000000000000000000000000000000000% increase.

Lenovo is rapidly disintegrating into a cheap pile of Chinese trash. The Thinkpad line is being gutted with cheap components and bad designs. Lenovo is too cheap to make a curved mouse button, for instance. If you are that cheap, then it spells only doom for anything called "quality" or "design". Bye bye Thinkpad.

Maybe Lenovo will flood the market with a bunch of cheap machines with yesteryear's processors that it bought at AMD's fire sale.

While is this technically "market share", it is ephemeral. Once Lenovo is done dumping all the trash chips, then the real "market share" figures will stabilize.

If AMD had a good chip, I'd be all for AMD ramping up market share. But all AMD does today is dump crappy chips on the market. AMD has not done any real chip design work in 5 years. AMD's original Opteron team is long gone and the only "innovation" AMD has to offer is vaporware specs for platforms and processors that are years in the future.

AMD is a sad company that is just a chip dumper. They are just as evil as Intel. AMD and their f-bois cannot even say "Intel did a good job with the Conroe". It is this lack of character that defines AMD.

12:32 PM, August 07, 2006  
Anonymous Anonymous said...

I've been trying to figure out the cpu volumes. AMD did 45 Million in 2005 and they planned to do 65 Million in 2006. It seems like current rate would only hit about 50 Million.

Considering that AMD was only at 22% in Q2 06 and they project 30% by end of 2008 how do you arrive at 40% by end of 2006? Do you have any hard numbers for cpu production by Intel, AMD, and VIA?

Scientia from AMDZone

12:37 PM, August 07, 2006  
Anonymous Anonymous said...

Conroe chips will not be available in any quantity until mid-August (or later)

Unless the manufacturers are cranking out Conroe chipped pc's, it seems like intc may miss the school market big-time.
As for me, at least 3 of the 26,000,000 AMD chips will be in my computers.

After all the hype, did intc have any inventory of their new chips, or was it all marketing ass gas.

12:54 PM, August 07, 2006  
Blogger Sharikou, Ph. D. said...

As in... you thought it was less realistic before...?

I should have said this target is being realized.

12:55 PM, August 07, 2006  
Blogger Sharikou, Ph. D. said...

how do you arrive at 40% by end of 2006

I was talking about the run rate.

12:57 PM, August 07, 2006  
Anonymous Anonymous said...

Sharikou, I'm interested to know what you think about this recent INQ comment:

http://www.theinquirer.net/default.aspx?article=33524

My opinion is that it's a load of baseless, speculative BS.

1:02 PM, August 07, 2006  
Blogger Sharikou, Ph. D. said...

Sharikou, I'm interested to know what you think about this recent INQ comment:

http://www.theinquirer.net/default.aspx?article=33524


I already analysed the AMD+ATI merger before and after the announcement. The merger will accelerate Intel's collapse.

1:07 PM, August 07, 2006  
Anonymous Anonymous said...

I really don't know how you count these numbers? To gain 18% of the market (AMD has 22%) something should happend to DELL first. They should stop production for example.
After 3 years performance lead AMD is able to get 22%. Now you are saying that 18% will come so easy in a half a year???

1:17 PM, August 07, 2006  
Anonymous Anonymous said...

I really don't know how you count these numbers? To gain 18% of the market (AMD has 22%) something should happend to DELL first. They should stop production for example.
After 3 years performance lead AMD is able to get 22%. Now you are saying that 18% will come so easy in a half a year???

1:22 PM, August 07, 2006  
Anonymous Anonymous said...

read his blog.........run rate

1:33 PM, August 07, 2006  
Anonymous Anonymous said...

"AMD's original Opteron team is long gone"

Somebody please confirm this? Where is the original Opteron team now?

1:33 PM, August 07, 2006  
Anonymous Anonymous said...

Hello... We are over at Intel...

j/k guys...lol

1:48 PM, August 07, 2006  
Blogger netrama said...

I found this interesting piece of old news

This is another reminder to the fact the Intel has successfully killed competitors in the past with their lies, boasting and fraudulent tactics.
They still keep trying this over and over again... (Why.. it is still the same pigeon heads running the show over there at Intel..) .. But they should realize that times have changed!! And these tricks work no more. You can only fool people a limited number of times.

2:14 PM, August 07, 2006  
Anonymous Anonymous said...

30% is more realistic...

2:56 PM, August 07, 2006  
Anonymous Anonymous said...

"AMD's original Opteron team is long gone"

this statement is extremely interesting. opteron is no more than server version of K8 core. its not different core or stuff like that. K8 core is in its own turn no more than tweaked K7 design. and there were no K7 based opterons.

so, either "original opteron team" is team of ppl who converted K8 into opty (thats real amount of work I would say), or if this ppl designed K7, they are strictly speaking not opty ppl.

wo this long gone ppl amd released next core revision with DDR2 IMC.

this statement is repeated constantly in comments of this blog. ppl are coming in and coming out. I can remember some brilliant ppl hired for amd. I dont remember shock sign offs (but some ppl did left company, just because).

so any comments on what do u mean?
some bright names and dates?

3:15 PM, August 07, 2006  
Anonymous Anonymous said...

Careful, careful ... Sharikou is saying AMD will have 40% of x64 server market share not 40% of the PC processor market. I don't think this will happen though. Intel has products that now compete pretty well in this space and will not let that happen. Intel can control it's prices better than AMD can and will not let its share continue to free-fall. AMD may exit with 30% share but not 40%. No way, no how. As Intel ramps the Xeon 5100 line and drops prices on older Xeon, this will take market share back or at least stem the tide... So no, Sharikou your predictions (as well as your other one that has Intel being bankrupt in 5 quarters) will be FALSE. I'll wait and see. Will you have the balls to admit you are wrong when you are proven wrong??? I don't have much hope though since your integrity is pretty much worthless at this point. You truly are the Shari-FRAUD!

3:23 PM, August 07, 2006  
Anonymous Anonymous said...

By the way check out the current poll going on at GamePC... see sidebar on the left

Summary, more people want Core 2 Duo than anything AMD has to offer by nearly a 3 to 1 margin. I would have to add my result to that as well since I don't want any Athlon crap polluting my PC. Cool and Quiet? Then you want Core 2 Duo!

3:27 PM, August 07, 2006  
Anonymous Anonymous said...

What is this conclusion based on?

Any articles / research you can point to?

Oh, I forgot, your analysis is so brilliant mere mortals couldn't understand it.

Just like when the "advisors" were reviewing your "Ph.D." thesis they found it too advanced for their understanding....

Thanks for the laughs today.

3:45 PM, August 07, 2006  
Anonymous Anonymous said...

Great Core 2 Duo review here:
Starting Off Strong : Intel’s Core 2 Duo E6400 and E6300

I just hope that Sharifraud has the integrity to post this since he likes to censor some things that make his arguments look bad. Keep in mind Gamepc.com sells both AMD and Intel products and probably don't have a real bias one way or the other (unlike a certain Blog I know of). The comparisons were with the smaller 2MB Allendale Core 2 Duo chips so these aren't even the Core 2 Duo heavy hitters. A quote from their conclusions:

Intel is certainly off on the right foot, as one can clearly see by reading the last few pages of benchmarks. The Core Duo E6400 processor is priced slightly less than the Athlon64 X2 4600+, but bests the 4600+ in nearly every benchmark. There are a few edge case scenarios where AMD’s architecture shines through, but those scenarios are becoming fewer and fewer as of late. AMD’s big selling point used to be exceptional gaming performance, whereas that is now entirely bested by the Core 2 Duo. At stock speeds, the $300 Core 2 Duo E6400 easily bests comparable AMD products in gaming scenarios, and is competing amazingly close to levels of the Athlon64 FX-62, AMD’s $1,000 flagship gaming processor. With AMD’s flagship FX-62 seemingly overpriced and their high-end Athlon64 X2 5000+ missing in action, Intel has a chance to take a significant chunk of the enthusiast market away from AMD.

Wow! I can't wait to get one of those. Cool and Quiet, low price and Intel can still show a profit on them because of superior fabrication process. AMD may have to sell at a loss just to stay in the game for the next few quarters. I predict AMD losses starting in Q1'07 and continuing for the rest of the year. ATI won't help that!

3:52 PM, August 07, 2006  
Blogger "Mad Mod" Mike said...

Welcome the AMD Radeon (LLink).

4:22 PM, August 07, 2006  
Anonymous Anonymous said...

Graham - I have a hard time taking someones recommendations when they suggest to buy a nearly impossible item to get instead of an item you can easily get anywhere.

The benchmarks are there - we get that, the chip just isn't yet.

5:12 PM, August 07, 2006  
Anonymous Anonymous said...

The bickering between AMD fans and Intel fans is humerous, but not particularly enlightening.

I see that some think that AMD is now crap and Intel kicks butt, while others are in denial about Intel's performance advantage.

The newly refreshed Katmai architecture (known as Core) is efficient and will have an impact. The long term question is what does AMD release next?

It took Intel ~2 years to respond. How long will it take AMD? How much change will they make? How much performance and feature upside will the new design provide?

Time will tell, but one thing is certain. Given that Intel has blown its wad for the next couple years with "Core", if AMD responds soon enough with enough feature and performance upside, they will have Intel's Otellinis in a vice.

This all derives down to what the real difference is between the companies, both of which employ some great engineering talent.

The real difference between Intel and AMD comes down to a simple factor: Leadership

Intel just doesn't have any...unless you count bad leadership as leadership. All you have to do is read some of the executive level interviews and look at the product decisions to see that.

Sad part is that Intel's employees will have to suffer massively until the investment community forces the Intel board to take major action to fix the problems...question is: will that happen before its too late? After all, it takes years to design, build, validate and launch a new design.

-Jack

6:09 PM, August 07, 2006  
Anonymous Anonymous said...

"What is this conclusion based on?”

AMD has been gaining mark share (obviously) Dell wants in big time. HP, Sun, IBM and countless other partners want more. Without even quoting numbers what would you say the trend is?

Who is laying off 10000 employees due to lost market share?

When Vista is released and Intel is struggling with 64 bit, who do you think will gain market share?

Class in logical conclusion 101, please get your head out of you’re a$%# and come up to speed.

7:15 PM, August 07, 2006  
Anonymous Anonymous said...

This blog is getting boring. All I ever read are:
1. Dell laptop burned, battery's fault but Intel to blame.
2. AMD market share to rise, Intel to go BK.
3. AMD server processors are so fast.
4. Conroe's nothing but paper launch.
5. 4x4, AM2, everything that AMD has will the standard of the future.

At first I thought that this blog is great but everyday is a rehash of the 5 items mentioned above.

I thought this was an intelligent discussion but its not. Every answer to a rebutal is "frag this, frag that..."

It would be nice if there are links to other sites that would support the claim.

Thanks to everyone, it was "hu-hum" interesting.

7:20 PM, August 07, 2006  
Blogger Michael said...

I can easily see AMD have 40% market share soon. I've perdicted this also. I can't wait until it happens though it would make me feel like I have helped them some how because I buy all AMD.
CyberSurge

8:18 PM, August 07, 2006  
Anonymous Anonymous said...

"AMD and their f-bois cannot even say "Intel did a good job with the Conroe". It is this lack of character that defines AMD."

Good job? Success? Those are all relative to whom and how it is analyzed.

Now here is my perspective of your so called success/good job!

YOUR Intel, the company you brag about that is so profitable. The company you claim has capacity, manufacturing and technological superiority. The company that has $10 billion in cash, higher margins, the best designers, etc...

Came up with a product that is ONLY 10% faster, slighly better on power consumption, has enourmous caches and is manufactured using superior technology...

Compared to a product (Athlon/Opteron) that is 3 years old, is manufactured on a process 6 months old (90nm), and has a fraction of the cache.

I don't see HOW in the world you can say that is amazing?

But I'll say this much: Intel surprised me at how fast their response was... that's it!

8:34 PM, August 07, 2006  
Anonymous Anonymous said...

"AMD has gone through stages of progression, where each new generation of architecture has been built off of the old. You can see vast similarities in the transitions. Intel did a complete overhaul and took a completely new direction in their methodology."

That's exactly why I predict a major Intel recall will eventually happen before they move to their new architecture. BILLIONS of $$$$$$!!

Intel has a monopolist culture embeded right into its roots. This company knows nothing about great designs, pressure, deadlines... they fucked up somewhere along the lines on this Core architecture. Now its just a matter of time before it's revealed.

8:51 PM, August 07, 2006  
Anonymous Anonymous said...

"Jim Cramer is a fool. If he did his research, he would know that Intel [...]"

Jim Cramer is brilliant! Normally this guy will always say:

"Stick with the leaders", "Stick with the one that owns the market share", etc.

Just the fact that he did his research, crunched the numbers and he is telling everyone to sell Intel and buy AMD says ALOT!!!

You've got half the world invested in Intel (most don't even know it), so just give it time... Intel's shares will be high teens by the end of the year!

Oh, and havn't you ever heard of 'Dollar Cost Averaging'? If you're sure about a stock in the long run, isn't it better to keep buying at a discount? A 30% discount?

8:56 PM, August 07, 2006  
Anonymous Anonymous said...

opteron is no more than server version of K8 core. its not different core or stuff like that. K8 core is in its own turn no more than tweaked K7 design. and there were no K7 based opterons.

The K8 core is the name of the core for Athlon 64, Sempron, Turion and Opteron. The first K8 available for sell was the Opteron. The k8 arch. is server-focused, with an IMC, giving the possibility of one memory bank per processor, and Hyper Transport, offering great scalability in multi-processor computation. Athlon and Turion are respectively the desktop and notebook implementations of the Opteron .
Regarding the transition from K7 to K8, it is true that it was an evolutionary not revolutionary transition, mainly because the K7 designers had done a great job. The execution units were left more less the same, but with major tweaks in the parts that come before the execution: memory subsystem (IMC, HTT, and larger caches with lower latency and bandwith) and the fetch and decode stages of the microproccessor were greatly improved and 2 pipelines were added in those stages, precisely.
The transition from k7 to k8 is similar with that to PIII to PM.
The Server class microprocessor within the K7 arch is the Athlon MP, basically an Athlon with a dual bus, a bus pretty similar of the actual Xeons, as sharikou has correctly stated.

9:13 PM, August 07, 2006  
Anonymous Anonymous said...

For all your questions there is one simple answer. AMD will respond with a new architecture where K8 will be obsolete. I mean - AMD will start loosing marketshare and profits.
And it is not going to happen till second half of 07. Why?
Because 65nm stupid! Lower cost mean the same margins for same processors with higher clock and even lower prices. Dual core business is not going to change that.

This is business, not fanboism...

10:38 PM, August 07, 2006  
Anonymous Anonymous said...

"At stock speeds, the $300 Core 2 Duo E6400 easily bests comparable AMD products in gaming scenarios, and is competing amazingly close to levels of the Athlon64 FX-62, AMD’s $1,000 flagship gaming processor."

Given that GamePC purposely put out misleading pricing on FX-62, I really don't give a damn to its bad CPU review. I'm only interested how much marketing money it got from Intel...

11:25 PM, August 07, 2006  
Blogger Scientia from AMDZone said...

What do you mean by "run rate"? Are you talking about chip volume? In other words, 40% of by count of all x86 processors sold?

The digitimes article mentioned at the top is in line with what I have seen. It says 46 Million for desktop and 12 Million for notebooks. This would be 58 Million total. With another 7 Million for servers this would reach the 65 Million that AMD estimated at the beginnning of the year.

AMD was only at 22% by volume in the second quarter so I don't see how this could go up to 40% by years end. However, if you are not talking about chip volume I'm not sure what other measure there is.

You probably realize now that your earlier estimates of 50% of the desktop market, 25% of the notebook market, and 45% of the server market are rather optimistic. I can't criticize your estimates too much because they were BC (Before Conroe) and what Intel had available in January of this year wasn't really competitive.

You probably also realize that your earlier estimate of 130 Million chips by AMD is also wrong by a factor of 2. Again, this can be very difficult to estimate without specific information on yields, future chip sizes, and ramping rates. My best current estimates are that FAB 36 will provide 1/3rd of FAB 30's production during 2006. For most people the ramping of FAB 36 is counter-intuitive because 3/4 of the chip production show up in the second half of the year. This should create a small flood of 65nm at years end.

11:55 PM, August 07, 2006  
Anonymous Anonymous said...

Sorry but the POWER6 leaves no room for Intel. It will only be IBM and AMD. Woodmistake can't even come close to POWER6 or K8L. Intel will last 6 more months at best. So realisticly benches don't matter in the real world. Intel can't even do 4 sockets like AMD can. Intel has a lot of problems with 2.

Opterons are the best. Why are so many companies coming out with the AMD solution over the Woodmistake problem? Google, HP, DELL, SUN, IBM. I don't see intel in their future. DELL will be all AMD soon after they get rid of all the intelcrap they where given. Google is the largest search engine in the world. All I see is 100,000 servers with AMD in them. Intel... whats that anymore? Not what they use to be thats for sure.

AMD hust took have their chipsets too. No more good chipsets for Intel.

12:27 AM, August 08, 2006  
Anonymous Anonymous said...

I think anybody should stop quoting GAMEPC.COM.
Because in their benchies they always use slow 5-5-5-15-2T memory, which seriously degrades Athlon performance.

3:10 AM, August 08, 2006  
Anonymous Anonymous said...

I have to say 30% is a much realistic target. 40-45% (maybe) in H1/2007. It is very sad, but as a AMD fan, I have to say that Conroe is eating much more ground that I have expected. I saw 3 or 4 polls where more than 50% of users prefer Conroe. AMD must do something to force Intel to slow down, if not, we will see a messy situation and all AMD effort to gain more market share will be in vain... I'm sorry but thats the aweful truth. Cheers AMD bros ;)

3:21 AM, August 08, 2006  
Anonymous Anonymous said...

i was wandering...how possible is integrating DX10 capability into A64s? no high end stuff, just the minimum to run vista smoothly. could it be part of the integrated northbridge on new turions?

and btw...why noone points out, conroe is 2x the size of A64 X2s. i mean...WTF!? what a crappy architecture it has to be, 2x the size and just +20-30% performance (and i am very generous with that :) and close to zero scalability it seems

...if half the rumours about k8l are true...and torrenza kicks in... hope intel has some nice cpu for thin clients, for all the amd servers :)

4:55 AM, August 08, 2006  
Blogger Sharikou, Ph. D. said...

What do you mean by "run rate"?

I was talking about the trailing unit share.

8:09 AM, August 08, 2006  
Anonymous Anonymous said...

Looks like AMD is keeping the financial pressure on INTC.

http://www.theinquirer.net/default.aspx?article=33536
If true, this would bolster AMD's court case against INTC.

8:20 AM, August 08, 2006  
Anonymous Anonymous said...

Some anonymous coward wrote:

Graham - I have a hard time taking someones recommendations when they suggest to buy a nearly impossible item to get instead of an item you can easily get anywhere.

The benchmarks are there - we get that, the chip just isn't yet.


You can get the E6300 and E6400 at this little distributor (I've never heard of them before :-) ) called NewEgg.com. In stock today. Another small company that sells printers and PDAs but has a side job in PCs is selling Core 2 Duo systems as well. Go over to Dell.com before too many people hear about them. Shhhh... Keep it quiet so they sell out.

Man, the AMD fanboys are really trying hard to deny the existence of Core 2 Duo. Keep sticking your heads in the sand and play on your old slow Athlons. First excuse was that the benchmarks were made up. Nope, they were real. Now AMD fannies (pun intended) are saying that you can't buy Core 2 Duos. No. I just busted that myth. What next? I can hardly wait for the next good excuse. Maybe the dog ate AMDs homework next time?

8:59 AM, August 08, 2006  
Anonymous Anonymous said...

AMD must do something to force Intel to slow down, if not, we will see a messy situation and all AMD effort to gain more market share will be in vain... I'm sorry but thats the aweful truth.

The interesting thing about AMD expanding capacity is that even if they don't sell every chip they make, they will still take market share from Intel. If users start to accept core 2 duo big time, and shying away from K8, the capacity will be there when K8L is released and AMD will be able to sell alot of them if it turns out to be as good as it looks.

9:08 AM, August 08, 2006  
Anonymous Anonymous said...

Oh, and havn't you ever heard of 'Dollar Cost Averaging'? If you're sure about a stock in the long run, isn't it better to keep buying at a discount? A 30% discount?

You show no intelligence Jeach. Although I don't expect much from someone calling themselves Jeach. Cramer in May was saying that AMD was a great buy... when their price was around $30/share. He had no idea there would be a huge fall so that wasn't factored into his advice. He also didn't do the research and recommend waiting out the price war that EVERYBODY knew was coming which really makes him look dumb. Forget $ cost averaging... that's for retirement accounts. If I had taken Cramer's advice (which I wouldn't since he is a stark, raving lunatic) and plunked down $60 Grand on AMD when he was pumping their stock, I would have around 33% less money today and would have to wait God knows how long just to get back even again. I pity the fools who DID take his advice. At least with Intel stock you get paid a dividend (albeit small) to wait for the stock to crawl its way back to positive. No dividend for AMD but that's what happens when you invest in a company with no profits.

9:13 AM, August 08, 2006  
Anonymous Anonymous said...

With 600% increase in x2 volume, AMD may even see ASP increase.

Maybe. Bad timing though... the drastic price reductions that AMD had to take just to keep market share will take its toll. Also there is nothing saying that they will actually SELL evertying they produce.

10:05 AM, August 08, 2006  
Anonymous Anonymous said...

"If users start to accept core 2 duo big time, and shying away from K8, the capacity will be there when K8L is released and AMD will be able to sell alot of them if it turns out to be as good as it looks."

Correct! That's the beauty when you release a product only at mature yield. It seems to me that AMD has calculated the Core 2 Duo effects well and planned ahead.

12:07 PM, August 08, 2006  
Anonymous Anonymous said...

Intel's Worst Nightmare:

"http://money.cnn.com/magazines/fortune/fortune_archive/2006/08/21/8383598/index.htm?source=yahoo_quote"

And if the lawsuit proves Intel abused its monopoly practices a la Rockefeller, expect that market share to boom even more...

2:52 PM, August 08, 2006  
Anonymous Anonymous said...

YES, AMD IS MAKING TONS OF CPU, BUT WHERE CAN I BUY THE AMD ATHLON X2 3800 EE SFF (35W)? IT IS AS GOOD AS VAPORWARE SINCE IT CANNOT BE FOUND IN US.

HINGSUN

4:32 PM, August 08, 2006  
Anonymous Anonymous said...

"But I agree. AMD's marketshare is steadily increasing, capacity is steadily ramping up, and CONroe with all its devious marketing schemes has made NO impact whatsoever on this."

Conroe just started shipping.. give it time before you reach a conclusion..

5:22 PM, August 08, 2006  
Anonymous Anonymous said...

"Conroe chips will not be available in any quantity until mid-August (or later)"

Actually, I think the back to school season is aimed at reducing the glut at Intel rather than selling Conroes!

5:24 PM, August 08, 2006  
Anonymous Anonymous said...

"Graham - I have a hard time taking someones recommendations when they suggest to buy a nearly impossible item to get instead of an item you can easily get anywhere.

The benchmarks are there - we get that, the chip just isn't yet. "

You are right.. It seems you, along with others, missed the announcement of the release of CONROE.. Intel declared that people will start seeing it after August 7th.

5:33 PM, August 08, 2006  
Anonymous Anonymous said...

You are right.. It seems you, along with others, missed the announcement of the release of CONROE.. Intel declared that people will start seeing it after August 7th.

Considering that comment was made on August 7th, where no conroes were to be found, it was still valid. Now that its August 8th, and the low end parts are out and about at inflated prices, it changes the AMD-Intel ballgame a little, we'll see how it pans out for both sides. But yes, you can buy an E6300 or E6400 now, yay!

Newegg - 229 (6300)
TigerDirect - 230 (6300), 270 (6400)
ZipZoomFly - 219 (6300)

The street prices are not as competitive as AMD's X2 line (depending on which benchmarks you look at and what motherboard you plan on buying), but at least it exists! Hopefully in a couple more weeks/days the prices will hover down closer to their distributor price.

Its a good time to start thinking about buying a PC. I'm waiting until Crysis comes out, so by then this whole shenanigan will be mature anyways.

5:55 PM, August 08, 2006  
Blogger Sharikou, Ph. D. said...

Newegg - 229 (6300)
TigerDirect - 230 (6300), 270 (6400)
ZipZoomFly - 219 (6300)


The fact Intel can only release Conroe at low speed and 2MB cache indicate that it has a lot of problems with yield and thermals.

6:34 PM, August 08, 2006  
Anonymous Anonymous said...

Sharikou, Ph. D said...

"The fact Intel can only release Conroe at low speed and 2MB cache indicate that it has a lot of problems with yield and thermals."

Did you consider that all the midrange chips have already been bought by the likes of Dell, HP, and Gateway.

Looking at your link for BLT ,it would seem that the 16th is gonna be the day.

If after the 16th, there is still a shortage, you may be on to something.

7:22 PM, August 08, 2006  
Blogger Sharikou, Ph. D. said...

Did you consider that all the midrange chips have already been bought by the likes of Dell, HP, and Gateway.

Out of the mouth of Paul O, we know Intel can only ship 1 million Conroes in 7 weeks. That volume is negligible. AMD sells 1 million CPUs per week.

7:38 PM, August 08, 2006  
Anonymous Anonymous said...

Look at the FX in the forecast.
Why 2x FX in Q4 than in Q1 or Q2?
Maybe it's for the 4x4 platform...

7:45 PM, August 08, 2006  
Anonymous Anonymous said...

Conroe Not Available:

Dell XPS 700: Oct. 18 http://configure.us.dell.com/dellstore/config.aspx?c=us&cs=19&l=en& oc=DXPS700B2&s=dhs

Alienware A51 ALX: Sep. 12
http://www.alienware.com/configurator_pages/Area-51_ALX_r5.aspx?syscode=PC-AREA 51-ALX-R5&subcode=SKU-DEFAULT

Aria.com 6300: not available
http://www.aria.co.uk/ProductsList.asp?Category=9&SubCat=CPU-AA7

ZipZoom 6400: no stock
http://www.zipzoomfly.com/jsp/ProductDetail.jsp?ProductCode=80859

NewEgg 6600: No stock
http://www.newegg.com/Product/Product.asp?Item=N82E16819115003

NewEgg 6700: No stock
http://www.newegg.com/Product/Product.asp?Item=N82E16819115002

ZipZoom 6800: No stock
http://www.zipzoomfly.com/jsp/ProductDetail.jsp?ProductCode=80862

Evesham 6300: Mid-October
http://www.evesham.com/products/info.asp?e=8D34ACED-8A01-44E4-80FC-D88021C293A5

Velocity DCX Raptor: Aug 29
http://www.velocitymicro.com/wizard.php?iid=18

How can Intel hope to take share away from AMD like this?

7:46 PM, August 08, 2006  
Anonymous Anonymous said...

The fact Intel can only release Conroe at low speed and 2MB cache indicate that it has a lot of problems with yield and thermals.

OR, Intel will wisely offer its lowest end priced Conroe first, and gradually release more powerful processors as time goes on, which only makes sense.

Doesn't make sense for any company to release high cache high Ghz cpu's and then release a smaller cached slower frequency CPU, does it.

You will take any little shred of a notion and make it into an Intel failure, won't you. That's not a question. That's a fact.

7:53 PM, August 08, 2006  
Anonymous Anonymous said...

They are totally available. Bzzzt. Try again.

PLZ SHOW ME THE WEBSITE WHERE THEY ACTUALLY SHOW THE PRODUCT TO BE IN STOCK, I WANT TO BUY IT NOW YET PAY LESS THAN 220.

HINGSUN

9:15 PM, August 08, 2006  
Anonymous Anonymous said...

NCIX IS SELLING THE ADO 65W VERSION, NOT THE ADD 35W VERSION. ENTROPHOS, PLEASE SHOW ME WHERE I CAN GET ME THE ADD VERSION, THX.

HINGSUN

9:17 PM, August 08, 2006  
Anonymous Anonymous said...

Doesn't make sense for any company to release high cache high Ghz cpu's and then release a smaller cached slower frequency CPU, does it.

Under ordinary circumstances, yes - you'd want to release low end chips first, and then transition up to the high end chips when you know you are able to get decent enough yields on them so that you don't lose money.

In this case, Intel has stockpiled inventory so high, that I think it would be a smarter move to simply have not released the low end chips and try to get the high end ones out the door first. The low end E6300 and E6400 are in direct competition with Intel's own Pentium D line.

Sales of the low end C2D chips take directly away from sales of pentium D, and since they aren't in very high volume, they don't really take away from sales of Athlon X2 chips. By the time they do ship in that kind of volume, yields will have matured and they can produce whatever chips the market demands.

Intel doesn't want to lower sales of pentium D - it needs to sell as many of those chips as it can.

Thus, it is in Intels own best interest to be producing high end chips, not low end chips. The fact that they aren't (or we have incomplete knowledge) likely means that they can't get high enough yields.

The other implication is that since the high end chips are scarce (based on our limited knowledge), all or most of intel fabs are currently producing chips that are either getting thrown on the inventory heap (ie P-D) or getting sold instead of the chips in inventory (C2D low end). If true - that is not a good situation to be in for the third quarter with AMD ramping production as well. Either mass production lines will go offline (which will help them in the long run by getting converted to C2D faster) or alot of chips are going to the dumpster in 6-12 months.

9:37 PM, August 08, 2006  
Anonymous Anonymous said...

Intel has brought exitemnet to the marketplace with CONROE. Its a bad idea to ship Conroe in volume when you know that netburst needs to be cleared out. So, Intel is right in doing a slow ramp. Q3/Q4 results will vindicate Intels strategy.
Can AMD bring the same level of exitement this year? I dont think so...

11:14 PM, August 08, 2006  
Anonymous Anonymous said...

From the article posted by Sharikou:

"AMD is not nearly as strong in the worldwide mobile PC market, rising from an 8.9 percent market share in the fourth quarter of 2004 to 12.2 percent in the fourth quarter of 2005, according to AMD."

I found this entertaining, as pretty much anyone with any insight into the semiconductor industry knows that Intel all but owns the mobile market. It's always enjoyable to watch Sharikou to mislead his readers by showing and manipulating data that makes AMD look good. Like giving information about AMD having a ~45% share of the retail mobile market, but then forgetting to mention how small a fraction retail sales account for in the mobile market. Sigh.

5:48 AM, August 09, 2006  
Anonymous Anonymous said...

"NCIX IS SELLING THE ADO 65W VERSION, NOT THE ADD 35W VERSION. ENTROPHOS, PLEASE SHOW ME WHERE I CAN GET ME THE ADD VERSION, THX.

HINGSUN"


Does anyone know what the release date on the EE models is/was? I'm just curios.

6:28 AM, August 09, 2006  
Anonymous Anonymous said...

A friendly caution goes out to my AMD friends who are making 'Conroe success/failure' judgments so early after its release... give it time!

But I would like to point out this very interesting fact though. One which I strongly believe reflects the average Conroe consumer (I hope).

My friend wanted to buy a new PC for a little while now. Being an Intel fan, the day the Conroe was released he went to his local computer store and placed an order.

He was told that he'll have to wait 1 week (2 weeks tops) for his new Conroe based PC.

The other day when I saw him (after waiting his first week), I told him that he probably would have to wait another few weeks before he got his computer.

Without a single hesitation he replied:

"If next week they don't have my PC I will be canceling my order and buying an X2 system".

That surprised the hell out of me!

My point is let time fly and let the dust settle before making success/failure projections.

9:40 AM, August 09, 2006  
Anonymous Anonymous said...

Jeach! said...
A friendly caution goes out to my AMD friends who are making 'Conroe success/failure' judgments so early after its release... give it time!

A while back, Intc poo-pooed their horrendous inventory of 'worthless' chips, saying $800 million of it was C2D.
I think we can now safely say this was just more B.S.
As Core 2 Nullo seems in very short supply, with street chip prices considerably over cost, I expect some OEM's to dump C2D on the street, leaving customers like your friend 'Hard Orders' to wait until street price drops. 'Market Ethics'
Advise your friend to cancel NOW, wait until this shootout settles down, or get an AM2 MB and a (relatively) inexpensive AM2 X2 4XXX processor and 2GB of ddr2 800 ram and help him build a new machine he can enjoy NOW.

12:22 PM, August 09, 2006  
Anonymous Anonymous said...

"Advise your friend to cancel NOW, wait until this shootout settles down, or get an AM2 MB and a (relatively) inexpensive AM2 X2 4XXX processor"

Plus a good AM2 SLI MB has the ability to upgrade to true quad core early next year, lik the socket 939 2 years ago can easily upgrade to dual-core today. Such value & performance extensivility is simply unmatched.

Of course, if you desire the best desktop performance right now, then you should wait/go for Core 2 Duo today. If you also want good upgradability, then you should choose wisely a motherboard certified for Kentsfield as well. It's a lot more efforts, and probably some more money, but I guess that's what defines an 'enthusiast'. ;-)

5:10 PM, August 09, 2006  
Anonymous Anonymous said...

Any insight on lower power X2's? I'd love a 35W version for my next build.

5:14 PM, August 09, 2006  
Anonymous Anonymous said...

"Does anyone know what the release date on the EE models is/was? I'm just curios."

It's June 1st.

Initially you can place orders for it on pricegrabber, but now you can't (probably due to low demand and availability).

It seems fair at this point to call it vaporware - at least to the end users. I think it's much more profitable for AMD to sell low-power chips as Turion X2 or as low-power Opteron to SUN, for example.

5:22 PM, August 09, 2006  
Blogger Sharikou, Ph. D. said...

Of course, if you desire the best desktop performance right now, then you should wait/go for Core 2 Duo today.

No. In that case, you should wait for 4x4, or get a 4P 8 core AMD64 workstation.

5:54 PM, August 09, 2006  
Blogger Sharikou, Ph. D. said...

With AMD’s flagship FX-62 seemingly overpriced and their high-end Athlon64 X2 5000+ missing in action, Intel has a chance to take a significant chunk of the enthusiast market away from AMD.
The FX62 market is only 10,000 CPUs per quarter. With Conroe volume less than 5% of the market, AMD will simply sell all it can make by beating the 95% Netburst crap. That's why AMD will reach 40% of the market exiting 2006. In 1Q 2007, Conroe will finally reach 30% of the volume, but AMD will release K8L.

Intel BK in 5 to 7 quarters.

6:01 PM, August 09, 2006  
Anonymous Anonymous said...

Sales of the low end C2D chips take directly away from sales of pentium D,

You make a very good point.

However, the people that buy a C2D today will not be people that buy a Pentium D. However, if the prices diverge greatly on these two chips, only high end consumers will go after the C2D and others will buy up the Pentiums.

6:43 PM, August 09, 2006  
Anonymous Anonymous said...

"2.AMD is with ~6% more efficient, than Intel from x32 to x64 transition. The reason is that the Core2 is "native" x32. This is because the EM64T hasn't "Macro-Fusion"...
Some peoples said, that x64 isn't important. At this point of view I want to recommend these people to play with 10GHZ quad core i80386:)))"

6% more efficient, eh.....out of curiosity which chip ACTUALLY performed better on those 64bit benchmarks when comparing the 2 AGAINST EACH OTHER?

If I make a crappy 32bit chip and it gains 100% going to 64bit does that really matter if it still gets beat by a chip that "only" improves by 10%?

It would matter only if the 2 chips were performing at the same level on 32bits (which they are not), why can't people get that?

10:12 PM, August 09, 2006  
Anonymous Anonymous said...

"Correct! That's the beauty when you release a product only at mature yield."

Does anyone actually have any data on this? I keep hearing it but have yet to see any actual #'s backing this up.

Even AMD's own analyst foils show the yield rate plots ramping up as plotted AGAINST PRODUCTION VOLUME! Meaning when they start production on a given technology node, they are not at "mature yield" as everyone continues to state, incorrectly.

I'm not saying this pertains to specific products (i.e if you start a product 2 years into the 90nm ramp say) but K8L is projected to come out when AMD starts ramping 65nm so I don't think it is accurate to state the product will start at mature yield.

If they truly are at mature yields at the start of a production ramp - why are not all of those plots horizontal lines, instead of creeping up toward AMD's own "mature yield target line" over time/production volume?

10:23 PM, August 09, 2006  
Anonymous Anonymous said...

"The fact Intel can only release Conroe at low speed and 2MB cache indicate that it has a lot of problems with yield and thermals"

How so?

Perhaps this is where Intel sees greater sales volume. As evidenced by the 0.12% AMD FX volumes, while the top of the line products have the highest margins, the sweet spot for demand is generally not there.

10:26 PM, August 09, 2006  
Anonymous Anonymous said...

"Intel BK in 5 to 7 quarters."

4 to 6 quarters now - you made these prediction last quarter!

To put this in perspective - you are saying Intel will be bankrupt between July-Dec next year.

10:33 PM, August 09, 2006  
Anonymous Anonymous said...

Sharikou: "I estimated that AMD's Q3 chip unit to be 118.5% of its Q1 level. Now, look at this table, which shows AMD's Q3 desktop CPU unit (11.815 million) exactly 118.5% of its 1Q06 level (9.965 million). This was an exact match. Did I have AMD inside information? No. It's just a matter of correctly using the available data and construct the right formulation of the problem to be solved."

While your estimates are admirable - you conveniently ignored mobile and server space. You stated 118.5% factory / CHIP OUTPUT and compared it to DESKTOP CHIP projections. How does your analysis hold up against OVERALL Q3 AMD estimated chip output? To say "exact match" is not accurate unless mobile and server estimates are also up 118.5%, no?

Your analysis is in the right ballpark, but to call your analysis an exact match is both arrogant and factually incorrect.

11:17 PM, August 09, 2006  
Anonymous Anonymous said...

"Even AMD's own analyst foils show the yield rate plots ramping up as plotted AGAINST PRODUCTION VOLUME! Meaning when they start production on a given technology node, they are not at "mature yield" as everyone continues to state, incorrectly."

Right... I should've said "close to mature yield", not just "mature yield". ;-)

Either way, that's a good/reasonable thing for AMD to do, given they have only 2 production fabs.

9:17 AM, August 10, 2006  
Anonymous Anonymous said...

Edward: "Right... I should've said "close to mature yield", not just "mature yield". ;-)

Either way, that's a good/reasonable thing for AMD to do, given they have only 2 production fabs."

With no units on the graph (and I'm not sure if it was clear if it was log scale or linear scale), I'm not sure how you can state "close" either.

I do appreciate your ability to think through opposing points of view (unlike others on the forum).

5:22 PM, August 10, 2006  
Anonymous Anonymous said...

"With no units on the graph (and I'm not sure if it was clear if it was log scale or linear scale), I'm not sure how you can state "close" either."

Please... Who would use log scale on a yield percentage? Wouldn't that be negative all the way (like a dB graph)? You probably meant "square root" or "cubic root" which would make worse yield look better on the graph? But does that make sense?

10:02 PM, August 10, 2006  
Anonymous Anonymous said...

Sharikou,
Sell Intel Flash to Spansion?????
You must be out of your mind!!
Spanison is atleast 2 generations behind Intel AND has poor reliability.
Flash is about to take off....

10:48 PM, August 12, 2006  
Anonymous Anonymous said...

An idiot said

"Sell Intel Flash to Spansion?????
You must be out of your mind!!
Spanison is atleast 2 generations behind Intel AND has poor reliability.
Flash is about to take off...."

About to take off, where have you been? Beating off in the bathroom? NAND flash has been growing like crazy the past 4 years. Samsung, Toshiba and even Micron-INTEL are playing. NOR is so yesterday like you! Stuck in the bathroom

10:31 AM, August 13, 2006  
Anonymous Anonymous said...

Edward: "Please... Who would use log scale on a yield percentage? Wouldn't that be negative all the way (like a dB graph)? You probably meant "square root" or "cubic root" which would make worse yield look better on the graph? But does that make sense?"

Log plots are used frequently in technnology and in yield analysis to look for tails, look at cum plots that look at via chain resistance.

Perhaps, and I'm just spitballing here, it was a ONE OR TWO DECADE log plot (meaning intercept of X-axis would either be 1% or 10%.

So no, I did not mean square root or exponential... My point is neither you nor I can say what the plot means in terms of actual yield because AMD chose to keep it unlabeled for that very reason. The only point of the graph in AMD's presntation (which it proves) was that AMD now gets to higher yields more quickly on new technologies.

The problem is a bunch of folks like you have re-interperted that point (with no data) to mean they start prodution at yield (which you now concede is wrong) or close to mature yield (which you have no data to backup). With no scale how can you say what the values are? I know I can't, perhaps you work for AMD and have privy to their yield #'s or know someone who does?

Furthermore you have absolutely no idea what the scale of the x-axis on production volume - it could be 100units, 10000unit, or 10,000,000 units across that span (and that scale could be linear or log; but likely linear)

2:59 PM, August 13, 2006  
Anonymous Anonymous said...

"Wouldn't that be negative all the way (like a dB graph)?"

Log scale doesn't imply negative! Do you have an engineering background?

For simplicity, you might want to check out some Yahoo finance charts; these can be done linear or logarithmic scale, and last I checked they weren't plotted in negative dollars.

3:12 PM, August 13, 2006  
Anonymous Anonymous said...

Sharikou: Any comments on the above post which asked why you were comparing your chip capacity analysis, to DESKTOP chip ESTIMATES and proclaiming it to be an "exact" match.

This only holds if mobile and server also went up by "Exactly" 118.5% too, no? or has AMD stop using some of their capcity to make these chips too? :)

3:17 PM, August 13, 2006  
Blogger Sharikou, Ph. D. said...

Any comments on the above post which asked why you were comparing your chip capacity analysis, to DESKTOP chip ESTIMATES and proclaiming it to be an "exact" match.

The growth was basically capacity driven. Mobile should undego the same level of growth. Server will grow at a faster rate, but server units are low, therefore, it won't make much difference in total units.

4:16 PM, August 13, 2006  
Anonymous Anonymous said...

"Log scale doesn't imply negative! Do you have an engineering background?"

Log scale of percentage, which is the "unit" for yield, MUST have negative.

You obviously have no engineering background, or if you do, you have no clue of what I was talking about.

11:29 PM, August 13, 2006  
Anonymous Anonymous said...

"The problem is a bunch of folks like you have re-interperted that point (with no data) to mean they start prodution at yield (which you now concede is wrong) or close to mature yield (which you have no data to backup). With no scale how can you say what the values are?"

You can't say it's close to mature yield, doesn't mean it's not. As a matter of fact, AMD's production did not start at mature yield, that at least was obvious from the graph they showed. While it was no proof that they started close to mature, it was no proof that they didn't, either. Your guess on "log scale," without further evidence, is just another twist from some particular point of view, and surely you can speculative as you wish.

Besides, I knew that log scale were "frequently used in technology" and probably "yield analysis" as well. But applying log scale to a percentage - who would do that if he wasn't trying to mislead the audience? Log scale amplifies the smaller values; for a curve that converges to the highest value (yield maturity), how does log helps looking for trails?

OTOH, there could be many reason for AMD to conceal the y and x axis unit; secrecy, for example; or plain convenience. The point is you (or whoever) have yet to show one evidence for log scale to prove that point.

Even if the gap of AMD's initial yield to mature is HUGE, one thing is still sure - they ramp 90nm SOI to mature more quickly than 130nm and 90nm. How many units did it take for them to ramp 90nm? You can keep spinning your x-axis conspiracy theory you like...

11:48 PM, August 13, 2006  
Anonymous Anonymous said...

"OTOH, there could be many reason for AMD to conceal the y and x axis unit; secrecy, for example; or plain convenience. The point is you (or whoever) have yet to show one evidence for log scale to prove that point."

My point is you cannot conclude anything about absolute yield value WITHOUT ANY LABELS, UNITS OR NUMBERS! You can ONLY compare one curve vs another. You are trying to draw conclusions as to one specific plot by stating "at mature yield" or "close to mature yield", without actual #'s this could mean anything. You have yet to show evidence that it is a linear scale - my point is it could be anything! (although you apparently have some inside info that the rest of us don't).

Also some companies do not simply plot yield in terms of specific % of good die/bad die. Some use more complex analysis, which often are in arbitrary units, which factory in die size and differences between the technology nodes. This is done to compare a technology against a previous technology (Hmmm...kind of like the AMD graph does...). Are you certain the y-axis is in % or are you speculating this? Could it be # of good die on a certain test chip? A normalized # of good die based factoring in product mix? It could be anything...

I'm still looking for YOU to backup your original point that "AMD starts production at mature yield" (which you have since modified to "close to" when that error was pointed out). Unless you define close to mean "within 1 inch as measured when superimposing a ruler on a graph with no units on it", I'm not sure how you can point to that graph and make any conclusions about absolute yield or state close to mature yield.

BTW - the reason there are no units on the graph is that no company releases yield info as you can then back out production costs, etc... a reason why all of Sharikou's CPU cost are just interesting speculation.

2:12 AM, August 15, 2006  
Anonymous Anonymous said...

"My point is you cannot conclude anything about absolute yield value WITHOUT ANY LABELS, UNITS OR NUMBERS! You can ONLY compare one curve vs another."

True, and I didn't base my assumption on the graph, but on what AMD publicly claimed. Look at Daryl Ostrander's analyst's day presentation, which says AMD started 90nm production "at mature yield."

Base on that, plus some leap-of-faith extrapolation, I assumed AMD's 65nm production would start close to mature yield.

Some anonymous raised the graph issue; while graph shows 130nm and 90nm non-SOI not started at mature yield, the gap is smaller with each generation. With again some leap-of-faith extrapolation, I assume AMD's 65nm production would start closer to mature yield than both 130nm and 90nm non-SOI.

9:07 AM, August 15, 2006  
Anonymous Anonymous said...

What is mature yield?

How do you get to a mature yield?

Please pray tell us Doctor, and now riddles just facts us flies can understand please!

6:01 PM, August 15, 2006  

Post a Comment

<< Home