Thursday, September 28, 2006

AMD sold 13000 quad core Opterons

Patty must hate this picture. First, his options are not worth the paper they print on, now his stock grants worth exactly zero.

AMD sold 13,000 quadcore Opterons in one shot. Previously, it sold 19,000 2.6GHZ quad core Opterons. Together, that's over $30 million profit, enough to feed 10K Intel employees for 2 weeks.

These days are getting boring, because AMD is so secretive. There is more on Rev H quad and APM on INQ. A Rev H core will be 60% faster than K8 (40% faster than Conroe) on integer and 200% faster on FP.

110 Comments:

Anonymous Randy said...

Hey Pat, feast your eyes on the first true quad-core, sans super glue and duct tape.

9:39 PM, September 28, 2006  
Anonymous Anonymous said...

His stock grants will be worthless because of this too.

http://www.theregister.co.uk/2005/10/28/intel_whitefield_india/

I was so happy to read that article. It meant two things for me:
1. My investment in AMD is sound.
2. The Intel people in my home town got to keep their jobs.

-Longan-

9:41 PM, September 28, 2006  
Anonymous theKingRich said...

Nice!
Didn't Intel say they will ship 1 Million Quads before AMD ships a single one?
HA! All they have to do is sell a couple wafers worth of Quads to beat Intel to the punch, no matter how soon the full Intel production kicks in! Beautiful!

Hey, how about HP buying VoodooPC?

10:06 PM, September 28, 2006  
Anonymous Anonymous said...

No matter how good 4x4 is supposed to be, no matter how good AMD 65nm dual core are supposed to be, no matter how good K8L are supposed to be you only make money buy selling things and not showing pictures.

AMD looks to have some good upcoming products but the time it will take to get them rolling will let Intel make hay while the sun shines.

10:54 PM, September 28, 2006  
Anonymous The Sheepshagger said...

Interesting article and great picture! Glad to see they're in the game but let's not forget a few things:

The article referenced the wafer to be PRE-PRODUCTION. That being said and all things being fair and equal, all companies can produce PRE-PRODUCTION wafers to pass around at press conferences. The real test is showing POST-PRODUCTION chips... in a system.

I'm not looking for a beat down for pointing that out but I didn't happen to see in that article or the other articles I've read when the AMD will be shipping the aforementioned monster. One would think The INQ and Charlie in particular would have been all over heralding that news.

The other thing I'm left with thinking is "HOLY SCHNIKEY!!! THAT'S A BIGGER CHIPPER THAN EXPECTED!!!" Which should roughly equate to "HOLY SCHNIKEY!!! HOW IS AMD PRODUCE THAT IN VOLUME *AND* WITHOUT ROBBING FROM OTHER PRODUCT VOLUMES *AND* YIELD ENOUGH VOLUME TO MAKE IT COMPETETIVE?!"

Seriously... that's a big chip and it won't take more than a defect here or there on one wafer to render the equivalent number of die unsellable. I just hope they know what they're doing.

12:27 AM, September 29, 2006  
Anonymous Anonymous said...

Yup, that's helliuwa huge chip even on 65nm... AMD is gonna be in the world of pain.

1:31 AM, September 29, 2006  
Blogger hyc said...

Seriously... that's a big chip and it won't take more than a defect here or there on one wafer to render the equivalent number of die unsellable. I just hope they know what they're doing.

If you actually count the dies, you'll see that it's nearly 50% more than AMD was getting out of Opterons in 130nm on 200mm wafers. I think they're doing fine.

1:37 AM, September 29, 2006  
Blogger Sharikou, Ph. D said...

If you actually count the dies, you'll see that it's nearly 50% more than AMD was getting out of Opterons in 130nm on 200mm wafers. I think they're doing fine.


I estimated the die size to be about 225m^2. Each 300mmafer can produce about 220 dies. To take 100% of the server market with K8L, AMD needs 3000wspm. FAb36 is reaching 17000 wspm by 1Q07.

2:37 AM, September 29, 2006  
Anonymous Anonymous said...

hmmm.. These are dies on wafers.. Not actual processors! Get the difference?

3:56 AM, September 29, 2006  
Blogger S said...

At last some hardware to show after all the vaporware

3:56 AM, September 29, 2006  
Anonymous Anonymous said...

100% of the server market? Ha ha ha....oh wait I forgot, intel will be BK before the end of next year. Your too much. Thanks for the laugh DR. Clueless.

4:00 AM, September 29, 2006  
Anonymous Anonymous said...

225mm^2 is quite low IMHO.

From what I've measured it comes a little above 300mm^2 - that's ~204 complete dies, as I count so.

4:01 AM, September 29, 2006  
Anonymous enumae said...

Sharikou said...

"I estimated the die size to be about 225m^2. Each 300mmafer can produce about 220 dies. To take 100% of the server market with K8L, AMD needs 3000wspm. FAb36 is reaching 17000 wspm by 1Q07."

Could you define when the first quarter starts?

It has already been established that they will try and be at 15,000 wspm at the beginning of the year with there goal to reach 25,000 by the end of the year 07.

10,000wspm/12months=833 wspm

Jan. = 15,000
Feb. = 15,833
Mar. = 16,666
Apr. = 17,499

So around the middle of March would be the best case scenario to reach 17,000 wspm.

Also keep in mind they will not be fully converted to 65nm until July 07.

7:09 AM, September 29, 2006  
Blogger Stephen said...

Photochop. No way in hell the Inq of all people got an exclusive view of quad-core AMDs when AMD has been so tight-lipped about them.

9:09 AM, September 29, 2006  
Anonymous The Sheepshagger said...

I'm not one to split hairs about things I don't know about so I Googled to see exactly what is AMD's current share of the market for server? What I found is that on July 31st, 2006 AMD had 26%. Also in that article I read that was an impressive 9% and change jump from where they were at the end of Q2'05.

Okay... so now I'm really starting to wonder... if it took 12 months from Q2'05 to Q2'06 to get 9% more of the market... Can AMD supply 100% of the server market with the K8L while simultaneously capturing the 74% that Intel currently owns?! It sounds very altruistic to me. I mean seriously… the K8L isn’t even ramped to 3000 WSPM. Unless you can find something to help educate myself, I’ll have to wait until Q2’07 to see how well this race pans out will do.

I’d also be curious if you could find somewhere (because I sure cannot) the number of server chips sold weekly. I suppose once we find the number sold yearly we could just divide by 52 weeks to get a rough estimate.

9:52 AM, September 29, 2006  
Anonymous The Sheepshagger said...

Not to add fuel to the fire, but... did you take a look at this gem this morning?

Toshiba joins massive recall of Sony batteries

I believe this is the 3rd company to announce a Sony battery recall behind Dell and Apple. Or is this just the 3rd company that was using exploding Intel chips from your point of view?

I stand corrected... yesterday Lenovo annuounced a recall, too.

9:53 AM, September 29, 2006  
Anonymous Edward said...

AMD didn't "sell" 19,000 quad core Opterons. It sold 19,000 dual core Opterons which are upgradable to quad cores.

There's no reason to believe that the upgrade won't take place some time in the future, though.

On the TACC supercomputer: "The initial configuration of this system will go into production on June 1, 2007, with the final configuration in operation by October 2007." This is about in-line with previos knowledge that quad-core Opteron systems will be available 2Q 2007.

11:03 AM, September 29, 2006  
Anonymous Anonymous said...

All I see is AMD taking pre-orders for vaporware. Seen any working processors yet?

AMD quad cores have not even started production yet. Is this how AMD is going to make its number? By selling products it cant yet make?

Sure is nice to see the Pee H Dee reduced to desperate hopes for future products. Its a bakhanded admission that AMD's current products suck.

12:13 PM, September 29, 2006  
Anonymous Anonymous said...

http://www.engadget.com/2006/09/28/intel-announces-specs-production-dates-for-classmate-pc/
http://www.engadget.com/2006/07/27/olpc-update-india-isnt-buying/
India doesn't like AMD.
http://www.engadget.com/2006/09/28/intel-announces-specs-production-dates-for-classmate-pc/
"the machines are slated to hit full scale production in "Q1 2007," and orders have reportedly already been placed by governments in Mexico, Brazil, Nigeria and India (they've found a winner, apparently"
Score Intel!

12:30 PM, September 29, 2006  
Anonymous Anonymous said...

"Together, that's over $30 million profit, enough to feed 10K Intel employees for 2 weeks."

This ironically indicates the profit is too small instead of huge.

12:35 PM, September 29, 2006  
Anonymous Anonymous said...

Yo PhD pretender and flies.

What a company..

The first picture of a rumor'd but not annouced product is a picture of a wafer in a clamshell sitting in someones car. Tell you something about either the control of information or how they do business with the valuable wafers.

2:52 PM, September 29, 2006  
Anonymous Anonymous said...

Obviously SUN have quads in the lab up and working

2:57 PM, September 29, 2006  
Anonymous Anonymous said...

A preorder for 10's of thousands of pre-production units, and you think it's grand.

Core2 CWM was preordered in the millions about the time your were claiming it was all vaporware.

You are one dizzy lizzy.

6:32 PM, September 29, 2006  
Blogger Sharikou, Ph. D said...

Core2 CWM was preordered in the millions

No. DELL only sells a dozen Core 2 Duos a day. that's why it went AMD.

6:35 PM, September 29, 2006  
Anonymous Anonymous said...

Obviously AMD is trying to get all the good press it can get, right around IDF time, when INTC shot up, AMD dove down.

6:37 PM, September 29, 2006  
Anonymous Anonymous said...

No. DELL only sells a dozen Core 2 Duos a day. that's why it went AMD.

Link? Evidence? At least make it sound remotely believable. If you said 12,000 a day, that might at least reek of believable. Or is your logic exceeding my limited mental capability?

You wonder why no one takes you seriously? Hint: Go back and read your statement.

8:15 PM, September 29, 2006  
Anonymous The Sheepsshagger said...

"No. DELL only sells a dozen Core 2 Duos a day. that's why it went AMD."

Good to see you stirring the pot in an entertaining way. I was disappointed you didn't have replies to my previous comments.

That being said... I don't suppose you'll share with us a link support your claim about a dozen C2D's a day by Dell. As I'm sure you already calculated 12 C2D's a day for 60 days is only 720 chips that Dell moved of the 5 million C2D's that Intel produced/sold/shipped in the first 2 months it was available. I reckon things in Round Rock are worse off than we all initially suspected.

8:50 PM, September 29, 2006  
Anonymous Anonymous said...

...of the 5 million C2D's that Intel produced/sold/shipped in the first 2 months

I don't suppose you would share where you got your numbers from.

9:50 PM, September 29, 2006  
Anonymous Graham said...

100% of the server market? Ha ha ha....oh wait I forgot, intel will be BK before the end of next year. Your too much. Thanks for the laugh DR. Clueless.

Amen! Bravo!
Couldn't have said it better myself...

11:31 PM, September 29, 2006  
Anonymous Anonymous said...

Pat Gelsinger, at IDF, stated Intel have shipped 5 million Core 2 CPUs in the sixty days since they shipped the product.

1:12 AM, September 30, 2006  
Anonymous The Sheepshagger said...

I don't suppose you would share where you got your numbers from.

But of course I will. I always back up things I say when asked. You can Google "Intel 5 million 60 days" or you can read the following snippet from this link:

As part of its counter attack to win back market share, Intel launched a triple-chip assault against AMD over the summer, unveiling a new microprocessor architecture dubbed Core. Intel is pushing those new chips for laptops, desktops, and servers out of it factories faster than ever before.

"It's the fastest 60-day ramp in Intel's history," Otellini said. He said Intel has shipped more than 5 million Core 2 Duo processors for PCs since late July and 1 million server chips called Xeon 5100. All the processors run on dual-core technology.

2:51 AM, September 30, 2006  
Anonymous The Sheepshagger said...

BTW... Here are links to 65 total news stories quoting the same 5 million C2D chips moved in 60 days. Needless to say, with that amount of coverage Intel will probably be in for a world of hurt if it does pan out when they announce the're earnings in a few weeks.

3:01 AM, September 30, 2006  
Anonymous Anonymous said...

Those numbers are everywhere. Just watch any IDF presentation at a hardware-site !

4:33 AM, September 30, 2006  
Anonymous Anonymous said...

http://www.marketwatch.com/news/story/Story.aspx?guid=%7B0127628C-D488-4C56-80BB-42752CE09286%7D&siteid=


"It's the fastest 60-day ramp in Intel's history," Otellini said. He said Intel has shipped more than 5 million Core 2 Duo processors for PCs since late July and 1 million server chips called Xeon 5100. All the processors run on dual-core technology."

5:53 AM, September 30, 2006  
Blogger 180 Sharikou said...

While we're talking about cheap PCs for emerging markets. AMD's 50x15 took a hit when their PIC bombed in India. Tata - the first ISP to come on board had to dump the product when they found even value conscious consumers in India were unwilling to take a product that doesn't deliver the rich experience they expect from a PC. That and something that anyways looks like a plastic toy put them off.

Till date - I don't think anybody has found the solution to this problem. Even Intel's Classmate PC which Otellni showed off during IDF is not necessarily going to cut it. But it's a darn sight better than the 100$ piece of junk Negreponte is peddling.

12:12 PM, September 30, 2006  
Blogger Sharikou, Ph. D said...

180 Sharikou said

Hi. I think you need to change your name to avoid dilution of my trademark.

4:26 PM, September 30, 2006  
Anonymous Anonymous said...

"If you actually count the dies, you'll see that it's nearly 50% more than AMD was getting out of Opterons in 130nm on 200mm wafers. I think they're doing fine."

So 50% more after gong to 300mm (which is 2.25 Si area increase) and 65nm (which is theoretically a 2X improvement - except of course IBM's reported literature which shows more like a 1.6X improvement).

So they are getting 50% more die when they should be getting 200% more? Might also want to keep in mind that 200mm fab has been fully paid for for some time....300mm (specifically F36) has 4 years of payments coming due...

4:56 PM, September 30, 2006  
Anonymous Anonymous said...

The most recent laptop recall at Toshiba includses laptops with AMD and Tramsmeta processors in them.

Looks like the "dokter" is going to have to eat a little crow.

9:36 PM, September 30, 2006  
Blogger Mikolaj said...

It looks like 4x4 is a complete crap. AMD is going to sell Opteron series with registered DDR2 memory as FX series. This also means that AM2 is a mistake. AMD should stay with 939 socket and focus on Athlon core improvents. Instead of AM2 "achievment" we should be able to get K8L faster - how about this month with first DDR2 memory support.
So - AMD management and engineers are humans...

6:32 AM, October 01, 2006  
Anonymous Anonymous said...

From what I understand 4x4 "Quadfather" will use LGA1207 processors, the same socket as the Opteron. They will not, however, require registered or ECC memory. Just plain DDR2 ram will do fine. You will need the FX series processors though, which will not be cheap.

9:01 AM, October 01, 2006  
Anonymous enumae said...

Sharikou said...

"A Rev H core will be 60% faster than Conroe on integer and 200% faster on FP."

You had said previously...

"On performance of the cores, we already know, 60% increase in integer speed, and 200% increase on floating point. Dirk Meyer told us AMD will have a faster core in 2007. He was true to his promise."

K8L (Rev H) server/desktop for 2Q07
Saturday, September 16, 2006

Now it seems you are saying the same but over Conroe, do you have a source/link for these assumptions or are you just trying to make conversation?

I call bullsh!t.

I have run these numbers based on K8L vs Kentsfield, and they were run factoring in your 60% and 200% improvements on K8 not Conroe...

The AMD K8L Quad core has a 28% performance lead in Multimedia Floating Point over Kentsfield.

And the Intel Kentsfield Quad core has a 45% performance lead in Multimedia Integer over K8L.

The other factor is clock speed, the numbers above are based upon 2.8GHz, but can K8L reach 2.8GHz?

Theoretcally a 2.5 - 2.6GHz K8L is possible in the 125 - 130W envelope that is currently used by AMD.

Keep in mind we have not factored in the 4MB of L2 and L3 cache or the improved manufacturing process AMD plans to use, the numbers could get better or worse.

Looking at all of this Kentsfield and K8L are each going to outperform the other in certain applications, so the whole K8L is going to destroy Kentsfield and visa versa is a bunch of crap.

The processors will/should be very competitve against one another.

9:43 AM, October 01, 2006  
Blogger Sharikou, Ph. D said...

Now it seems you are saying the same but over Conroe

You are right. I stand corrected.

10:01 AM, October 01, 2006  
Anonymous Anonymous said...

AMD hasn't "shipped" anything. They showed a WAFER.

Intel has been showing the QuadCore since January and a few sites have been testing.

One site took 2 Xeon Quadcores and dropped them in a MacPro. Guess what?

They worked with no problems. OS X showed 8 processors.

Still, we haven't seen squat from AMD on K8L (should be called K9, and this dog don't hunt).

I've noticed that the Doc has been posting more this week.

Are the kiddies out of school for holiday?

1:31 PM, October 01, 2006  
Anonymous Anonymous said...

http://www.theinquirer.net/default.aspx?article=34433

The Q4/07 part is called Budapest, and it is the Barcelona core with HT3.0 in 12xx Opteron and A64 guise

http://www.hkepc.com/bbs/news.php?tid=675864&starttime=0&endtime=0

"The model is codenamed Yorkfield and expected to release in the second half of 2007. Similar to Conroe, Yorkfield will be a hybrid Quad-Core processor which shares the same L2 Cache, such that it can increase the cache hit and lower the loading of FSB.

Yokrfield features 1333MHz FSB, DDR3-1333/DDR2-800 and dual PCI-Express 2.0 interface. Coupled with Bearlake X chipset, Yokrfield perhaps is the strongest CPU in 2007. Intel is now co-operating with software developer to optimize Quad-Core products.

http://www.dailytech.com/article.aspx?newsid=3830

Bearlake-X coming Q307. Intel has pulled up products 1-2 quarters early lately.

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2768&p=3

AMD quads are designated for 'performance'[FX($999)] only.

http://www.xbitlabs.com/news/cpu/display/20060919051744.html

Intel plans on 'mainstream' quad.
---
Summarization, Intel has desktop native quads by[since they've released early recently] Q307, AMD at earliest[since they delayed Socket F] Q407.

1:56 PM, October 01, 2006  
Anonymous Anonymous said...

TACC is partnering with Sun Microsystems to deploy a supercomputer system specifically developed to support very large science and engineering computing requirements. In its final configuration in 2007, the supercomputer will have a peak performance in excess of 400 teraflops, making it one of the most powerful supercomputer systems in the world. It will also provide over 100 terabytes of memory and 1.7 petabytes of disk storage. The system is based on Sun Fire x64 (x86, 64-bit) servers and Sun StorageTek disk and tape storage technologies, and will use over 13,000 of AMD's forthcoming quad-core processors. It will be housed in TACC's new building on the J.J. Pickle Research Campus in Austin, Texas.


All I have to say lots can change in a year. Dont count your egg until they hatch.

4:50 PM, October 01, 2006  
Anonymous Anonymous said...

"The most recent laptop recall at Toshiba includses laptops with AMD and Tramsmeta processors in them.

Looks like the "dokter" is going to have to eat a little crow."

Crow? What crow - the recent issues are CLEARLY battery issues (as they impact AMD systems). He's already conclusive proven the explosions were the actual chips explooding, remember 2 explosions - 2 cores? I say it is clearly the battery because we all know by now that Intel chips explode and AMD chips, through their superior design/IMC/HT/APM3.0, don't explode.

All the "doktor" needs is to make up another BS explanation draped in bad logic to cover up his previous lies.

6:08 PM, October 01, 2006  
Blogger Sharikou, Ph. D said...

"The most recent laptop recall at Toshiba includses laptops with AMD and Tramsmeta processors in them.


If this is true. It's further proof that Intel CPUs were the direct cause of explosion. When you pair an Intel CPU with a defective battery, the notebook explodes. When you pair an AMD CPU with a defective battery, everything is fine.

8:47 PM, October 01, 2006  
Anonymous The Sheepshagger said...

"If this is true. It's further proof that Intel CPUs were the direct cause of explosion. When you pair an Intel CPU with a defective battery, the notebook explodes. When you pair an AMD CPU with a defective battery, everything is fine."

No offense, Sharikou... but the quote the previous anonymous person left read:

""The most recent laptop RECALL at Toshiba INCLUDES laptops with AMD and Tramsmeta PROCESSORS in them."

I concede I didn't see a link from that person to say it is true or not, but I just want to make sure you understood the point they were making.

When you offered the following:
"When you pair an Intel CPU with a defective battery, the notebook explodes. When you pair an AMD CPU with a defective battery, everything is fine."

I was left to think that you wanted your readers to believe that it was an accident for an Intel CPU to explode, but AMD it was a design rule for AMD chips to explode and they were flawlessly executing on that role.

I know that is all tongue and cheek at your expense, but I bet Intel would love if AMD created exploding chip benchmarks. AMD could spin this latest announcement in their favor by announcing that they (AMD) are winning the explosion per watt benchmarks.

10:45 PM, October 01, 2006  
Anonymous Anonymous said...

This whole core duo exploding and intel going BK has lost all the respect that I had once for you. You sir are a fraud and disgrace. You are doing more harm than good.

10:52 PM, October 01, 2006  
Anonymous Anonymous said...

"When you pair an AMD CPU with a defective battery, everything is fine."

Hmmm...makes you wonder why they are going through the trouble to recall the batteries that are in the AMD systems; one would think if "everything is fine" they would only recall batteries in the Intel models. Must be yet another part of the vast Intel conspiracy.

Boy, someone is really spending a lot of money covering up the fact that it is not the batteries exploding but the actual chips as you have previoulsy "scientifically" proven. Can you explain to us again your theory on how Si chips explode and the conclusive visual evidence that the batteries stay intact? It was just so elegant that my feeble mind couldn't comprehend how Si chips explode - something about Si vaporizing and gas not being able to escape?

1:46 AM, October 02, 2006  
Anonymous Anonymous said...

Intel FUD fan boys,
if Intel production is so great why did they announce 1/40 of their 65nm production is core 2.

Is that impressive? Not for me.

Want links? Find them your self.

1:51 AM, October 02, 2006  
Blogger S said...

"When you pair an AMD CPU with a defective battery, everything is fine."

So why the recall of the batteries of AMD based laptops ?

Can't you see the recalls were only for batteries. It the CPU had any role in the explosions, the whole laptop would hv been recalled. That's not been the case.

2:45 AM, October 02, 2006  
Anonymous Anonymous said...

That makes no sense. How is everything fine if you're running with a defective battery?

7:20 AM, October 02, 2006  
Anonymous Anonymous said...

I got a question, why waffers are rounded "circular" insteath of square?
I mean if you use square, it will be most likely to get a 0% broken dies. while I see that rounded waffers gets around 10 to 15% losses due of the laser cut

9:07 AM, October 02, 2006  
Anonymous Anonymous said...

Now just look what those idiots say..

http://www.theinquirer.net/default.aspx?article=34806

DJKyp wrote to you today with the wolloing diatrabe: "There's no way Intel CPUs would cause batteries to explode, period. If it draws too much power, it's just going to shut down, not explode."

But don't you worry Sharika. We know you are right about that it IS the Intel chips that made the batteries explode. Don't worry what those guys say, Go get'em!

9:46 AM, October 02, 2006  
Blogger Sharikou, Ph. D said...

I was left to think that you wanted your readers to believe that it was an accident for an Intel CPU to explode, but AMD it was a design rule for AMD chips to explode and they were flawlessly executing on that role.


Dudes, you can't even follow my simple logic, how can you do anything mildly productive. Battery recall doesn't mean there is an exploding laptop. They may recall a laptop because the cover is loose...or battery is defective. But as I reasoned before, a battery won't explode by itself, the explosion was caused by something else.

Try compute on the following facts

1) There is no AMD latop exploded
2) There are tons of Intel laptops exploded
3) Both AMD and Intel latops are actually using similarly "defective" batteries.

Now, the question, why Intel laptops exploded?

Go back to the original analysis I made. And the conclusion was that Intel CPUs were the direct cause of explosions of the intel laptops.

11:21 AM, October 02, 2006  
Anonymous Anonymous said...

Perhaps I'm not updated at all, but last I recall, Toshiba does not sell any AMD-based laptops. I had a "Satellite" series from them back in the mobile K6-2 days, but after that, things went downhill ...

3:37 PM, October 02, 2006  
Anonymous enumae said...

Sharikou said...

"A Rev H core will be 60% faster than K8 (40% faster than Conroe) on integer and 200% faster on FP."

ASUS FX-62 SPECint_rate_base2000

Dell X6800 SPECint_rate_base2000

DC = dual core

43.7 FX62 - Integer DC

60.9 X6800 @ 2.8GHz - Integer DC - 28% advantage for X6800

FX62 - 69.9 with the 60% improvement = 13% advantage for Rev H DC

I do not see a 40% advantage in integer performance over Conroe with Rev H.

Can you elaborate on where you get 40%?

7:21 PM, October 02, 2006  
Anonymous Anonymous said...

"But as I reasoned before, a battery won't explode by itself, the explosion was caused by something else."

Actually your REASONING before was that the battery was intact and the chip itself exploded due to vaporization of the Si - do you really want me to dig up the blogs again? Let me know and I will.

Your first conclusion was not that the chips caused the explosion, but it was the chips themselves that were exploding. Remember "2 cores,2 explosions, it doesn't take a genius to figure it out" comment. You've only adapted this conclusion when the complete idiocy of your first statements were pointed out.

7:38 PM, October 02, 2006  
Anonymous Anonymous said...

"if Intel production is so great why did they announce 1/40 of their 65nm production is core 2."

Got a real rocket scientist here. Intel just reached crossover between 90nm and 65 nm (menaing 65nm is now slighlty greater thatn 50%)

1/40 of ~50% is ~1.25%.

All Core 2 production is on 65nm.

So now Intel's Core 2 production (server/mobile/desktop) is ~1.25%?

Buddy your about 6 beers short of a six-pack....

7:50 PM, October 02, 2006  
Anonymous Anonymous said...

This blog is so tedious and boring with your ridiculous lies.

The reason no AMD laptops are exploding is because neither one happened to be in the right conditions. That's right. There are only 2 AMD laptops in existance, nobody wants them.

Compared to the number of Intel based laptops, there are _2_ AMD laptops out there.

9:20 PM, October 02, 2006  
Blogger nyx said...

AMD is definitely too secretive. They mostly allow others to publicize their products. It's funny to see the differences in approach towards marketing between AMD and Intel. Intel seems to be driven by sheer desperation with their marketing blitz which doesn't seem to be helping much for their stock price. AMD seems to be a lot more reserved, allowing others to praise them unlike Intel's constant self-praise lately. It hurts because I cannot rely on periodic earnings releases to raise AMD's stock value. Instead I have to rely on news releases to raise public interest to get AMD's stock price to charge. However, AMD rarely releases news compared to Intel's bombardment of how great they are. At least with AMD, I can expect an eventual stock price increase. Whereas with Intel, stock jockies have to pray to devine Gods that Intel won't screw up yet again and cause their stock price to slip several more dollars. Within the barrage of self hype, Intel investors always seem to get slammed with an equal amount of negative news. Failures tend to pile up in Intel's camp. Everything from mass firings, corporate amputation of some division or another, the existence of some flawed aspect to their architectural approach, market disapproval in their ability to maintain corporate alliances, inability to meet deadlines, disorganized distribution channels, insanely high inventories in archaic technologies, an increase in third world out sourcing, and the list goes on. Take your pick. A tribe of retarded pygmies strung out on a cocktail of lithium and heroine have better managerial skills than what Intel has been exibiting.

9:32 PM, October 02, 2006  
Anonymous Anonymous said...

"Go back to the original analysis I made. And the conclusion was that Intel CPUs were the direct cause of explosions of the intel laptops."

I understand you perfectly. You are saying you have no idea what you are talking about, but you will love AMD even when they are slower, run hotter, use more power and are all made overseas.

Did I miss anything, herr doktor?

10:06 PM, October 02, 2006  
Anonymous Anonymous said...

OH!!! come on this site must be a joke. Intel bankrupt and processors exploding. Really people get a life.

11:44 PM, October 02, 2006  
Anonymous Anonymous said...

"if Intel production is so great why did they announce 1/40 of their 65nm production is core 2."

“Next week Intel will have shipped its 40 millionth 65nm chip, and it was quick to point out that the rest of the industry has yet to ship a single 65nm chip. Of its 40M 65nm chips, only about 5 million of them have been Core 2 based processors, really putting into perspective the vast number of Intel processors that are still NetBurst or Yonah based.

the link:
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2840&p=2

"So now Intel's Core 2 production (server/mobile/desktop) is ~1.25%?"

Yes. Until month end (September).
Year end ~20%.

1:45 AM, October 03, 2006  
Anonymous Anonymous said...

Anonymous said...
...
The reason no AMD laptops are exploding is because neither one happened to be in the right conditions. That's right. There are only 2 AMD laptops in existance, nobody wants them.

Compared to the number of Intel based laptops, there are _2_ AMD laptops out there.


Well, ever think of that it is the CPU that draws that power and the bad desogn of the cell of the batteries caused the explosion?

BTW, are you saying there are only TWO AMD notebooks on this world?

Okay, here's a list of AMD Turion 64 X2 notebooks, here goes:-
1. HP Compaq nx6325
2. HP Pavilion dv6040us Notebook
3. Compaq Presario V6000Z
4. Acer Ferrari 1004WTMi
5. Voodoo ENVY Featherweight a:228
Should I list more?

And a list of AMD Turion notebooks:-
1. Fujitsu LifeBook S2110
2. HP Pavilion DV5223OM
3. Voodoo ENVY Heavyweight u:909
4. Acer Aspire 5003WLMi
5. Alienware Aurora m9700
More?

Oh, BTW, there are mobile sempr0n notebooks:-
1. Compaq Presario V2555US
2. Compaq Presario v2710us
3. Averatec 3715-EH1 Notebook

So, I managed to find all these products over the net, and you tell me there are only TWO AMD notebooks?
Please do some more research before you tell lies to the others. DUMBASS!!

2:18 AM, October 03, 2006  
Anonymous Anonymous said...

"I understand you perfectly. You are saying you have no idea what you are talking about, but you will love AMD even when they are slower, run hotter, use more power and are all made overseas.

Did I miss anything, herr doktor?"

wtf.. are you one of these idiotic "patriots"? what, is germany planning to teorrise a unknown city in some hick state?, or are they just spying on you?, They want you're BBQ sauce recepy!(spellcheck) :o OMFG!!

when will people learn that we need AMD to compete with Intel, and that Sharikou and other people like AMD because it's a innovative company and it seems like a better company when it comes to moral, but that we cannot be sure of..

//Kim Leo

3:01 AM, October 03, 2006  
Blogger S said...

"Now, the question, why Intel laptops exploded?"

Because Intel laptops were being used as computers. AMD laptops were being used as paperweights :)

3:14 AM, October 03, 2006  
Anonymous Anonymous said...

A recent Custom PC review of AM2 3800 X2 vs C2D 6300 found the C2D had about 10% overall performance advantage. However the AM2 was £100 while the C2D was £130. They concluded the C2D was "clearly the better option" ??

Another one in Intel's pocket...

4:57 AM, October 03, 2006  
Anonymous Anonymous said...

Intels run fine on a non defective battery..

7:42 AM, October 03, 2006  
Anonymous DaSickNinja said...

What I don't get is why you're so determined to see Intel go bankrupt.
If in fact Intel goes bankrupt, which is highly unlikely due too many factors, that would leave only AMD and VIA as competitors in the marketplace.
You think your beloved AMD is as white as snow but you forget that in order to be in business in this free market, you must be dirty. If you want to be in business and still have morals, you form a nonprofit entity.
AMD if Intel goes bankrupt would become a virtual monopoly. You can't even say VIA will be competition with a straight face, can you?
Without the opposing force of Intel, the development of CPU's will stagnate.
Surly if you open your eyes you'll see this.

11:43 AM, October 03, 2006  
Anonymous Anonymous said...

http://www.dailytech.com/article.aspx?newsid=4422

"The 65nm chips are expected to sample in December 2006, with official shipments still slated for Q1'07."

...

12:47 PM, October 03, 2006  
Anonymous Anonymous said...

http://www.aceshardware.com/forums/read_post.jsp?id=120066919&forumid=1
I've found the Bunny from the Inquirer!
His posts say 65nm K8s for sure.
http://www.aceshardware.com/forums/read_post.jsp?id=120066245&forumid=1
9.11.06..
"No, AMD is not about to put out anything new with any new goodies for over a year. By then Intel will have the Penryn cores out and those WILL have a lot of goodies. Have your mother put some Intel roadmaps on the top basement step when she places your food there."
http://www.aceshardware.com/forums/read_post.jsp?id=120067118&forumid=1
He at least got the Sun/AMD common socket thing right:D
http://www.aceshardware.com/forums/read_post.jsp?id=120066879&forumid=1
"I still expect Conroe to beat it core for core in all cases, but AMD has a chance when the platforms are taken into account, IE >1 core systems."
Intel still ftw in desktops!
http://www.aceshardware.com/forums/read_post.jsp?id=120067743&forumid=1
Says ATI working on raytracing, as is Intel.
http://www.aceshardware.com/forums/read_post.jsp?id=120065810&forumid=1
Claims lots of K8L ES in Taiwan.
http://www.aceshardware.com/forums/read_post.jsp?id=120066243&forumid=1
Anyone doubting Intel's smarts are in for a shock[in regards to phase change RAM]

7:17 PM, October 03, 2006  
Anonymous Anonymous said...

All the laid off Intel'ers are here in full force hitting their daily pinata... :)

"I got a question, why waffers are rounded "circular" insteath of square?"

If you understand how wafers are made, then you'll know that it's slowly pulled from a silicon ingot... if you can figure out a way to make square wafers, you got yourself a nice patent. ;)

Never mind that the whole industry will have to change their tools that serve these "round" wafers.

9:51 PM, October 03, 2006  
Blogger nyx said...

Four-core vs. 4x4

"AMD's 4x4 system features two coherent HyperTransport (cHT) AM2 sockets, each of which can gluelessly support a dual-core Athlon CPU and a pool of DDR2. So a 4x4 system gives you two cores and one DDR2 bank per socket. In contrast, a Kentsfield system puts all four cores in the same socket, and connects them to a single DDR2 pool via the chipset. This means that the 4x4 system is going to have a higher aggregate system bandwidth, because it has more buses and more sockets. More importantly, the 4x4 system will also have more memory and frontside bus bandwidth available to each individual core.

What all this means is that the four cores in the 4x4 system will be able to access data at a higher rate and with lower latencies than the cores in a comparable Kentsfield system, a factor that could very well overcome Kentsfield's microarchitectural advantages for some bandwidth-intensive categories of applications.

However, the tradeoff here is that all those sockets and buses don't come for free. The single-socket Kentsfield will be cheaper to implement, so Intel will probably be able to give you a quad-core system for less money than AMD. In fact, depending on how both systems are priced, the Kentsfield may offer more bang for the buck than a comparable 4x4 system.

On the level of system topology, my other concern with Kentsfield is that it pairs a quad-core socket with a pool of DDR2. There's no doubt that Kentsfield is going to be starved for memory bandwidth, because a quad-core Conroe-based part like that will have a ton of execution hardware in a single CPU package. Kentsfield would be better of with FB-DIMMs or even an RDRAM flavor, but that's not going to happen at launch. In this respect, the Clovertown server part, because it's paired with FB-DIMMs, will be much better equipped to face off against competing quad-core, two-socket Opteron systems."

http://arstechnica.com/news.ars/post/20060720-7310.html

And it looks as though one will be able to build an AMD 4X4 system for under a $1,000 dollars taking into account projected prices. Very exciting times. I can't wait until the Nov-Dec time frame. I wish AMD pushed out their 65 nm line faster, but oh well. At least now they'll be able to step it up to 45 nm faster than it took them to go from 90 nm to 65 nm. They have the infrastructure in place now. Software developers need to step up and push multi-threaded programs. I don't know why they have been so complacent. Only a few of my programs and games are multi-threaded and yet we've had dual core processors for quite some time. With the advent of quad core processors and enthusiast 4X4 multiple dual core processor systems, software developers better step up the pace with better support for newer technologies. There's no excuse for this complacency and I blame AMD and Intel for not pressuring software developers enough. At IDF Intel proclaimed there was a great deal of support for multi core systems with software developers focused on multi-threaded programs, yet I have seen no evidence of any abundance of support. All I have seen are a few bright enthusiastic developers taking the initiative, yet surrounded by complacency as the majority push antiquated single threaded 32 bit programs. Totally unacceptible and it's in the best interest of AMD and Intel to encourage more interest in this matter.

10:00 PM, October 03, 2006  
Anonymous bgt said...

was just using my sons Acer Travelmate 800 series laptop with a 1,6Ghz Centrino CPU yesterday and doing a Windows/Accesoires/systemtools/ diskcleanup when after having done about 20% the laptop switches off completely. Temp reading in speedfan 95C. Battery boiling hot. Coincidence?? The air output of the CPU cooler was smelling burned. The unit was running on batteries.
I was really shocked, my son said he never noticed. But than he uses it not very highly loaded. Laptop is about 3 years old.
Maybe a good burn in test? Laptop had a huge temp. file.

12:55 AM, October 04, 2006  
Blogger realgenius said...

It really does not matter how many chips you produce and ship, what matters is how many you sell. Intels production system seems to create massive inventories while AMD is just in time production. Sales figures are what count and I think it is safe to say that if Intel continues to loose market share to AMD this quarter, there will be big trouble at Intel.

1:01 AM, October 04, 2006  
Anonymous Anonymous said...

DaSickNinja said...
What I don't get is why you're so determined to see Intel go bankrupt.
If in fact Intel goes bankrupt, which is highly unlikely due too many factors, that would leave only AMD and VIA as competitors in the marketplace.
You think your beloved AMD is as white as snow but you forget that in order to be in business in this free market, you must be dirty. If you want to be in business and still have morals, you form a nonprofit entity.
AMD if Intel goes bankrupt would become a virtual monopoly. You can't even say VIA will be competition with a straight face, can you?
Without the opposing force of Intel, the development of CPU's will stagnate.
Surly if you open your eyes you'll see this.


Yeah, yeah, markets are like that, bankrupcies happens everyday, no need to be so angry like that.
Well, to see some company BK is his own opinion.
Why do someone need to show you his/her reasons when telling his opinions?
I don't get it also.
Maybe you don't want to see Intel go BK, cuz you want to see competitions in the market? or you just want to see that Intel beating up AMD like hell?
Reasons?

Can't think of any. :P
BTW, Good job doc, go make everybody laugh.

3:24 AM, October 04, 2006  
Anonymous Anonymous said...

nyx, do you own a blog?
cuz I'd like to see you bloging with the information you have on hand, not in replies, but in your own blog.

3:26 AM, October 04, 2006  
Blogger Mikolaj said...

AMD is late, very late.
K8L is going to be introduced in 5 months in Opteron series and 11 months for desktop. How can you explain
that?


They plan to produce 55mln CPUs next year - only 25% more than this one.
During Q1 and Q2 07 we are going to sleep.

8:29 AM, October 04, 2006  
Anonymous Edward said...

"If in fact Intel goes bankrupt, which is highly unlikely due too many factors, that would leave only AMD and VIA as competitors in the marketplace."

I think he means Chapter 11, reorganization, not liquidation.

Either case, investors lose all. In fact, Chapter 11 usually strongly favors the management.

12:58 PM, October 04, 2006  
Blogger Scientia from AMDZone said...

I keep hearing how wonderful the C2D architecture is. As far as I can tell, C2D is good on the desktop but not so competitive in servers. The current problem is FBDIMM but Intel then has the problem of MCM with Clovertown. It wouldn't surprise me if either Anandtech or Tom's characterized an Intel architecure as the best. Both of these websites have lost their credibility. Neither has done adequate load testing on the dual core chips.

The big question though is where are the HPC designs for C2D derived processors? If Yorkfield is going to be that good why hasn't a Yorkfield based supercomputer been announced? AMD has three scheduled, two use K8L quad core. And, the Oak Ridge machine was just overhauled and upgraded from Opteron to dual core Opteron.

1:08 PM, October 04, 2006  
Blogger Scientia from AMDZone said...

AMD is definitely too secretive. They mostly allow others to publicize their products. It's funny to see the differences in approach towards marketing between AMD and Intel. Intel seems to be driven by sheer desperation with their marketing blitz which doesn't seem to be helping much for their stock price. AMD seems to be a lot more reserved, allowing others to praise them unlike Intel's constant self-praise lately.

I have an article that talks about his in my blog. Basically, I think that Intel is talking a lot and even about unlikely future products because they have no real foundation for their server platform. It appears that they are hoping that their lead on the desktop will be associated with everything else. Yet, Intel faces increasing pressure in notebooks, servers, corporate sales, and the Asian market.

In contrast, AMD has been incredibly secretive. I think AMD is quiet because although they are improving in most other areas, they are struggling on the desktop. I think they are trying to maximize their announcements because it wouldn't take much to flip the spotlight back to Intel. The desktop is most visible and Intel is currently ahead.

1:17 PM, October 04, 2006  
Anonymous Anonymous said...

The PhD pretender has slowed down... no posts in a while. Actually the post were getting tiring.

INTEL going BK
INTEL CPUs causing fires
Pat stock options underwater.

All are irrelevant, could the pretender actually be studing and aborbing the frigthening facts

Core2 actually outperforms AMD across the board

Careful analysis of AMD's profit/loss and new revenues factoring in 50% price drop looks frigthening

INTEL 4 core annouced, released, and benchmarked while K8L and AMD all they can do is make pretty powerpoint foils.

INTEL ramping 65n in 3 factories and pumping core2 out millions a week. AMD not one 65nm production CPU and volume not till Q1'07. Damm I'll wager INTEL could have samples of 45nm CPUs by then...


Looks like INTEL is on a role again and AMD is hunkering down under the bombardment. Of course they have their fallback. Hector will call out the lawyers. Damm I can't compete in the marketplace but I'll look to compete in the courtroom. Did you see those fanatic Oklahoma Schooner fans try and take Oregon Duck to courts? NO even them sorry assed narrow minded, blind, mid-westerns are proud and manly enough not to go so low as AMDers when they loose.

Why is AMD quiet? Because they got nothing.. no rabbit to pull out of no hat

Why is Sharikou so quiet? Because he has gone made from all his BK delusions

LOL

5:51 PM, October 04, 2006  
Blogger nyx said...

Something that perplexes me is: If Intel's Core 2 line has such a dominating presence as most seem to proclaim nowadays, why in the hell is ATI and Nvidia not supporting it? I realize AMD is in the process of aquiring ATI, so some might claim ATI will no longer support Intel, but I thought ATI mentioned it still would with an XPRESS 3200 chipset. Nvidia doesn't seem to even acknowledge Intel's existence in the gaming market. I have not found any motherboards for the Core 2 line which has a nForce 590 or XPRESS 3200 chipset. Both chipsets have been out for awhile now, but none of the online stores I frequent (such as Newegg) have any Core 2 supported boards with either or those chipsets. With AMD you have quite a nice selection, yet with Intel you are forced to get a 975 chipset based board if you want the latest chipset. I thought at least Nvidia would be quick to support Intel since they are supposedly playing both fields to maximize profits. Could that be one of the many reasons why some speculate that Intel could be moving soon to aquire Nvidia? In my opinion it won't happen. I agree with most people that Nvidia would cost Intel far too much, though Intel could afford it. The aquisition would probably not fly anyway for legal reasons too. Nvidia's stock price sky rocketed due to the speculation, but it has to fall as soon as investors realize Intel's not going to dish out $ 10 billion in the midst of firing personnel and selling off parts of itself. Intel's graphic chipsets are not going to cut it and their northbridge chipsets are not as attractive as the top line ATI or Nvidia chipsets. They need Nvidia now more than ever.

"Imagination chief exec Hossein Yassaie explained that the companies will work together to put graphics and video capabilities into Intel microprocessors."

http://www.theinquirer.net/default.aspx?article=34822

-> that's one of the reasons why AMD aquired ATI right? Is Intel trying to combine both fields in anticipation of what AMD will be doing with its future line of processors? It's hard to believe Imagination Technologies can beat out ATI in making this concept possible.

6:34 PM, October 04, 2006  
Blogger core2dude said...


A recent Custom PC review of AM2 3800 X2 vs C2D 6300 found the C2D had about 10% overall performance advantage. However the AM2 was £100 while the C2D was £130. They concluded the C2D was "clearly the better option" ??

Another one in Intel's pocket...


The reason is that, even though the processor costs $140 to $180, the entire computer costs about $800. At that point, the price difference between the two systems is just 5%, and you are getting 10% performance advantage.

7:49 PM, October 04, 2006  
Anonymous Anonymous said...

anonymous said:
"If you understand how wafers are made, then you'll know that it's slowly pulled from a silicon ingot... if you can figure out a way to make square wafers, you got yourself a nice patent. ;) "

If the Japanese farmers can grow square water-mellons, it's possible to make square wafers. I think it's harder to make circular-dies. Haha...

You see, circular-dies have better average distance from pins to circuitries than square-dies!! I you figure out a way to make circular-dies...never mind my ranting...haha...

-Longan-

10:51 PM, October 04, 2006  
Anonymous Anonymous said...

"Four-core vs. 4x4"

I will ask two trivial questions:

1. Which one is more costly to design? The socket for 4x4? Or the chip-package with 2 glued-on dies?

2. Which way is easier, less problem? A socket for 4x4 or the chip-package for 2 glued-on dies?

I bet on the 4x4 solution any day.

-Longan-

The 4x4 should also give a free path to 4x8 in the future. Try that with the glued-on approach, two-die quad. Pffft!

11:00 PM, October 04, 2006  
Blogger S said...

If Intel's Core 2 line has such a dominating presence as most seem to proclaim nowadays, why in the hell is ATI and Nvidia not supporting it?

The reason why ATI is not supporting Core 2 should be pretty clear - Intel would not want to use a competitor's chips and AMD would not want to support the ramp of a competitor's chips.

Why Nvidia don't yet have chipsets for Core2 ? That's because they were so heavily into AMD chipsets when AMD-ATI announcement caught them by surprise. They haven't had enough time to develop a chipset optimized for Core2 yet. Nvidia do have some chipsets for Core2 but they are more like adapted from AMD chipsets than developed for Core2. So I guess in time Nvidia will have stronger products targetted for Core2 platform. Otherwise they will lose the opportunity to capture the market segment that ATI so recklessly gave up.

What Core2 is doing to the CPU market will become clear in a couple of weeks, once Q3 results are out.

4:12 AM, October 05, 2006  
Blogger S said...

I bet on the 4x4 solution any day.

Common sense tells me that a PC with two socket motherboard will have a bigger bill of material and running cost than one with a single socket motherboard and TCO would be way higher.

I wonder what way a two socket quad core is better than one socket quad core, even architecturally.

4:25 AM, October 05, 2006  
Blogger Kalle said...

"I will ask two trivial questions:

1. Which one is more costly to design? The socket for 4x4? Or the chip-package with 2 glued-on dies?

2. Which way is easier, less problem? A socket for 4x4 or the chip-package for 2 glued-on dies?

I bet on the 4x4 solution any day.


Do you mean what is easier, two AM2 sockets or one 775 socket with support for dual-die CPU's? If so then I say the latter is clearly easier since you don't have to worry about any kind of NUMA issues.

5:30 AM, October 05, 2006  
Anonymous Anonymous said...

"Four-core vs. 4x4"

1. Which one is more costly to design? The socket for 4x4? Or the chip-package with 2 glued-on dies?

2. Which way is easier, less problem? A socket for 4x4 or the chip-package for 2 glued-on dies?

>> 4x4 uses an existing socket (S1207 for Opterons, much like how first gen FXs used Socket940 for Opterons), uses existing chipsets (all based on same HyperTransport technology.

>> Incremental cost is minimal, IMHO

7:23 AM, October 05, 2006  
Anonymous Anonymous said...

If the design of the 4x4 were so easy, AMD should have had this available well before Conroe was released. Fact is that the "glued" dual-core quad is easier to implement....the advantages to yield are also obvious to this approach.

Conceptually it (4x4) should be easy.....either AMD engineers are morons or there is a problem that AMD has not disclosed.

My bet is that there is a problem. Lately it seems like every Intel roadmap has been accelerated and AMD is just stagnating and screwing up on all fronts.....I hope they have something for 2007 with K8L/4x4 and do a better of job of bringing it to market than they have done recently or it is AMD and not Intel who will be BK in near future.

If AMD has another couple quarters like this, they will be behind in every single market segment (where they currently are ahead with certain Opteron configurations which is making AMD profitable) and seriously behind on Desktop and light years behind on mobile platform. Intel has a ton of headroom with Conroe and as soon as AMD releases something, Intel could easily crank out another 5-6 speed grades the next day with C2D....AMD better come up with something soon or it is game over. Merom isn't even mainstream yet.....Intel doesn't need it to take 90+% of market.

2007 could kill AMD.

9:35 AM, October 05, 2006  
Anonymous Anonymous said...

Have any of you intel fanboys run conroes with your own visual basic code? I run linux and write gambas(visual basic for linux) code. So before you get too excited about 2 conroes glued together you better check conroes performance in basic code execution for accuracy.
Although intel may say that they will have quad cores to test, if the product is not right what is it worth. Quad wrong is probably worse than quad none.
However, I have not found any problems with AM2 chips and since you can buy a AM2 5000 and put it on a $50 motherboard with ddr2 800 why bother with the expensive conroe e6600 with a $250 mother board and maybe run only ddr2 800. You just cant beat the amd power price performance curve.
Its too bad intel was not able to provide fast cheap motherboards like amd has.
I suspect that the intel quad core motherboards will be some more ancient hammer mill technology we have come to expect from intel of late.
Shipping quads is not the same as selling quads....inventory is not sales.
Intel came very late to the 64 bit multi core processor game so maybe they will make it or maybe they wont.
Intel glued together 2 netburst chips for the 800 series which still is not selling, so why would you glue together 2 conroes when glueing gave recent past sales failures?

11:25 AM, October 05, 2006  
Anonymous enumae said...

Longan said...

"1. Which one is more costly to design, the socket for 4x4, or the chip-package with 2 glued-on dies?"

Unless your work for either Intel or AMD neither you or I will know.

But more importantly you should compare ...

1. The cost of creating 2 seperate processors packages (4x4) vs 1 processor package (Kentsfield).

More materials more money.

2. The cost of moterboards.

Kentsfield with LGA 775 (Newegg $270 is the most expensive), or 4x4 with LGA 1207 (Newegg $275 is the least expensive Socket F motherboard).

The cheapest motherboard that supports Kentsfield is...

Asus P5B $150.

"2. Which way is easier, less problem, a socket for 4x4 or the chip-package for 2 glued-on dies?"

I would have to relate back to a previous staement, 2 processors (4x4) vs 1 (Kentsfield), your only validating 1 chip vs 2, meaning less problems and easier.

"Try that with the glued-on approach, two-die quad. Pffft!"

The thing I feel you are overlooking is these are not Netburst cores.

12:37 PM, October 05, 2006  
Anonymous Anonymous said...

Checking performance for accuracy???
...
Why would anyone buy a $50 motherboard;)
http://www.newegg.com/Product/ProductList.asp?N=2000200280+1070622728&Submit=ENE&SubCategory=280
If you are that cheap though, there are Core 2 motherboards below $50. And with AMD, you're required to use DDR2 800 or risk a big decrease in performance, where as with Core 2, DDR2 800 is only needed if you overclock.
Yes AMD has a very nice lineup of single cores;)
And any board that supports X6800 should support QX6700 with a BIOS update or so.
And I'd rather build up stock to sell later, rather than potentionally miss out on sales that'll never come back.
64 bit? They came when 64 bit wasn't[and still isn't] needed.
PS, Pentium D 800s are single die, it's the 65nm Pentium D 900s[which are cooler and faster] that are dual die.

2:31 PM, October 05, 2006  
Anonymous enumae said...

Anonymous said...

"you can buy a AM2 5000 and put it on a $50 motherboard with ddr2 800 why bother with the expensive conroe e6600 with a $250 mother board and maybe run only ddr2 800."

You should have looked up the prices :)

AMD setup $566

AMD X2 5000+ $519

PC CHIPS A33G V1.0 Socket AM2 $47

Intel setup $465

Intel Core 2 Duo E6600 $317

ASUS P5B $148

Intel is $99 dollars cheaper :)

You also could have lowered the Intel price more, but this board will be able to support Kentsfield.

"You just cant beat the amd power price performance curve."

Looks like I did, and so could anyone who goes to Newegg.

"Its too bad intel was not able to provide fast cheap motherboards like amd has."

Like I said, the price could have been lowered, without support for Kentsfield, the price total would have been $412 ($319 + MSI P965 Neo-F $93)... Intel is now $154 cheaper.

Also keep in mind there is a premium price on these boards and chips right now, they will come down even more.

"I suspect that the intel quad core motherboards will be some more ancient hammer mill technology we have come to expect from intel of late."

Well... nevermind :)

2:39 PM, October 05, 2006  
Anonymous Anonymous said...

I have both AM2s and Conroes, the conroes are just too expensive to run ddr2 800. Only the expensive conroe boards support ddr2 800 but all AM@ boards regardless of price support ddr2 800.
I have tried the $50 conroe boards and they are really bad and only support ddr2 533 while all AM2 boards support ddr2 800 and even the cheap boards run circles around the conroe boards.
You get much more performance per dollar with AM2 than you do with conroe.
My AM2 5000 on a $55 board beats the pants off of my Conroe e6600 on a $75 board and I am not going to spend another $200 to $350 for another conroe board just to prop up the lack luster conroe performance.
Like i said, there are no cheap fast boards for conroe only AMD AM2 has cheap fast motherboards.
You cant build a walmart computer with over priced conroe parts but you can build a high performance computer with AM2 amd parts cheaply and sell it to the typical end users who wont hold still for abusive prices.
If more of you would use linux you could see where Conroes cpu power is going.
Just say no to quad conroes, until intel designs a real 4 core chip with integrated memory controller skip the conroe quad.

4:06 PM, October 05, 2006  
Blogger Scientia from AMDZone said...

Kentsfield with LGA 775 (Newegg $270 is the most expensive), or 4x4 with LGA 1207 (Newegg $275 is the least expensive Socket F motherboard).

The cheapest motherboard that supports Kentsfield is...

Asus P5B $150.


Hmmmm. The problem is that you are comparing the wrong boards. Socket F (1207) is not equivalent to socket 775. You should have compared socket F to socket 771. There are more socket 771 and 775 boards because those standards have been out longer. However, when you do the proper comparison on NewEgg you get:

771 - $310 - $600
1207 - $275 - $445

Socket 775 would be equivalent to socket AM2.

775 - $130 - $350
AM2 - $200 - $305

6:56 PM, October 05, 2006  
Anonymous DoggieHowser said...

I believe there will still be a premium on 4x4 boards. So it wouldn't be cheap compared to entry level 965 based chips, but it would be comparable to Intel's 975X (extreme edition class boards).

But with 4x4 boards, you get quad core (2xFX) today, upgradeable to octa core (2 x quads) next year.

And you get Quad SLI support. Something you don't get on Intel Core 2 Duo/Quads yet.

As for support for Kentsfield, well... I am not sure how many Intel fanboys who bought the earlier versions of 975X (ie before BatX aka Rev 304) would agree with you. If you bought those versions for yr Pentium D Extreme Editions, you would have been disappointed to find out later that it DIDN'T support Core 2 Duos. Tough luck.

7:29 PM, October 05, 2006  
Anonymous enumae said...

Scientia from AMDZone said...

"The problem is that you are comparing the wrong boards."

Actually the reason for comparing the socket F was because 4x4 will use a socket F, I also believe that 4x4 boards will be at least $250, so the $270 is not far off.

That is speculation based on the fact that 4x4 is going to be a nitche market.

"You should have compared socket F to socket 771. There are more socket 771 and 775 boards because those standards have been out longer."

Tha problem with that is Kentsfield is LGA 775, not LGA 771 and we were comparing Kentsfield to AMD 4x4.

9:18 PM, October 05, 2006  
Anonymous Anonymous said...

Kalle said:
"If so then I say the latter is clearly easier since you don't have to worry about any kind of NUMA issues."

Good point. NUMA issues would be a lot difficulties for desk-top apps.

-Longan-

10:28 PM, October 05, 2006  
Anonymous Edward said...

"I believe there will still be a premium on 4x4 boards. So it wouldn't be cheap compared to entry level 965 based chips, but it would be comparable to Intel's 975X (extreme edition class boards).

But with 4x4 boards, you get quad core (2xFX) today, upgradeable to octa core (2 x quads) next year.

And you get Quad SLI support. Something you don't get on Intel Core 2 Duo/Quads yet.

As for support for Kentsfield, well... I am not sure how many Intel fanboys who bought the earlier versions of 975X (ie before BatX aka Rev 304) would agree with you. If you bought those versions for yr Pentium D Extreme Editions, you would have been disappointed to find out later that it DIDN'T support Core 2 Duos. Tough luck."


Totally agree.

It is a moot point to discuss any upgrade path for Kentsfiend, anyway, since Intel's current FSB/northbridge approach doesn't support octo-core very well. Even at 1.6GHz bus speed each core has only 200MHz share, not to mention contention overhead. In contrast, 4x4 scales better, and it will scale.

Also, while NUMA programming will be more difficult than UMA, the increased difficulty is not nearly as much as that of multi-threading itself.

10:35 AM, October 06, 2006  
Anonymous Anonymous said...

Your AM2 5000 price is way too high, mine cost $349 and is now $369 your $519 price is just hogwash like the performance of a e6600.
So this am2 5000=$349 and the ECS ATI-RS485M motherboard special of $44.00 plus 1gb of ddr2 800 ($132) so the board and chip are $393...its cheap and super fast and a super deal and runs memory bandwidth better than conroe and runs math programs better than conroe, and can actually run a basic program with out screw ups.
So I really dont see how a muffler clamped conroe core2 sort of quad chip will fix the problems intel has already, if they cant make 2 cores work right how do you expect 4 of the same problematic cores to work right?
If intel is already drowning in core2 conroe inventories what would be the sense to double their troubles with quads?
There are no good, cheap fast, core2 motherboards so there are no economical intel based computers, dual, quad , or otherwise.
A pentium 3 is not the processor of the future, its only a historic footnote regardless if you put a super charger on it or not, its still yesterdays old news core2 antiques.
Regardless of how many supercharged pentium 3s you put under one cover, its still yesterdays old technology which cant operate present or future computers correctly.

11:04 AM, October 06, 2006  
Blogger core2dude said...


Have any of you intel fanboys run conroes with your own visual basic code?

Visual Basic??? Isn't that for babies???

3:27 PM, October 06, 2006  
Anonymous enumae said...

Anonymous said...

"Your AM2 5000 price is way too high, mine cost $349 and is now $369 your $519 price is just hogwash like the performance of a e6600."

It would be hogwash if I did not have a link, prices change, availabilty changes, maybe AMD makes very few of them now that Intel has released Conroe, I don't know, but thats the price.

"So I really dont see how a muffler clamped conroe core2 sort of quad chip will fix the problems intel has already, if they cant make 2 cores work right how do you expect 4 of the same problematic cores to work right?"

Well I do not do codeing, and if you are having such a problem, write Intel a letter and explain how there chip is screwing up.

"If intel is already drowning in core2 conroe inventories what would be the sense to double their troubles with quads?"

I,and many others here would doubt that Intel has inventories of Conroe, again write a letter... post there response here, I am curious.

"There are no good, cheap fast, core2 motherboards so there are no economical intel based computers, dual, quad , or otherwise."

Looking at the price I gave you it would seem you are wrong.

Is there something wrong with an ASUS P5B motherboard? Please explain.

"A pentium 3 is not the processor of the future, its only a historic footnote regardless if you put a super charger on it or not, its still yesterdays old news core2 antiques."

What is K8L then, from what I understand it is basically K8 with alot of enhancements (superchargers :)), the new core (K10?) is not due until 2008.

"Regardless of how many supercharged pentium 3s you put under one cover, its still yesterdays old technology which cant operate present or future computers correctly."

Again write the letter and post the reply here.

9:36 PM, October 06, 2006  
Anonymous Anonymous said...

Whats wrong with the p5b asus, 1st its a 965 chipset which in reviews has been rated rotten to horrible in various performance tests.
2nd its price is beyond $50-$60 dollars which makes it useless to the white box manufacturers(too expensive, no profit margin).
3rd it only has experimental 64 bit drivers for windows only you know, the gamebox os.
The real os software Linux, Solaris, and Unix have little to no support from intel so this chip set will never be popular and US Govt tests also found problems with this chipset as well as conroes.
The same linux gambas code was tested with smithfields, preslers, conroes, X2 amd, AM2 amds and only the conroes failed to run the same code correctly.
Intel has no process for problem reporting they just stick their heads in the sand and hope for the best.
Further, intel said core 2 was ready, but I have had problems, why?
AMD said AM2 was ready, and there were no problems.
Intel motherboard chipsets are very expensive compared too ATI-AMD, VIA, Nvidia-Uli and of course SIS.
Here is a nice ATI-Uli motherboard for the conroe, 775Twins-HDTV R2.0 asrock, but intel wont let asrock market it in north America.
Same question, why would you want a quad core conroe when the dual core conroe is problematic and too expensive to build a user price friendly system?

1:44 AM, October 07, 2006  
Anonymous Anonymous said...

All those zippy core2 bench marks were done on 975 chipset boards nobody used the 965 chipset boards.
The only reviews of the 965 chipset boards revealed they are not as fast as the 975 chipset boards so why build it so cheaply?
You need those 975 bad ax boards with 1066 ddr2 at a cost of $200-$350 to get that old conroe core 2 running at high speed.
Maybe those quad cores will run on a board with two 975 chipsets?
Intels benchmark results are usually dependent on external hardware that cant be purchased so nobody can contest their results that can only be duplicated with engineering samples that are not for sale on this planet.
So regardless of the Kentfield benchmark result reports its only on equipment which will never see daylight in regular production.
So rave about there quad core benchmarks all you want you will never see the hardware they used to get those benchmarks and you will never be able to repeat there results.

9:35 AM, October 07, 2006  
Blogger S said...

"why would you want a quad core conroe when the dual core conroe is problematic and too expensive to build a user price friendly system?"

Good systems are inevitably more costly to build. People who want better performance do have to pay more than what a budget user will pay.

Intel has a range which competes in price & performance with AMD - the P4s. AMD doesn't have a product which matches Core 2. So don't expect it to be in cheaper Systems.

10:55 AM, October 07, 2006  
Anonymous Anonymous said...

P4s comparable to Athlon 64s? In tests I have Athlon 64s run circles around Pentium Ds, even before we get into the heat/thermals and power consumption.

Geez, what alternate universe do you live in?

6:33 PM, October 08, 2006  

Post a Comment

Links to this post:

Create a Link

<< Home