Sunday, November 26, 2006

AMD eating Intel's notebook market share

See past "transcripts" for previous conversations, especially, this one -- which I wrote one year ago.

Intel BK by 2Q08.

44 Comments:

Anonymous Anonymous said...

Dude,

Are you running out of argument?

1:02 PM, November 26, 2006  
Anonymous Anonymous said...

http://www.reghardware.co.uk/2006/11/23/amd_faces_patent_infringement_claim/

Seems to me that AMD technology appears to be stolen from somebody else.. It is not INTEL only, is it Doc?

1:22 PM, November 26, 2006  
Anonymous Anonymous said...

LOL now your news items are stuff that you wrote last year :-)

Talk about a reliable source.

3:37 PM, November 26, 2006  
Anonymous Anonymous said...

You are dusting off some old articles vs. writing about the huge flop about to happen -- AMD "4x4"???

Somehow you have avoided talking about AMD's dumbed-up 2P Opteron workstation for "gamer$". I wonder why??

Could it be that the economics of having to run two chips vs. one quad-core Intel chip just don't make sense?

Or could "4x4" be another sign that AMD is the market leader in processor vaporware? AMD cannot ship on time something that has been working a long time (2P Opteron systems). It's sad.

And then there is KL8 -- "K-Late". AMD has been working on this chip for nearly 10 years. It had better be very very good. This chip is the only new thing AMD has done (or 'will have done') in a long time. The rest of AMDs chips are just Opteron warmovers and DDR2 hacks.

Really, AMD is competing on price with Intel. But this is because AMD tweaks their one good design, doing next to no R&D. It really is a sad story.

7:42 PM, November 26, 2006  
Anonymous Anonymous said...

Are those transcripts real?

Or its just fictition?

They look real, nothing strange in their conversation, but how was that cought if they are really real?

5:55 AM, November 27, 2006  
Anonymous Anonymous said...

“Seems to me that AMD technology appears to be stolen from somebody else.. It is not INTEL only, is it Doc?”

It seems nVidia paid out 11 million to Opti for the same thing. The article claims AMD was approached 4 years ago about the infringement. It smells like Opti hasn’t been very aggressive on purpose.

The cache snooping technology sounds like a steal, maybe AMD should just buy the company if they’re guilty.

6:51 AM, November 27, 2006  
Anonymous Anonymous said...

“Are you running out of argument?”

Slow news week. David killing Goliath isn’t news anymore; however I predict that Paul won’t even be able to lick stamps for the post office after the Xmas numbers come in.

I haven’t seen much Intel support in the resent posts. Intel’s constant decline and lose of momentum must be having a devastating affect on the faithful.

No news is good news for Intel fanboys lately.

7:27 AM, November 27, 2006  
Anonymous Anonymous said...

From C/net news from 11/21/06 here is some boring news


“In x86 servers, AMD saw server revenue grow nearly 80 percent.”

“Intel lost share on a year-to-year basis”

8:16 AM, November 27, 2006  
Anonymous Anonymous said...

STILL 6 quarters from BK???!!? When does that clock start ticking???

9:19 AM, November 27, 2006  
Anonymous KingRichard said...

I like these news:

AMD's 65nm dual-core Athlons to debut 5 December?

http://www.reghardware.co.uk/2006/11/27/amd_65nm_x2_debut_5_december/

2:59 PM, November 27, 2006  
Blogger Markus said...

And then there is KL8 -- "K-Late". AMD has been working on this chip for nearly 10 years.

You're a fucking idiot.

3:37 PM, November 27, 2006  
Anonymous Anonymous said...

"STILL 6 quarters from BK???!!? When does that clock start ticking???"

Come on, remember it's 5-7 quarters away - you weren't aware this, like all his other dumb predictions, was a rolling one? If you go back to the real early predictions BK was as early as Q2/Q3'07 - of course AMD wouldn't have been able to handle the market capacity by then but Shari-kook had his own calculations to prove it!

3:55 PM, November 27, 2006  
Anonymous Anonymous said...

Nice Intel ads on the top of your page Skairkou - but aren't you help delaying the inevitable BK by doing this...or are you just interested in ad money, not principles?

3:57 PM, November 27, 2006  
Anonymous Anonymous said...

Dr. Sharikou

About the BK thing. I also believe Intel will soon be in deep bandini, however I’m perplex as to why Intel will soon build another plant in Vietnam. Either they are extremely stupid and ignorant of their pending demise or incredibly wise and the new plant is their ace in the hole (extreme tax relief).

After K8L’s release the rules change again, how will Intel clone the new instructions and extension sets? If x86 is the standard, Intel will never be able to dig them selves out after K8L; or am I wrong?

In the past I figured it would be the manufacturing skills that come with K8L that would be the beginning of the end for chip-zilla.

You are holding back something or you wouldn’t be so sure of the BK. I’m getting close, huh.

6:12 PM, November 27, 2006  
Anonymous Anonymous said...

Wow, Athlon 64 X2 65nm next week, Dec 5th, I wonder if Santa knows how bad I want that.

7:20 PM, November 27, 2006  
Anonymous Anonymous said...

You're weird. Who talks to themselves?

7:35 PM, November 27, 2006  
Anonymous Anonymous said...

Someone better tell Intel they're going bankrupt....they are already sampling 45nm processors.

7:59 PM, November 27, 2006  
Anonymous Anonymous said...

Must be really slow on the news hook, you missed this story on the Register

amd to announce 65NM by Dec 5th?

http://www.reghardware.co.uk/2006/11/27/amd_65nm_x2_debut_5_december/

8:04 PM, November 27, 2006  
Anonymous Anonymous said...

“You are dusting off some old articles………vaporware……."K-Late"…….DDR2 hacks……….It really is a sad story”

Excuse me friend; I liked your post, but you need to keep up with current events. In 7 days Athlon 64 X2 65nm launches ahead of schedule big time. K8L is alive and early bench tests put it 64% over Xeon MP and is also on schedule. The rest of your fluff isn’t worth debate.

You are correct on the dusting off old articles.

8:18 PM, November 27, 2006  
Anonymous Anonymous said...

“STILL 6 quarters from BK???!!? When does that clock start ticking???”


Each tic takes three months. This quarter ends December 31st then listen for another tic. Trust me; you’ll know when time is up. Here’s your sign; when Paul applies for a job as a salesman at a used car lot. LOL

8:46 PM, November 27, 2006  
Blogger Sharikou, Ph. D said...

You are correct on the dusting off old articles.

There isn't much news any way.

8:50 PM, November 27, 2006  
Anonymous Anonymous said...

“Dude,
Are you running out of argument?”


Give the Doc a break; he’s most likely getting a life during the holidays. It’s the time of year to be cheerful, dude.

Yeh, Doc are you running out of argument, or what?

8:59 PM, November 27, 2006  
Blogger Mikolaj said...

Georg Soros bought today AMD shares for 20 million. There must be a reason for that.

10:43 PM, November 27, 2006  
Anonymous Anonymous said...

I don't see how AMD's new K8L will be revolutionary. Except for the opterons they look pretty sexy with a 128-bit wide SSE and floating point units (ars-technica) so now they match the xeon, however I'm not sure if AMD can catch up. However much people want to say that all the C2D is, is a reinvented pentium 3 with more cache, it really hasn't retained enough of the same characteristics. They have introduced a lot of features, and even though they have a large cache it doesn't matter when its faster, because thats all that matters. I think AMD's best chance is the opteron, this is the first time that they don't have the 2p crown in a long time, at least in the mainstream. Since the opteron is notoriously strong in 2p because of htt I think they might be able to take it back. But that is the only one that I am giving more than a 50 percent chance of catchup, everything else is just a coin-flip, we will only see when it comes out.

11:32 PM, November 27, 2006  
Anonymous Anonymous said...

"K8L is alive and early bench tests put it 64% over Xeon MP and is also on schedule"

Interesting, only 64% faster than Xeon MP. Do you know that curreny MP's are P4 based?

Also, could you link to these benches?

11:34 PM, November 27, 2006  
Anonymous Anonymous said...

I think its only natural that AMD is gaining portable market share.
After the Intel/sony flaming notebooks people will naturally prefer a non intel/sony brand.
Of course AMDs best performance and best price and best overall deal dont hurt those sales either.
AMD does have the best value and the safest track record so its only natural that AMD portables now out sell all others combined.
AMD best processors, best platforms, best deals.

12:44 AM, November 28, 2006  
Anonymous Anonymous said...

http://www.dailytech.com/article.aspx?newsid=5108

"Penryn will also be the last product based on Intel’s Core architecture. "

I don’t understand, so Core architecture was just a bandaged product? A Pentium 3 like people say?

I thought Core was the new Intel architecture.

1:53 AM, November 28, 2006  
Anonymous Anonymous said...


Excuse me friend; I liked your post, but you need to keep up with current events. In 7 days Athlon 64 X2 65nm launches ahead of schedule big time. K8L is alive and early bench tests put it 64% over Xeon MP and is also on schedule. The rest of your fluff isn’t worth debate.


Really sad that AMD has not even started selling 65nm before Intel have prototypes of 45nm out.
http://yahoo.reuters.com/news/articlehybrid.aspx?storyID=urn:newsml:reuters.com:20061127:MTFH94130_2006-11-27_22-55-33_N27466640&type=comktNews&rpc=44

4:20 AM, November 28, 2006  
Anonymous Anonymous said...

Someone better tell Intel they're going bankrupt....they are already sampling 45nm processors.

Yes. What a jerks. Some one must tell them that.

Or maybe they are going bankrupt because of that.
The price to pay for being first than everyone.

7:18 AM, November 28, 2006  
Anonymous Anonymous said...

Sales are what drives a companies success.
Introductions, samplings, and down sizing do not make for successful operations.
Intel conroe(reworked pentium 3s) have not sold well since introduction.
65nm quad core with bad design also not selling well.
So how will 45nm sampling help or just make further inventory build ups of more and more unwanted intel products.
Endless platform problems.
Millions of intel stock shares are being pumped and dumped daily.
AMD replacing intel market share is a daily happening.
How many bad decisions in a row does it take to sink a company? Intel may give us that answer quite soon.

9:18 AM, November 28, 2006  
Anonymous edward said...

"I don’t understand, so Core architecture was just a bandaged product? A Pentium 3 like people say?"

Core is Pentium-3 added with a lot of features. The basic structure remains the same, though. The largest change is probably the shared L2 cache. There are also better branch prediction, macro fusion, better memory prefetch, 64-bit, better power management, not to mention the smaller feature size.

These are just the changes that people have given a name an
used for marketing purposes. There are countless other smaller changes, just like any other "next generation" chip would have.

Looking at the picture from far away, Core 2 Duo is a modified Pentium-III more or less like Pentium-III is a modified Pentium-Pro (mostly with SIMD and integrated cache). The p6 architecture is a classic one - it's based on the best of the RISC core and OOO executions.

10:22 AM, November 28, 2006  
Anonymous Anonymous said...

Interesting, only 64% faster than Xeon MP. Do you know that curreny MP's are P4 based?

Also, could you link to these benches?

http://www.theinquirer.net/default.aspx?article=34581

This link just makes reference to the 64%, can’t find my original search, I’m still looking.

10:37 AM, November 28, 2006  
Anonymous Anonymous said...

“ Really sad that AMD has not even started selling 65nm before Intel have prototypes of 45nm out.”


Its not sad, its economics. Intel will need more than 45nm chips to keep up.

10:49 AM, November 28, 2006  
Anonymous Anonymous said...

“ Sales are what drives a companies success.
Introductions, samplings, and down sizing do not make for successful operations……////……How many bad decisions in a row does it take to sink a company? Intel may give us that answer quite soon”

Excellent post, logical and main points made clear. A+

10:56 AM, November 28, 2006  
Anonymous Anonymous said...

“ I don’t understand, so Core architecture was just a bandaged product? A Pentium 3 like people say?”


Wow, good deduction. I would of missed that.

11:01 AM, November 28, 2006  
Anonymous Anonymous said...

Anyone seen Amd 65nm, or 4X4 yet??? I thought that they were out already but can't seem to find them anywhere. I hope sharikou gets ES samples of K8L soon so that we can see the might y chip in action.

2:04 PM, November 28, 2006  
Anonymous Anonymous said...

I thought one of AMD's engineers gave a big presentation abt their native QuadCore Barcelona architecture that was covered in a lot of the major news sites. Guess not everyone read the news.

So what's new?

IMHO, it does look like AMD did look at overall system architecture efficiencies, and not just hacking on big slices of cache to the designs.

Core IPCs are being improved. SSE128/SSE4 and FP128 are the biggest changes that most people understand. These are similar to what Intel did from Core to Core 2.

But there are also other major changes. Core 2 uses more decoders + macro ops fusion to improve IPCs. Barcelona will move more instructions to run through the Fastpath decoder pipelines (which results in faster ops) rather than thru the simpler microcode engine. This should put it on par with Core 2 MacroOps.

Other changes include better branch prediction and out of order execution. Cache lines have also been widened and the cache hierarchy has also been tweaked to prevent thrashing for the shared cache. The L3 shared cache is a neat idea, IMHO. For apps that benefit from shared big caches, like databases, it can be turned on. But for a lot of HPC apps where it hurts performance, you can turn it off, but still have the dedicated L2 cache for performance gains. In Core 2, once you disable the shared L2, you only have the tiny dedicated L1 caches.

But what really interests me are the changes to the memory controller. Besides upping support for higher spec memory speeds, they are now independent DRAM controllers to support concurrency. This means that different cores can have access to the memory controllers at the same time. DRAM burst lengths are also improved along with improvements to pre-fetchers.

True, a lot of these are evolutionary changes rather than revolutionary, but IMHO, the Core 2 is also an evolutionary change over Core, which in itself was an evolutionary change from P3.

But IMHO, a "revolutionary" change ala NetBurst isn't always a good thing.

6:07 PM, November 28, 2006  
Anonymous Anonymous said...

"Penryn will also be the last product based on Intel’s Core architecture. "

I don’t understand, so Core architecture was just a bandaged product? A Pentium 3 like people say?

No. It will be the last Intel processor.
Intel is going bankrupt after that.
No more processors from Intel. ;)

4:38 AM, November 29, 2006  
Anonymous edward said...

"For apps that benefit from shared big caches, like databases, it can be turned on. But for a lot of HPC apps where it hurts performance, you can turn it off, but still have the dedicated L2 cache for performance gains."

Nice post. Do you have the link(s) to these information?

Can one actually enable/disable the L3 cache per-process on-the-fly? That'd be really interesting.

All look good on paper. We just need to wait to see how it competes with Intel's quadcore at that time (mid-2007).

11:23 AM, November 29, 2006  
Anonymous Anonymous said...

"Really sad that AMD has not even started selling 65nm before Intel have prototypes of 45nm out."

Prototypes are just that. A first silicon spin test chip with MANY bugs and does not mean that they are ready for manufacturing and mass production. Even their own press say 45nm won't be out till end of next year at the earliest.

AMD is and has been working on 45nm for some time now in colaboration with IBM.

AMD has been competitive with 90nm 300mm wafers. 65nm will only help accelerate their market share gains as you can squeeze more dies/wafer.

I see AMD as an innovative company that drives technology for their customers.

Intel? I had a networst P4 and all the benchies said they were the best thing since sliced bread. Push platforms, kill innovations, and commodotize the entire PC eco-system...

11:25 AM, November 29, 2006  
Anonymous Anonymous said...

If pictures are worth 1,000 words, then the language barrier shouldn't change the impact of these benchmark graphs. Impress has pitted AMD's new 4x4 (3ghz) setup against Intel's Core 2 Extreme QX6700 (2.67ghz) in a variety of tests, and the results are eye-opening.
All of their results are, um, interesting, but pay special attention to the very last graph on the page (Power Consumption).

http://pc.watch.impress.co.jp/docs/2006/1129/tawada92.htm

translated


http://translate.google.com/translate?u=http%3A%2F%2Fpc.watch.impress.co.jp%2Fdocs%2F2006%2F1129%2Ftawada92.htm&langpair=ja%7Cen&hl=en&ie=UTF-8&oe=UTF-8&prev=%2Flanguage_tools

2:43 PM, November 29, 2006  
Anonymous Anonymous said...

If pictures are worth 1,000 words, then the language barrier shouldn't change the impact of these benchmark graphs. Impress has pitted AMD's new 4x4 (3ghz) setup against Intel's Core 2 Extreme QX6700 (2.67ghz) in a variety of tests, and the results are eye-opening.
All of their results are, um, interesting, but pay special attention to the very last graph on the page (Power Consumption).

http://pc.watch.impress.co.jp/docs/2006/1129/tawada92.htm

translated

http://translate.google.com/translate?u=http%3A%2F%2Fpc.watch.impress.co.jp%2Fdocs%2F2006%2F1129%2Ftawada92.htm&langpair=ja%7Cen&hl=en&ie=UTF-8&oe=UTF-8&prev=%2Flanguage_tools

2:44 PM, November 29, 2006  
Blogger LeeCooper said...

If pictures are worth 1,000 words, then the language barrier shouldn't change the impact of these benchmark graphs. Impress has pitted AMD's new 4x4 (3ghz) setup against Intel's Core 2 Extreme QX6700 (2.67ghz) in a variety of tests, and the results are eye-opening.
All of their results are, um, interesting, but pay special attention to the very last graph on the page (Power Consumption).

http://pc.watch.impress.co.jp/docs/2006/1129/tawada92.htm

translated

http://translate.google.com/translate?u=http%3A%2F%2Fpc.watch.impress.co.jp%2Fdocs%2F2006%2F1129%2Ftawada92.htm&langpair=ja%7Cen&hl=en&ie=UTF-8&oe=UTF-8&prev=%2Flanguage_tools

2:47 PM, November 29, 2006  
Anonymous Anonymous said...

Sharikou did you predict this one?

http://www.nytimes.com/2006/12/02/technology/02chip.html?_r=1&oref=slogin

12:18 AM, December 08, 2006  

Post a Comment

Links to this post:

Create a Link

<< Home