Wednesday, January 31, 2007

K8L to frag Clovertown and Intel's next generation

AMD says K8L will frag Clovertown by 40% on a wide range of benchmarks. Some think AMD was talking about floating point performance, which is simply wrong.

As I have shown here, a dual core Opteron system outperforms a quad core Clovertown system by 14% on SpecFP_rate. That is, 4 Opteron cores are 14% faster than 8 Clovertown cores. Now, we know a K8L is 3.6x faster than a K8 on FP. This would lead to the conclusion that K8L is about 5x the speed of Clovertown, as far as FP performance is concerned.

So, the 40% performance lead has to be integer performance.

In conlusion, K8L will be 400% faster than Clovertown on FP, and 40% faster than Clovertown on integer.

Watch this AMD interview on K8L yourself. Basically, AMD is saying that K8L is so advanced that nothing in Intel's known roadmap can catch up. In other words, not only Clovertown and Conroe will be fragged, the next generation of Intel is pre-fragged, even with Intel going 45nm.

AMD will soon launch four K8L parts, the fastest will be clocked at 2.5GHZ, the slowest will be about 2GHZ. Therefore, within the AMD line, the performance spread is 25%. This means the slowest AMD chip will be at least 15% faster than the fastest Intel chip.

Intel BK by 2Q08, that what I said in 2006, and that is what will happen in 2Q08.

83 Comments:

Anonymous Anonymous said...

Whatever it was that you were smokin', seems like you just doubled it.

1:15 AM, February 02, 2007  
Anonymous Anonymous said...

Come on, where is the blog about the Q4'06 40% market share runate. Was it right like ALL of you other predictions?

You made that prediction so many times I would think you would update your blog on that!

1:52 AM, February 02, 2007  
Anonymous Anonymous said...

Sharikou, you fool ! You can't link your own blog as proof. They are 0% trustworthy.

2:06 AM, February 02, 2007  
Anonymous AthlonBoy said...

As pro-AMD as I am... you're stretching things a bit here.

4:37 AM, February 02, 2007  
Anonymous Anonymous said...

Hey, Sharikou, you're going to love this. THG and AMDZone are having a battle of words and a THG insider has called-out JumpingJack, a THG poster,as an Intel shill. Apparently JJ is paid by Intel to spread FUD against AMD!

7:10 AM, February 02, 2007  
Anonymous Anonymous said...

400% FP I believe (one trick pony), but 40% integer would be spectacular and I’m skeptical. We will all know soon enough. Algebraically speaking it all depends on what whole number is used and who’s number is it?

7:33 AM, February 02, 2007  
Blogger PENIX said...

This will be the nail in the coffin for Intel. 400% is more than Intel will have a year from the release of K8L. BK may come sooner than 2Q08.

8:14 AM, February 02, 2007  
Anonymous Anonymous said...

Intel's recent glory days are soon to be cut short Intel fanboys, enjoy it while you still can.

http://www.youtube.com/watch?v=G_n3wvsfq4Y

8:42 AM, February 02, 2007  
Anonymous Graham said...

Sharikou, you are delusional. Will you print a retraction when you are proven to be wrong or will there just be more excuses for why your predictions failed to be accurate?

11:38 AM, February 02, 2007  
Anonymous penix said...

i like your post in Rahul's VoodooPC blog.

Nice to see you've now changed camps.

12:45 PM, February 02, 2007  
Anonymous Anonymous said...

That sounds quite right to tell you the truth, K8L's arcitecture is amazing. AMD has something to be proud of because of the results they are talking so much about over.

http://www.hkepc.com/bbs/news.php?tid=718770&starttime=1164902400&endtime=1167580800

Everybody should just cool it and accept the truth now before they start commenting, hold you tongue because you might regret what you say later after all it is true K8L is faster, so what? Don't be such a baby about it.

God what losers getting so mad over how much a CPU blows anothers away. Thats just how technology is. Conroe isn't going to last long. What did you expect?

1:49 PM, February 02, 2007  
Anonymous Anonymous said...

Any update on the 40% market share runrate prediction - you seem to be dodging this...what was AMD's runrate exiting Q4'06

3:08 PM, February 02, 2007  
Anonymous Anonymous said...

Once again, Sharikou, your readers demand accountability (at least one of them). I would never ask you to shut your blog down, it generates too much food for thought. I will acknowledge your piercing insight and admit you were right if Intel BK's by or before 2Q08. But if Intel does not BK in 2Q08, will assess no blame and acknowledge your prediction was incorrect? You are quick to accept credit when your predictions are correct. Are you capable to doing the opposite?

3:45 PM, February 02, 2007  
Anonymous Anonymous said...

I have discovered the Intel crowd to be odious. Most likely Gen-X or younger. Intel must be proud.

4:09 PM, February 02, 2007  
Anonymous Anonymous said...

It appears AMD believes we are now in a Vista world and need new bench mark standards and if the new standards eliminate the old XP standards then Intel will loose its crown over night. I suppose AMD might be correct but just maybe a tad early. How about two standards; one for 32-bit and one for 64-bit Vista? The K8L runs 128 bit algorithms, what then?

4:40 PM, February 02, 2007  
Blogger PENIX said...

Anonymous said...

God what losers getting so mad over how much a CPU blows anothers away. Thats just how technology is. Conroe isn't going to last long. What did you expect?

Yes, there are some truly pathetic Intel fanboys that visit this forum. This forum encourages completely unbiased discussion, but the Intel fanboys cannot seem to grasp this basic concept.

5:24 PM, February 02, 2007  
Anonymous Anonymous said...

Err.. ah... oh yes. AMD has pre-fragged Intel's next architecture that's not even out yet No one in the general public even knows anything about it yet, other than the fact that it will feature CSI.

5:52 PM, February 02, 2007  
Blogger PENIX said...

Anonymous said...

It appears AMD believes we are now in a Vista world and need new bench mark standards and if the new standards eliminate the old XP standards then Intel will loose its crown over night. I suppose AMD might be correct but just maybe a tad early. How about two standards; one for 32-bit and one for 64-bit Vista? The K8L runs 128 bit algorithms, what then?

The 128-bit refers to width of the pipes, not int/fp precision. This is not the same 32-bit/64-bit used for Vista.

5:58 PM, February 02, 2007  
Blogger Jeach! said...

Can you believe this! DELL is getting sued by its investors... possibly direct proof for AMD in its anti-trust case:

The lawsuit accuses Dell of artificially inflating profits "by secretly receiving approximately $250 million a quarter in likely illegal rebate kickbacks payments" from Intel in return for an exclusive deal to purchase Intel's microprocessors [...]

Dell faces investor lawsuit over Intel pact

A new class-action lawsuit claims that computer maker Dell Inc. inflated profits with secret payments of about $1 billion a year from chip maker Intel Corp.

Dell Accused of Hiding Intel Payments

Also, look at what AMD lawyers are saying...

AMD was at one point trying to give away 1 million free chips to Hewlett-Packard to sell machines based on AMD CPUs, and HP only took 160,000 of them because of its fear of reprisals

AMD Sues Intel for Antitrust Violations

The more that is coming out of the discovery process, the easier it will be for AMD to prove its case. It's only the beginning folks!!

6:22 PM, February 02, 2007  
Blogger enumae said...

AMD - Windows, Red Hat, and SuSe

http://www.spec.org/cpu2000/results/res2006q4/cpu2000-20061030-07857.html
Operating System: Red Hat Enterprise Linux 4 Advanced Server Update 4 64-bit
SPECfp_rate2000 92.1

http://www.spec.org/cpu2000/results/res2006q4/cpu2000-20061120-08180.html
Operating System: SuSE Linux Enterprise Server 9 (x86_64) SP 3
SPECfp_rate2000 90.6

http://www.spec.org/cpu2000/results/res2006q4/cpu2000-20061121-08250.html
Operating System: SuSE Linux Enterprise Server 9 (x86_64) SP 3
SPECfp_rate2000 90.8

http://www.spec.org/cpu2000/results/res2006q4/cpu2000-20061120-08154.html
Operating System: SuSE Linux Enterprise Server 9 (x86_64) SP 3
SPECfp_rate2000 90.4

http://www.spec.org/cpu2000/results/res2006q4/cpu2000-20061120-08155.html
Operating System: Windows Server 2003 Enterprise SP1
SPECfp_rate2000 80.1

http://www.spec.org/cpu2000/results/res2006q4/cpu2000-20061113-08029.html
Operating System: SuSE Linux Enterprise Server 9 (x86_64) SP 3
SPECfp_rate2000 90.8

http://www.spec.org/cpu2000/results/res2006q4/cpu2000-20061113-08030.html
Operating System: Windows Server 2003 Enterprise SP1
SPECfp_rate2000 80.1

Average SPECfp_rate2000 = 87.8

AMD SPECfp_rate2000 on Solaris = 119

This would then conclude that Solaris is about 35% faster than Windows, Red Hat, and SuSe.

Intel - Windows, Red Hat, and SuSe

http://www.spec.org/cpu2000/results/res2006q4/cpu2000-20061002-07438.html
Red Hat Enterprise Linux 4 Advanced Server Update 3 EM64T
SPECfp_rate2000 81.0

http://www.spec.org/cpu2000/results/res2006q4/cpu2000-20061110-07871.html
Windows Server 2003 Enterprise Edition (32 bits), Service Pack1
SPECfp_rate2000 78.9

http://www.spec.org/cpu2000/results/res2006q4/cpu2000-20061030-07804.html
Windows Server 2003 Enterprise Edition (32 bits), Service Pack
SPECfp_rate2000 78.4

http://www.spec.org/cpu2000/results/res2006q4/cpu2000-20060915-07272.html
Windows XP, 64 bit Edition
SPECfp_rate2000 80.6

http://www.spec.org/cpu2000/results/res2006q4/cpu2000-20061016-07630.html
RedHat Enterprise Linux 4.0 Advanced Server for AMD64/EM64T, Update 3
SPECfp_rate2000 81.2

http://www.spec.org/cpu2000/results/res2006q4/cpu2000-20061113-08060.html
RedHat Enterprise Linux 4.0 Advanced Server for AMD64/EM64T, Update 4
SPECfp_rate2000 87.4

http://www.spec.org/cpu2000/results/res2006q4/cpu2000-20061113-08061.html
Windows Server 2003 Enterprise SP1
SPECfp_rate2000 83.0

Average SPECfp_rate2000 = 81.5

Now we have an average of 87.8 for AMD, and an average of 81.5 for Intel, this makes AMD about 7% faster while not using Solaris 10, and shows that without Solaris, AMD's is ahead, but not by 19%.

8:26 PM, February 02, 2007  
Anonymous Anonymous said...

it amazes me how amd fan boys accept AMD claims of Barcelona performance hook line and sinker WITHOUT proof yet even when you see intel benchmarks run by 3rd party websites they accuse it of being rigged

10:37 PM, February 02, 2007  
Anonymous Anonymous said...

it amazes me how amd fan boys accept AMD claims of Barcelona performance hook line and sinker WITHOUT proof yet even when you see intel benchmarks run by 3rd party websites they accuse it of being rigged

Of course. That's standard procedure here. Sharikou will accept Randy Allen's word that Barcelona will kill Clovertown. But, last year, when Intel said Conroe will own an Athlon 64 and provided benchmarks proving so he said the benchmarks were obviously faked. Oh the irony of that...

1:34 AM, February 03, 2007  
Anonymous Anonymous said...

About the spec score, the AMD - Windows was taken with 32 bits ICC, you should drop them.

3:46 AM, February 03, 2007  
Blogger Jeach! said...

it amazes me how amd fan boys accept AMD claims of Barcelona performance hook line and sinker WITHOUT proof

Generally speaking, you are right about accepting claims without proof. BUT, what is different here are three things:

1. In contrary to Intel - which has a history of doing paper launches and spreading FUD - AMD is quite conservative and honest. This allows us to take AMD's words seriously.

2. AMD has been making similar statements about the quad while holding the performance crown. These are not just new statements because Intel outperforms AMD on a per-core/per-clock basis. AMD predicted this way, way back!

3. Most of the architectural enhancements put into C2D to make it outperform its predecessor by such a large margin will be put in the K8L. So this allows us to 'assume' with great certainty that if it boosted the C2D by such a large factor, there is no reason not to believe it won't do the same for AMD.

yet even when you see intel benchmarks run by 3rd party websites they accuse it of being rigged

But the ARE rigged! Look at the facts:

Disclaimer: I'm not an expert at benchmarks, so I'm basing myself on Scientia's last blog.

1. Use of Intel's compiler rather than an independent compiler. This artificially boosts Intel's performance in comparison to real-world while artificially reduces AMD's performance in comparison to real-world.

2. We are now in a multi-core, multi-threaded world now, so why do benchmarks still test only on a single core (and for the most part, single threaded)? This boost's Intel's performance artificially because the other cores do NOT have to share the cache. This tends to hide Intel's weaknesses (FSB) and does not show AMD's streanght (HT, integrated memory controller, etc). I suggest that ALL benchmarks should create several hundred threads in the background (just like the real world) and allow them to operate on very light work.

3. Non-real world environment, such as limiting the data-set being operated on such that it perfectly fits in the cache. Billions of similar and simple instruction sets which enhances performance. No memory thrashing (Windows seems to be doing this frequently).

4. Memory speed timings helping Intel's performance, degrading AMD's performance.

5. Etc, etc.

5:38 AM, February 03, 2007  
Blogger enumae said...

Correction to my previous post.

I had not re-read prior to posting.

The comment about the 19% would not apply to the benchmarks I have referenced.

Sorry.

7:46 AM, February 03, 2007  
Anonymous ed said...

"it amazes me how amd fan boys accept AMD claims of Barcelona performance hook line and sinker WITHOUT proof yet even when you see intel benchmarks run by 3rd party websites they accuse it of being rigged"

That's true. There is no reason to take what AMD claims as fact. But everyone now also realizes that the first "benchmarks" of C2D run by the so-called "3rd party" are 1) button-pusher and 2) biased. C2D is faster on those benchmarks when compared fairly, but not that much (15-20% instead of 25%+).

There's a big different between a claim and a falsification.

8:12 AM, February 03, 2007  
Anonymous Anonymous said...

quoted from here:
----------------------------------
If people spend the next 2~4 months running around hyping Barcelona, above what may achieve, it has the potential to cause the same if not worse disapointment. If everyone gets a 40% improvement set in their heads, and AMD only delivers 20%, people are likely to blame AMD and say they failed. And the ridiculous thing is, theres absolutely nothing wrong with a 20% improvment. I have seen no offical staements from AMD yet. Ive only seen what others have seen, Some unverified specs, some that Uarch analysis you talked about in another thread, "...some guy who works for AMD said some guy at Dell told him....bla bla bla" and a whole lot of fan boys making a whole lot of claims. Did you know the lost soul Sharikou is claiming a 400% increase?Too many people making to many guesses with far to few facts to back them up.....for now.

AMD doesnt need 'help' like that right now. It looks like they are going to continue lowering their prices. Yes, that will cut into thier 'wallet', but they can deal with that. The benefit of regaining the value crown and continuing to expand market share or retaining it or, worst case slowing its errosion, is much better for them long term then a few more bucks right now. Disapointing people could threaten any progress they make doing this.

Nevertheless, "pre-fragged" sounds mighty cool.

8:44 AM, February 03, 2007  
Anonymous Anonymous said...

I thought 4x4 was supposed to frag Clovertown by 40%. What happened to that prediction?

10:41 AM, February 03, 2007  
Blogger Scientia from AMDZone said...

Sometimes I wonder why I bother countering overly optimistic projections of AMD. At least according to the Tommies on ForumZ I'm an AMD fanboy regardless of what I actually say.

At any rate, I'm sorry Sharikou, but you are misunderstanding the numbers.

Barcelona is 3.6X faster in SSE. Half of this increase is just the fact that we are talking about 4 cores versus 2 cores on X2. So, divide that by two. The actual increase in SSE per core is 1.8X. Now, since C2D is roughly twice the speed of K8 on SSE this really just brings Barcelona up to about equal to C2D in terms of SSE. So, don't expect Barcelona to clobber C2D on SSE benchmarks; they should be about the same. Well, unless the "testers" compile with the Intel Compiler which will not generate SSE code on AMD processors. I would like to see the PGI Compiler become the testing standard since it can generate good code for both.

Also, anyone who is expecting Barcelona to run 40% faster than Intel on Integer is going to be badly disappointed. AMD did a resurvey of K8. This is similar to when K7 was resurveyed and this gave K8 about 20% more speed. My conversative estimate for Barcelona is 10%. 15% would be good and 20% would be great. 40% is impossible. When AMD gives the 40% number this is performance / watt. This number probably includes cpu, northbridge, and FBDIMM wattage versus Opteron and registered DDR2. The actual performance of K8L versus C2D should be in the range of 5% slower to 5% faster at the same clock. However, Intel may have a faster clock.

12:01 PM, February 03, 2007  
Blogger Sharikou, Ph. D said...

Sometimes I wonder why I bother countering overly optimistic projections of AMD. At least according to the Tommies on ForumZ I'm an AMD fanboy regardless of what I actually say.


why don't you finishing watching the 9 minute video interview of Randy Allen first??

12:23 PM, February 03, 2007  
Anonymous Anonymous said...

Response to Jeach!
> Disclaimer: I'm not an expert at benchmarks, so I'm basing myself on Scientia's last blog.
You picked a guy who write a pro-AMD blog as your source?


1. Use of Intel's compiler ...
So that covers EVERY benchmark by every third party tester? Nice try.

2. We are now in a multi-core, multi-threaded world ...
I have seen MANY multi-core benchmarks for C2D. Don't kid yourself.

3. Non-real world environment...
The whole POINT of a benchmark is to simulate a real world test that is REPEATABLE across multiple products. They are designed to imitate the real world

4. Memory speed timings helping Intel's performance, degrading AMD's performance.
So all the third-party benchmarks are wrong?!?!

5. Etc, etc.
Give up. Intel is way ahead. There is no way you, or Scienta or anyone else is going to be able to cherrypick your way out of that FACT.

1:19 PM, February 03, 2007  
Anonymous Anonymous said...

Your evidence for all this BS is a video produced in a french restaurant in San Jose. The AMD rep didn't even claim that they had silicon to support any of these claims.....rather he referred to some VHDL model they have of the core.

AMD looks like it has about 2-3 more quarters before it goes burger king. $500 some million this past quarter, $200+M in Q1....can't run those Ferrari's on IOU's and VHDL models.

1:59 PM, February 03, 2007  
Anonymous Anonymous said...

Haha. That's hilarious. Scientia is a fanboy extraordinaire and even he is mocking Sharikou's ridiculous numbers.

"AMD chips to be 1 MILLION times faster in integer."

"AMD chips to be 1 GIGABAZILLION times faster in FP!"

"AMD chips will use less than 1 MEGAQUADRIZILLIONTH of a watt, while Intel chips will use 29.3 GOOGAQUINTILLION watts!"

Sharikou is like Dr. Freaking Evil, but Dr. Evil actually has a Ph.D in Evil, while Sharikou is clearly about 20 IQ points short of ever being worth of a Ph.D (Sharikou's IQ about 95-100).

2:04 PM, February 03, 2007  
Anonymous Dr.Who said...

Why AMD lacks a good compiler as the Intel one? Are AMD programmers less skilled than Intel ones? And yes, Barcelona may be faster per clock than Conroe & Bros. however the latter has a HUGE oc potential, Intel can simply raise the clock to fill the gap matching or surpassing the competitor once more time... who knows?

2:35 PM, February 03, 2007  
Anonymous Anonymous said...

WTF is "Frag" Oh lets see...
Frag is .. assassination of an unpopular member of one's own fighting unit.

How does K8L "assisinate" a clovertown, what a moronic word selection. You sure you got a PhD.

All I know INTEL can't be beat..

Lets see, on track for between 4-6 billion in profits on 34-36 Billion in sales, worlds largest semiconductor company. Demonstrated HiK metal gate in 2003, fully fuctional 45nm SRAM test chip in 2006 and fully functional 45nm x86 CPU in December. Building 3 factories to pump out 200+ million x86 dual/quadcore 45nm in 2008 at 50% margins. Got 3 65nm factories pumping out chipsets, graphics, and celeron 65nms. Damm AMD can't even get 65nm ramped and INTEL is pumping out Core2 overclocking Celerons on 65nm. AMD sucking losses with ATI and billions in debt. Got a crappy silicon partner who looked like a fool at IEDM annoucing their great SOI, dual stress, lowK, strained silicon 45nm technology with super immersion lithograhpy only to find out INTEL got HighK metal Gate on their process....

AMD got its ass whipped in the market, in the stock market, on the ledger, and lastly in the benchmarks. Gamers prefer COre2 tells you where the performance is.

The only place AMD has any advantage is the 4 core+ market. And what does AMD do, instead the try and slug it out with INTEL in the 2 core below market and sucking read for all of 2007 and BK in 2008.

What was the most funny when Hector was asked why AMD choose to leave the nich profitable market he responded after careful study they had to compete in the broader market. Why is that, they can't survivi on server along, even they know that. But what is obvious is they can't make anymoney in the mainstream.

AMD is and will always be a bit player... kind of like Sharikou. Make lots of noise, sue lots of people, pretend they are bigger then they really are, but like sharikou they are nothing...

2:49 PM, February 03, 2007  
Anonymous Anonymous said...

L1=4x128KB/L2=4x512KB/L3=2MB

1266 2.3GHz 95W
1268SE 2.4GHz 120W
1270SE 2.5GHz 120W
2258HE 1.9GHz 68W

There is also a 2.8 dual-core for Opteron 12xx series as well.

3:59 PM, February 03, 2007  
Blogger Scientia from AMDZone said...

sharikou
why don't you finishing watching the 9 minute video interview of Randy Allen first??


Yes, I've seen the interview. What in particular are you referring to?

4:01 PM, February 03, 2007  
Anonymous Anonymous said...

Scientia, I personally consider you an AMD fan...that said I really don't think that you should stoop yourself down to this guys level, he is a fanatic. I have read your blog several times and though you do lean towards Amd, you do at least man-up when you are challenged, unlike sharikoward who does drive-by posts. As far as a favorite company goes, I tend to do what I do in politics, stay in the middle, but when I want an op piece that leans Amd I read your blog and when I want something that leans towards Intel I read 180. When I want humor I come here.

4:40 PM, February 03, 2007  
Anonymous Anonymous said...

Watched it, too much talk with no real demo to backup!

You would call that "vaporware" sharikou!

5:58 PM, February 03, 2007  
Anonymous Anonymous said...

I would like to see two machines, one based on clovertown and the other based on Barcelona with bench marks running..

Taking Allen's and sharikou's word for it does not cut it..

5:59 PM, February 03, 2007  
Blogger Gabe_Tamara said...

"In contrary to Intel - which has a history of doing paper launches and spreading FUD - AMD is quite conservative and honest. This allows us to take AMD's words seriously."

dude go back to December analyst daya and AMD said and then to AMD earnings last month and tell me who is spreading FUD

And soeaking of paper launches, AMD supposedly launched 65 nm in December, how long did it take before we 65 nm parts ar readily available?

On and where is AMD's quad-core

6:48 PM, February 03, 2007  
Anonymous Anonymous said...

Somebody showed these amazing K8L results. But the link has been fried now... :/

Kuma X2(K8L):
344 difference between estimated PR
673 difference between ALU
720 difference between FPU
Per Model Number estimates divided by half

ALU 40% clamed by AMD
FPU 3.6X clamed by AMD in Quad Core
FPU devide by half for Dual Core

2.0Ghz X2 6100+ PR:10217
ALU:20224MIPS
FPU:21965MFLOPS

2.1Ghz X2 6200+ PR:10905
ALU:20944MIPS
FPU:22638MFLOPS

2.2Ghz X2 6300+ PR:11249
ALU:21617MIPS
FPU:23358MFLOPS

2.3Ghz X2 6400+ PR:11593
ALU:22290MIPS
FPU:24078MFLOPS

2.4Ghz X2 6500+ PR:11937
ALU:22963MIPS
FPU:24798MFLOPS

2.5Ghz X2 6600+ PR:12281
ALU:23636MIPS
FPU:25518MFLOPS

2.6Ghz X2 6700+ PR:12625
ALU:24309MIPS
FPU:26238MFLOPS

2.7Ghz X2 6800+ PR:12969
ALU:24982MIPS
FPU:26958MFLOPS

2.8Ghz X2 6900+ PR:13313
ALU:25655MIPS
FPU:27678MFLOPS

2.9Ghz X2 7000+ PR:13657
ALU:26328MIPS
FPU:28398MFLOPS

K8L fragz conroe. OH YEAH!!! :D

7:06 PM, February 03, 2007  
Blogger Azary Omega said...

to mr. Sharikou:
could you please do something about hundreds of ravaging pro intel kids that reply or simply post FUD on this blog? Seriously, this blog is deteriorating thanks to posts like this one:
Anonymous said...
Is this blog dead?


....i mean freedom of speech is important, but this is really isn't much of a blog anymore. Do something.

11:49 PM, February 03, 2007  
Anonymous Anonymous said...

Sharikou : Any update on the 40% market share runrate prediction?

11:51 PM, February 03, 2007  
Blogger Azary Omega said...

Anonymous said...
Sharikou : Any update on the 40% market share runrate prediction?


Did you miss the part about 40% runrate being for desktop only?
But wait, didn't AMD had outstanding increase in desktop market share during 2006Q4? Yes it did. Furthermore we wont be able to discuss a runrate of 12-31-2006 'till 2007Q1 results are in.

You may pike off now.

12:46 AM, February 04, 2007  
Blogger Scientia from AMDZone said...

azary omega
to mr. Sharikou:
could you please do something about hundreds of ravaging pro intel kids that reply or simply post FUD on this blog? Seriously, this blog is deteriorating thanks to posts like this one:
Anonymous said...
Is this blog dead?

....i mean freedom of speech is important, but this is really isn't much of a blog anymore. Do something.


Actually, the simplest thing you can do is disable anonymous posting. This works very well for two reasons: First of all only more serious posters tend to bother setting up an account. Secondly, if someone has a consistently skewed point of view it is easy to tell since they cannot hide behind the anonymous sig. Basically, people have to stand up for what they post.

Incidently, turning off anonymous posting prevents spoof posting as Sharikou180 did on his blog when he pretended to be posting as Sharikou.

My blog seems to work pretty well. My settings are:

Who can comment? - Only Registered Users

Enable comment moderation? - No

Show word verification for comments? - No


There doesn't seem to be as much reason to moderate when you can tell who said what. Also, I've had my blog up for months and have only received two spam messages so word verification doesn't seem to be needed. It is much easier to post without word verication.

2:22 AM, February 04, 2007  
Anonymous Anonymous said...

""WTF is "Frag" Oh lets see...
Frag is .. assassination of an unpopular member of one's own fighting unit.

How does K8L "assisinate" a clovertown, what a moronic word selection. You sure you got a PhD.

All I know INTEL can't be beat..

Lets see, on track for between 4-6 billion in profits on 34-36 Billion in sales, worlds largest semiconductor company. Demonstrated HiK metal gate in 2003, fully fuctional 45nm SRAM test chip in 2006 and fully functional 45nm x86 CPU in December. Building 3 factories to pump out 200+ million x86 dual/quadcore 45nm in 2008 at 50% margins. Got 3 65nm factories pumping out chipsets, graphics, and celeron 65nms. Damm AMD can't even get 65nm ramped and INTEL is pumping out Core2 overclocking Celerons on 65nm. AMD sucking losses with ATI and billions in debt. Got a crappy silicon partner who looked like a fool at IEDM annoucing their great SOI, dual stress, lowK, strained silicon 45nm technology with super immersion lithograhpy only to find out INTEL got HighK metal Gate on their process....

AMD got its ass whipped in the market, in the stock market, on the ledger, and lastly in the benchmarks. Gamers prefer COre2 tells you where the performance is.

The only place AMD has any advantage is the 4 core+ market. And what does AMD do, instead the try and slug it out with INTEL in the 2 core below market and sucking read for all of 2007 and BK in 2008.

What was the most funny when Hector was asked why AMD choose to leave the nich profitable market he responded after careful study they had to compete in the broader market. Why is that, they can't survivi on server along, even they know that. But what is obvious is they can't make anymoney in the mainstream.

AMD is and will always be a bit player... kind of like Sharikou. Make lots of noise, sue lots of people, pretend they are bigger then they really are, but like sharikou they are nothing... ""

AMD with a almost 4 old cpu gains market share to a CPU that just came out...


Your post was deeply retarded but this link even makes it worse:

http://yahoo.reuters.com/news/articlehybrid.aspx?storyID=urn:newsml:reuters.com:20070131:MTFH03388_2007-01-31_00-12-24_N30384740&type=comktNews&rpc=44

4:21 AM, February 04, 2007  
Anonymous Anonymous said...

!!! Warning !!!

F.U.D. for your thoughts. Intel In-sight!

... couldn't help it.

5:48 AM, February 04, 2007  
Blogger Jeach! said...

....i mean freedom of speech is important, but this is really isn't much of a blog anymore. Do something.

I second that! Some 140 comments may seem impressive but when the majority of the comments are just Intelers blabing their frustration towards AMD or you without any logic or common sense it just gets to be a waste of all our time.

I'd rather read 14 great comments than 100 insults towards you or others!

6:50 AM, February 04, 2007  
Anonymous Anonymous said...

Isn't AMD still hiring people to design and build the K8L??

6:57 AM, February 04, 2007  
Anonymous Anonymous said...

Azary Omega said...
to mr. Sharikou:
could you please do something about hundreds of ravaging pro intel kids that reply or simply post FUD on this blog? Seriously, this blog is deteriorating thanks to posts like this one:


Yo you as stupid as your post. You really believe the PhD pretender and his BK message, the fire storys..

You are one dumb fuck aren't you.

9:26 AM, February 04, 2007  
Anonymous Anonymous said...

Azary Omega said...
to mr. Sharikou:
could you please do something about hundreds of ravaging pro intel kids that reply or simply post FUD on this blog?

Azary, why are you such an AMD fanboi? Is it because you like to cheer for the underdog because you were picked on in high school? Or is it because you prefer to see yourself as anti-establishment to shed your dork image? Either way you are a looser. Intel is crushing AMD, and there is nothing you can say on this board that will change that fact.

10:00 AM, February 04, 2007  
Anonymous Anonymous said...

Any update on 40% market share runrate prediction?

With this amount of fragging what will the runrate exiting 2007 be, 60%? 80%?

1:42 PM, February 04, 2007  
Anonymous Anonymous said...

"Azary Omega said...
to mr. Sharikou:
could you please do something about hundreds of ravaging pro intel kids that reply or simply post FUD on this blog?

Azary, why are you such an AMD fanboi? Is it because you like to cheer for the underdog because you were picked on in high school? Or is it because you prefer to see yourself as anti-establishment to shed your dork image? Either way you are a looser. Intel is crushing AMD, and there is nothing you can say on this board that will change that fact."

I don't normally post, but why are you personally attacking this poster? Do you know him in real life? He's expressing an opinion. If you don't like it, please refute it as opposed to making a personal attack.

If Intel has such an advantage it should be trivial for you to find the evidence and post it. Attacking the poster only makes people angry. It does not further your position.

2:12 PM, February 04, 2007  
Anonymous Anonymous said...

""Intel is crushing AMD, and there is nothing you can say on this board that will change that fact.""

Unbelievable how dumb intel fanboys can be... Just a few posts above your retarded remark...

""AMD with a almost 4 old cpu gains market share to a CPU that just came out... ""

http://yahoo.reuters.com/news/articlehybrid.aspx?storyID=urn:newsml:reuters.com:20070131:MTFH03388_2007-01-31_00-12-24_N30384740&type=comktNews&rpc=44""

2:27 PM, February 04, 2007  
Blogger Azary Omega said...

Anonymous said...
Yo you as stupid as your post. You really believe the PhD pretender and his BK message, the fire storys..


Get back to that 3rd world country from witch you came from with your bad English.

2:57 PM, February 04, 2007  
Anonymous Anonymous said...

Quote:"Isn't AMD still hiring people to design and build the K8L??"

No, you are thinking of the Fab in New York that Intel are going to pay for!

3:09 PM, February 04, 2007  
Anonymous Anonymous said...

Quote:"Intel is crushing AMD,..."

What you really is mean your third grade teacher is crushing your...

3:13 PM, February 04, 2007  
Anonymous Anonymous said...

Intelers are retarded.

6:23 PM, February 04, 2007  
Anonymous Anonymous said...

intelers are superretarded......................yeah...........

8:59 PM, February 04, 2007  
Anonymous Anonymous said...

Fanboy?

" Some 140 comments may seem impressive but when the majority of the comments are just Intelers blabing their frustration towards AMD or you without any logic or common sense it just gets to be a waste of all our time."

I got nothing to be impresed about damm INTEL totally fucked up and only made a billion dollars last quarter. It was an absolute disaster! Damm only 75% market share and a billion dollar profit during the middle of a price war. We had to dump all our PentiumIV space heaters.. it was so fucked we only made a billion dollars. Damm we had to outfit 3 65nm factoires, start paying for the 3 45nm factoires and spend that 2 billion on the 45nm HighK metal gate process. Damm we still only made 1 billion dollars after all that. Things are pretty grim, we laid of what a thousand managers and will be down another 4-10K people this year and be ramping 45nm.. we are so screwed. Damm we only made a billion dollars what a disaster. We shoudl fire that CEO as INTEL should be pulling in 8 billion a year instead of that palty 4 billion. Must be that pesky AMD with sucking benchmarks selling their CPUs to dell for under a hundred bucks.

Things are looking like a disaster we are only 18 months ahead of AMD.. they got that Barcelona coming and we got no asnwer till our 3 45nm factoires ramp on in a few months.. Damm I hope a billion dollar profits a quarter plus our billion dollar savings will carry over till we get 45nm going.. Boy things are looking pretty bad there at INTEL

9:56 PM, February 04, 2007  
Anonymous Anonymous said...

Interesting video, there are a few lies that he's telling and he shows this by touching his nose... the info on the AMD k8l is true but with reference to what Intel are doing with DC and IMC he shows that he now's more than letting on.

10:01 PM, February 04, 2007  
Anonymous Anonymous said...

Intel is pissing all over AMD for the last one year...I hope no one lost money by listening to this mindless nutcase..Intel filing for BK? wow..the company can buy AMD 15 times over..lol

12:51 AM, February 05, 2007  
Blogger Azary Omega said...

Anonymous said...
I got blah to be blah about damm INTEL blah fucked up and blah made a billion blah blah quarter. It was an absolute blah ! Damm only 75% blah blah and a billion dollar profit blah the blah of a blah war. We had to blah all our PentiumIV space heaters.. it blah so blah we only blah a billion dollars. Damm we blah to blah 3 65nm blah , blah blah for the 3 45nm blah and blah that 2 blah on the 45nm HighK blah blah process. Damm we blah only blah 1 blah blah after all that. Things are blah grim, we laid of what a thousand blah and will be blah another 4-10K people this blah and be blah 45nm.. we are so blah . Damm we blah made a billion blah what a blah . We shoudl blah that CEO as INTEL blah be blah in 8 billion a blah blah of that palty 4 blah . blah be blah blah AMD blah blah blah blah blah CPUs to blah for blah a hundred bucks.


So Sharikou, are you going to do something about this mess? C'mon, i WANT TO see this blog having discussions that actually mean something. Scientia is right about her settings - they work and not only for her blog, i'm sure those settings will work for you to.

Just try it for a week, what harm in that? Please.

4:22 AM, February 05, 2007  
Anonymous Anonymous said...

Quote:"Intelers are retarded."

Stop OVERESTIMATING them!

7:18 AM, February 05, 2007  
Blogger PENIX said...

K8L will set a new industry standard for 64-bit computing. I will be first in line to preorder as soon as that option becomes available.

I advise everyone to liquidate their Intel stock and be ready to purchase AMD stock. Once K8L hits the streets, AMD stock will go through the roof.

8:17 AM, February 05, 2007  
Anonymous Anonymous said...

Anon said:

"Unbelievable how dumb intel fanboys can be... Just a few posts above your retarded remark..."

Is that the best you can do? AMD gained marketshare by giving away processors for free to Dell. Lets see how long that lasts.

9:33 AM, February 05, 2007  
Anonymous Anonymous said...

The fact that Intel Crew makes so much noise speaks for itself. Something happening?

Btw. what's that IBM process guy doing in AMD? I read about that in AMDzone.

10:06 AM, February 05, 2007  
Anonymous Anonymous said...

"""
Is that the best you can do? AMD gained marketshare by giving away
processors for free to Dell. Lets see how long that lasts."""

The best you can do is saying Intel is crushing AMD, when in reality AMD is beating Intel with a almost 4 year CPU against a new CPU... so please stop anoying people with your stupidity.

10:08 AM, February 05, 2007  
Anonymous Anonymous said...

"The best you can do is saying Intel is crushing AMD, when in reality AMD is beating Intel with a almost 4 year CPU against a new CPU... so please stop anoying people with your stupidity."

Aha! So the same to you stupid AMDroids. Since you claim C2D is a hotrod P3, which on the other hand is a hotrod Pentium Pro. So a hothotrod Pentium Pro is beating K8? So it would seem. In other words, a 12 year old processor is beating a 4 year old one...

3:04 PM, February 05, 2007  
Blogger Scientia from AMDZone said...

Sharikou

The word verification is annoying because it never works the first time; I always have to do it twice to post anything.

Secondly, as long as you allow anonymous posting, your blog is susceptible to spoof posting where anyone can pretend to post as someone else. Sharikou180 did this on his blog.

I would definitely urge you to turn off both word verification and anonymous posting. This would improve the blog comments by about 400%.
--------

I'm puzzled that I keep seeing people talk about building three 45nm FABs. Intel is only building two, FAB 32 and FAB 28. D1D will produce 45nm first since it is the R&D FAB for new processes. However, it is not a volume FAB so output will be limited.

As far as Intel's supposed success in 2006 let's look a bit more closely. Intel's revenues declined to about what they made in 2004 while AMD's have doubled since 2004. That's a bit of a difference.

Also, let's look at history. When Intel released the northwood core in 2002, AMD got slammed. It's revenue share dropped by 66% in just two quarters. Although it did recover some the following quarter, it only came back up to what it had been 2 years earlier.

Now we have a similar situation where Intel launched a new architecture in 2006. However, this time, there was only a slight change in revenue share. What it looks like is that people understand that what Intel did in 2006 was not impressive. So, it seems that these people are basically projecting this forward to 2007 and sort of trying to pretend that Intel is having some kind of delayed benefit. However, in 2007 AMD will launch its own new architecture and Intel won't have anything substantially new until 2008.

The reality is that AMD got clobbered in 2002 when Intel released northwood but AMD started coming back in late 2003 with a new architecture. Comparing then to now I'm wondering why AMD would not then be able to gain in 2007 with a new architecture. Yes, I've seen people try to pretend that things are really dire for AMD now but by mid 2003 AMD had had 6 straight quarters of no profits. So, a single quarter with no profits doesn't seem that big a deal.

I've also seen the statement about Intel doing a 45nm memory chip but noticed that no one mentioned that AMD did the same thing just a few months later. Also, no one mentioned that IBM/AMD also have a high-K gate design that should be in production in 2008. So, Intel does not have a high-K advantage.

It is true that Intel is building two new 45nm FABs but these FABs will have to ramp so they won't help much in 2007; they will be a strong factor in 2008. I would also expect Intel to start upgrading one of its older FABs to 45nm once these two come online. AMD will have its own 45nm production in 2008 although obviously it won't ramp as quickly as Intel. However, the current best guess is that Intel will have lost some of its process lead on 45nm.

For anyone who still thinks that Intel is doing so well all you have to do is count up how many advantages Intel has lost since K5.

K6 - first X86 processor to match the performance of Intel.

K7 - first X86 processor to exceed the peformance of Intel.

3DNow! - first floating point extensions for X86.

K7-MP - first X86 processor to match Intel's dual socket server capability. first X86 with point to point interprocessor and I/O bus.

AMD 760MP - first dual FSB chipset.

K8 - first X86 processor with 64 bit extensions. first X86 with IMC. first time AMD was able to create new X86 standards.

X2 - first dual core processor.

FAB 36 - first time an X86 FAB matched Intel's 300mm capability. First time AMD had more than a single FAB capable of cpu production.

Purchase of ATI - first time AMD became capable of doing serious chipset and gpu planning beyond the limited chipsets that it produced for K7 and K8. first time Intel matched in terms of integrated graphics chipsets and stable image platforms.

AMD has caught up with Intel in many ways. AMD is far better off today in terms of manufacturing capability than it was in 2003. It is far better off in terms of chipset development. And, its financial position even after Q4 is still better than it was in Q1 2003. AMD is not going to go bankrupt in 2008 nor is AMD going to keep losing money all during 2007. In 2003, AMD still lost money for the first two quarters and they did just fine. What reason would there be to prevent AMD from doing well in 2007 with a new architecture?

3:55 PM, February 05, 2007  
Blogger brian222 said...

I agree with Scientia about disabling anonymous posting. The quality of the posts from anonymous posters has been deteriorating. I simply don't read many of them. They are a distraction and I would love to see them disappear. My two cents.

4:55 PM, February 05, 2007  
Anonymous Anonymous said...

Btw. what's that IBM process guy doing in AMD? I read about that in AMDzone.

That guys is Grossman. FYI he used to run Fishkill when IBM was still trying to do development for volume manufacturing. During his tenure IBM had to give up on DRAMs, contract out to foundry one of the two 8" fabs on the site. Now he is going to AMD.. How could INTEL get so lucky?

7:42 PM, February 05, 2007  
Anonymous Anonymous said...

"The best you can do is saying Intel is crushing AMD, when in reality AMD is beating Intel with a almost 4 year CPU against a new CPU... so please stop anoying people with your stupidity"

Frankly it doesn't matter whether INTEL with a one year old design is beating AMD's 4 year old design on 65nm against their 90nm. Damm even when INTEL was getting its ass kicked with 90nm with a 4 year old netburst design one thing was still the fact. INTEL 75% market share and 2 billion profit a quarter.

Damm things are rough these days. Core2 kicking AMD in every benchmark. Marketshare slide stopped at 75%, winning server share back, profits down to a horrible billion a quarter with gross margins a full 10% + over AMD.

Yup we got nothing to crow about but them profits and damm good benchmarks..

7:45 PM, February 05, 2007  
Blogger Azary Omega said...

The best you can do is saying Intel is crushing AMD, when in reality AMD is beating Intel with a almost 4 year CPU against a new CPU... so please stop anoying people with your stupidity.

Brilliantly said! Regardless on which side you might be, the numbers are hard to ignore. Maybe next quarter the numbers will change, maybe not, but now they show that C2D and C2Q may just not be enough to beat grandpa Athlon.

11:22 PM, February 05, 2007  
Blogger Ho Ho said...

Scientia
"Also, no one mentioned that IBM/AMD also have a high-K gate design that should be in production in 2008. So, Intel does not have a high-K advantage"

Problem with that is, AMD won't use it before it goes 32nm somewhere around 2011. IBM will use it on 45nm.

Btw, there are some news about Barcelona. Basically it is what I expected:
highest clocked CPU on midyear launch will be 2.3GHz, 2.5GHz following in Q307 and 2.6GHz in Q208.

Who wants to bet that by Q307, Intel will release its 3Ghz quadcores?

12:54 AM, February 06, 2007  
Blogger Aguia said...

ho ho,
The clock speed is not important; what’s important is the performance it achieves.

Or what you are trying to say is that Intel in order to compete needs to clock their processors at 3.0Ghz?

6:16 AM, February 06, 2007  
Blogger netrama said...

It pains me to see that several major OEMs and retailers are still selling the highly energy inefficient Celerons and Pentiums even now. Some examples include Celeron laptops from Toshiba and Sony and Dell (a push since last 5-6 months), where a clear choice is the Turion ...well only if these guys started thinking about their customers and buyers !!!

8:43 AM, February 06, 2007  
Blogger Ho Ho said...

aguia
"The clock speed is not important; what’s important is the performance it achieves."

I know that. It's just that everything I've heard so far put Barcelona and C2Q to roughly the same level when comparing core-to-core, clock-to-clock. Assuming I'm correct AMD will have some problems sinec Intel has released faster clocked quadcores couple of months ago.

aguia
"Or what you are trying to say is that Intel in order to compete needs to clock their processors at 3.0Ghz?"

No, I just say it can do it if there would be the need.

11:19 AM, February 06, 2007  
Blogger Lex said...

Some one said "It pains me to see "

I'll tell you it pains me not to see silly people with allegiance to a company with a banrkupt balance sheet, a banrkupt manufacturing position, a bankrupt market share, and a bankrupt product.

I'm sure in the PhD's eyes it is INTEL. How about that Q4 results from AMD and INTEL there! Billion profit vs a loss. Get used to it there is more of that coming.

9:20 PM, February 07, 2007  
Blogger Scientia from AMDZone said...

ho ho
Basically it is what I expected:
highest clocked CPU on midyear launch will be 2.3GHz, 2.5GHz following in Q307 and 2.6GHz in Q208.


Yes, this sounds correct.

Who wants to bet that by Q307, Intel will release its 3Ghz quadcores?

Wrong. Even by Q1 08, Intel's highest clock on quad core will only be 2.4Ghz. I think you'd better change your bet to Q3 08.

11:00 AM, February 08, 2007  
Blogger Dr. Nehalem, Deity Status said...

Whoa, whoa, whoa:

All Celeron laptops are based on Pentium M designs, that's why they're still called Celeron M in some places. Look at the GHz, if they're in the 1.5-2.0 Ghz as of 2/2007, that's what they are. NOT based on Netburst. Last Netburst in laptops that I'm aware of was the P4 Mobiles, about 3 years ago.

Overclockers have already proven you can get monstrously high overclocks on all Core2 processors. What this means is as the factory process is refined, procs can be clocked faster stock and still be within a certain power envelope. Betcha Intel already knows that and will quickly punch the gas if Barcelona is even close.

Think that core2 quad can't do 4 Ghz? Think again. The core2 architecture on 45 nm will have 4+ GHz parts. I'll bet Dr. Sharikou on that one. Q208 - either Intel BK's or Intel is healthy AND has 4+ GHz shipping parts. I'm staking my bet on 2 things occuring, not just one. Any takers?

9:52 PM, February 10, 2007  
Blogger coldpower27 said...

Scientia from AMDZone

Wrong. Even by Q1 08, Intel's highest clock on quad core will only be 2.4Ghz. I think you'd better change your bet to Q3 08.

Did you suddenly forget about the Core 2 Extreme QX6700 or the Xeon X5355 or am I missing something here?

Given what we know now, the likeliest scenario is a higher clocked Core 2 Extreme Yorkfield in Q4 2007 December, with the Core 2 Quad Yorkfield's coming in during Q1 2008. Clockrates will likely be increased, as Intel was immediately able to gain clockspeed with the Smithfield to Presler transition.

X2 - first dual core processor.

The above is simply incorrect, Intel beat AMD by about 4 days on this with the Pentium EE 840, AMD was first though with a "native" implementation meaning capable of interprocessor communication on the die.

hoho

Who wants to bet that by Q307, Intel will release its 3Ghz quadcores?

The current data we have suggests that Intel will release Core 2 Extreme in Late Q4 2007, to say they released 45nm parts in 2k7, with the bulk majority of product moving in Q1 2008. This would hold with average 2 yr cycle for process transitions.

10:55 AM, February 16, 2007  

Post a Comment

Links to this post:

Create a Link

<< Home