ie8 fix
madison

Getting to the bottom of the 3.0 GHz AMD quad-core demos

By | July 31, 2007, 5:03am PDT

Summary: [Update 8/16/2007 - It has been confirmed that the photographs were shot the same day and sent out without an NDA.  The photo was not a "canned for the press" shot that my sources told me.] There’s been some controversy brewing over the issue of the 3.0 GHz “K10 Barcelona” desktop-variant quad core processors that were demoed [...]

[Update 8/16/2007 - It has been confirmed that the photographs were shot the same day and sent out without an NDA.  The photo was not a "canned for the press" shot that my sources told me.]

There’s been some controversy brewing over the issue of the 3.0 GHz “K10 Barcelona” desktop-variant quad core processors that were demoed at last Thursday’s AMD analyst day.  Accusations on whether the chip was cherry picked or possibly even staged have been flying back and forth and Charlie Demerjian had some strong words to say about it and that they were definitely real.  There were questions raised because the initial photograph that made the rounds over the Tech news sites were identical and I was told that it was a “canned” photo sent out to various journalists the day before the analyst meeting with an NDA release date of noon the next day during the Analyst meeting [See update above].  AMD disputed the accusation that real photographs from the media were barred and pointed me to some of Charlie Demerjian’s blog and photos and indeed it was true.  [Update 8/2/2007 - Kristopher Kubicki of DailyTech responds to Charlie Demerjian]

Demerjian also pointed out that the demos had conventional air coolers indicating that it wasn’t a case of extreme overclocking and that these weren’t picked out of a batch of “hundreds of thousands of chips”.  But Demerjian admitted that these were sorted and tested out of a few chips to pick out the ones that would hit 3.0 GHz.  Furthermore, Demerjian’s colleague Nebojsa Novakovic pointed out that no CPU-intensive benchmarks like Linpack (which would have fully loaded all four cores) were run nor were any benchmarks released.  Running a game isn’t likely to put the same kind of stresses on an overclocked CPU so the demos, while real, weren’t truly put to the test.  Furthermore, the 3.0 GHz quad core AMD “Phenom” chips don’t appear on any roadmaps and the 2.6 GHz quad core Phenom processors won’t launch until Q1 2008 and by that time Intel will be shipping 3.33 GHz “Penryn” quad core processors.  The actual AMD Barcelona server-variant chips will only be clocked at 2.0 GHz when they launch sometime in September.

Note: To give some more perspective, various overclockers had demoed air-cooled 3.8 GHz Intel Core 2 Duos in Q3 2006 and 3.8 GHz Intel Core 2 quad cores in late 2006 and those were actually sold factory overclocked.  That doesn’t mean Intel could actually ship upper 3 GHz parts and the top performing Intel Core 2 chips officially run at 3.0 GHz.  But the latest “Stepping G” manufacturing revision Intel Core 2 chips will apparently run at a cool 3.6 GHz without any voltage boosts and the overclocking forums are raving about it.

So the fact that AMD showed off a 3 GHz desktop-variant of a quad core “Barcelona” CPU at analyst day does lay to rest some of the fears that AMD can’t scale Barcelona though the workloads that were shown didn’t really stress the demo systems and no benchmarks have been released for the 3.0 GHz parts.  While this is certainly positive news for AMD, I have a hard time agreeing with Demerjian’s conclusion that the folks at Intel where having “heart attacks” over this news given the roadmaps of the two companies.  The real challenge for AMD in the next six months is to start delivering some 2.6 GHz parts on time and even then it will have to face the onslaught of 45nm 3.33 GHz Penryn quad core processors.

Kick off your day with ZDNet's daily e-mail newsletter. It's the freshest tech news and opinion, served hot. Get it.

Topics

Disclosure

George Ou

http://blogs.zdnet.com/Ou/?page_id=557

Biography

George Ou

George Ou, a former ZDNet blogger, is an IT consultant specializing in Servers, Microsoft, Cisco, Switches, Routers, Firewalls, IDS, VPN, Wireless LAN, Security, and IT infrastructure and architecture.

57
Comments

Join the conversation!

Just In

Thanks for another proof
CFKane 5th Aug 2007
The people critizing George's biased reporting here are sticking to the topic, while you come up with a completely unrelated comment, which is just linked to the story through the "AMD vs. Intel" aspect. You couldn't have shown more clearly what your real motives are and the fact that you're also notoriously applauding George's blog posts with "Thanks for looking into this, George" and the like, tells a lot about what audience his perspective pleases.

Thank you.
0 Votes
+ -
Oh, it would be nice if...
nucrash 31st Jul 2007
AMD just surprised us all and delivered the 3 GHz Quad cores from the starting line. Hector Ruiz would stand up at the pulpet and say, "2.0 GHz" was a typo, we actually meant 3 GHz, but thought we would let it slide for the time being.

And then I woke up and got ready for work.

We can all dream though.
0 Votes
+ -
The 3.0 Phenom is not on the AMD roadmap for the next 5-6 months according to DailyTech.
0 Votes
+ -
They showed the chips and the horsepower via a video game, but they wouldn't allow any full benchmarks or anything of that nature. And they pulled a demo off of a chip that technically isn't to be released to the public for atleast another 5 to 6 months.

As I said before, it would be nice if AMD had something up their sleeve and attempted to pull a fast one on Intel. Hinting that they have something out there, but not allowing anyone to know how really bad or good it was. (Hence no heavy benchmarks)

Intel doesn't have to do much to scale up their CPU power, all they would have to do now is just re-lable some Quad Cores and watch AMD scrable to make up the difference.

AMD doesn't seem like the company that would draw up a fake roadmap, but if they did, it may just restore some consumer confidence that was lost after the Quad FX and the HD series video cards.
0 Votes
+ -
The type of roadmaps I'm talking about aren't fake and they can't be fake. AMD's partners rely on these roadmaps for internal planning and a lot of partners actually adjusted their plans to build more Intel based motherboards this year. They would be VERY angry if AMD lied to them about the roadmap because it would have VERY severe consequences to their planning and forecasting.
0 Votes
+ -
I know releasing a roadmap provides vendors with a way to better understand what lie ahead and to build up to the expected specification.

Although, I would think that if AMD actually moved up their release date on some processors that aren't a major structural change such as a different socket, then motherboard vendors shouldn't care all that much as long as the faster chips are guaranteed to work with their boards. From what I remember, most mainboards are designed to handle chips that may or may not be released for months. Hence the over-clocking of performance chips.
0 Votes
+ -
Reality
Nil Einne 31st Jul 2007
While you made some good points, I think you've missed some key details. Firstly, there is no chance in hell IMHO that the 3.0ghz is unstable whether you use linpack or Prime95 or Call of Juarez. If there any instability in LinPack, that would me there is always going to be a risk it will crash in any resonable load including Call of Juarez. And there is no way in hell that AMD will risk their demo comp crashing. No, I think we can be sure the 3.0 ghz was very stable. But just because it's very stable, doesn't mean it's something AMD is likely to be selling. There are a variety of reasons why AMD may not be willing or able to sell a 3.0ghz even if they are able to stably achieve that. From what I can tell from what Charlie said, they definitely are able to achieve this in sufficient quantities so that won't be an issue. But there are still a variety of other reasons which I won't go into.

Secondly, on roadmaps, it seems to me that a key thing you're missing is that it doesn't make sense for AMD to promise a 3.0ghz Phenom on roadmaps. Intel may very well deliver the 3.33ghz Penryn on time, all signs are pointing to it but Intel can and has failed before and this is a new part. If Intel does fail to deliver, it doesn't make sense for AMD to release a product to compete against a non-existant Intel product. AMD is currently in survival mode and they need to concentrate on what Intel is delivering now and be ready to compete with what Intel delivers rather then trying too hard to gain a lead over Intel. If and when Intel delivers the 3.33ghz Penryn then sure, it may make sense for AMD to deliver a 3.0ghz of higher Phenom. And if AMD put the 3.0ghz on their roadmaps then they HAVE to deliver or they just look bad.

Finally, and more importantly I think what Charlie was getting at is that the problem for Intel is that it looks like AMD is going to be able to seriously compete in the low to mid end again. For a long while, even with the Core 2 blowing AMD out of the water, via aggressive pricing AMD was able to compete with Intel in the low to medium-high end. But AMD are losing ground and if they didn't deliver something good soon, they're going to be stuck in the low end. With this, AMD has a very good chance of regaining the medium-high end. Whether or not AMD can compete in the ultra high end doesn't matter that much IMHO. Sure there is great glory in having the performance crown but ultimately, what matters most to AMD (and Intel) is where they make their money which is not the ultra-high end
0 Votes
+ -
I hope you are right.
nucrash 31st Jul 2007
I would love AMD to "Under-promise and Over-deliver"

Or even to return to the days of old and, here is this AMD 2.0GHz, watch it overclock to 3.0GHz and run stable.
0 Votes
+ -
'no chance in hell'?
georgeou 31st Jul 2007
"Firstly, there is no chance in hell IMHO that the 3.0ghz is unstable whether you use linpack or Prime95 or Call of Juarez"

And you know this for a fact? You do understand that most games don't come close to perfect scaling and they don't come close to full throttle across all four cores right?
0 Votes
+ -
AMD bashing is "in"
christopher@... 31st Jul 2007
Or has the company really fallen apart? It seems like the investment required to build new fabs is holding the company back from producing faster chips. Intel has the largest market share, and the biggest leverage as far as chip manufacturing goes, so it's really a no brainer that they produce faster chips, and since they have many more chips to sort through, the economies for picking the fastest chips is also reduced.

Let's face it, for brute force in manufacturing, Intel wins hands down, for cutting edge ideas, AMD has led for a long time (which is why Intel used many of the same ideas after the fact)
0 Votes
+ -
Thanks for looking into this, George
thetruthhurts 31st Jul 2007
AMD could have picked a benchmark that pegged the CPUs 100%
Umm, why didn't it choose to use POVray _in addition to_ whatever game they were demoing?

I understand AMD wanting to show a 'platform centric' demo but it seems like they have never recovered from their 'simulated benchmark PR fiasco' or the Computex PR fiasco with POVray.

Whatever. It will all be sorted out shortly. Intel remains failry quiet (short of the controlled leaks we've read about on their upcoming 45nm CPU frequencies and TDP numbers) which in my opinion means that it's preparing to trounce on AMD's planned Barcelona late-August 'intro/whatever' dog and pony, ASUMING it's still on track.
Intel won't talk about it publicly, but the leaked roadmaps are pretty clear about it. Based on leaked information, Penryn 3.16 GHz servers (probably 3.33 for 2P 8-core high-end workstations) will officially launch this year which means it will have to ship sometime in October. Desktop and mobile parts will probably ship around the end of this year but official launch and availability will be early next year. I can?t say this leaked information is accurate or not, but the leaked info is usually reliable.
0 Votes
+ -
.
AMD Analyst Day is not for tabloid bloggers

George Ou did not attend AMD's Analyst Day because it was for analysts, not tabloid bloggers who have an adversarial agenda.

Whine, whine, whine!

Now that AMD has actually DEMONSTRATED two Barcelona quad core systems running at 3.0 GHz (with no evidence of overclocking whatsoever), George Ou STILL has nothing good to say about it. This sure makes George Ou's pathetic rabid four-day blog-a-thon attack on AMD at the beginning of July look pretty silly.



How wrong was/is George Ou about AMD?

On July 6, 2007, George Ou wrote: "...The problem is that AMD still posted scores for a non-existent AMD [2.6 GHz Barcelona] product based on "simulated" results which is completely bogus and deceptive..." (1)

On July 26, 2007, AMD DEMONSTRATED a 3.0 GHz pre-production product, just 20 days later. This shows quite conclusively that George Ou is completely full of "Intel partisan" hot air. The only thing "completely bogus and deceptive" is George Ou and his idiotic tabloid blog.



What has AMD shown?

So, what did AMD show? It's mighty impressive:

1. These Barcelona CPUs are the only public quad core X86 CPUs on the planet.

Intel's "quad core" processors are two dual-core CPUs, in one package, connected by their front-side busses. When you put two V8 engines under the hood, you don't have a V16. Intel does not have a quad core CPU. Only AMD does.

2. These Barcelona CPUs are the fastest X86 CPUs on the planet.

Benchmark results pending, at 3.0 GHz, these pre-release AMD CPUs should also be the fastest X86 CPUs on the planet. AMD's Barcelona has superior performance-per-clock over Intel's Core 2-based CPUs. The fastest Intel "quad" runs at 3.0 GHz (but you cannot find and buy one).

Bravo AMD! Bravo! I know I speak for many people who LOVE to watch George Ou eat crow!



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

(1) AMD removes bogus benchmarks
http://blogs.zdnet.com/Ou/?p=570
0 Votes
+ -
Calm down Henri, er Drew, er, TechExec2...
thetruthhurts 1st Aug 2007
Psst, watch this: http://youtube.com/watch?v=vWG-BSuu7SE

Quiet period is over.

You got nothing.
0 Votes
+ -
.
Here's a video you, George Ou, and your Intel partisan friends might find embarrassing. Henri is right. Intel has been illegally manipulating the market for many years. You really should THANK AMD. Without them, the fast Intel CPU you want would cost $1500 and the slow one you settle for would cost $800.

PC industry hooked on Intel's drug money, says AMD exec
http://youtube.com/watch?v=s5-47NAfwik



P.S. It's quite sad how you guys pound the table about a little AMD hypocrisy and let Intel get off with criminal market manipulation.
0 Votes
+ -
That sums it all up right there
dragosani 1st Aug 2007
"It's quite sad how you guys pound the table about a little AMD hypocrisy and let Intel get off with criminal market manipulation."
0 Votes
+ -
Day late, dollar short
thetruthhurts 3rd Aug 2007
I can afford $1500 CPUs so it does not matter to me how much they cost.

Intel's the technology leader and they provide the best platform value today.

How many AMD 'simulated native quad cores' can I buy today in retail? Zippo. And it's not because of your manipulated market claims, etc.

Should I start discussing AMD's financial status next?
.
Step by step for those who rode the SHORT BUS TO SCHOOL:

Step 1. Intel manipulates the market so it sells more CPUs: Carrot and/or stick to entice component makers and OEMs to not do business with AMD.

Step 2. Intel sells more CPUs and makes more profit as a result of their illegal market interference. AMD sells fewer and makes less profit.

Step 3. AMD has less money to invest in R&D and astronomically expensive CPU FABs than they would without Intel's illegal market interference.

Step 4. AMD struggles to survive in the technology race (but, they still have the only public quad core X86 CPU).

Step 5. Intel lowers prices over and over to bleed AMD to death, hoping to drive AMD into bankruptcy and out of the market.

Got it now? Intel's actions are blatantly predatory and illegal.

I'm happy for you that you don't care what CPUs cost. Just remember: A fool and his money are soon parted! laugh laugh laugh
0 Votes
+ -
History of the World, Part MCMXVII
thetruthhurts 4th Aug 2007
Step 1. AMD, a scrappy also ran, lies still like Jabba the hut for 3+ years and gets caught with its pants down by Core 2 Duo.
Step 2. AMD panics and starts to undercut its poorly performing products in an attempt to retain market share, but fails especially on the DP market.
Step 3. Intel introduces "TRUE QUAD CORE AMD KILLER" at dual-core price points. What quad core product does AMD have to compete? Zip.
Step 4. AMD panics, knowing it does not have a Quad core product and an aging microarchitecture. AMD resorts to a price war with Intel.

Step 5. AMD gets into debt. Can't afford to pay its bills. Has to go ask the street for money to pay its utility bills.

Step 6. AMD in shambles. AMD realizes they have no one to blame but themselves, yet try and lauch a smear and mud slinging campaign against Intel, which backfires.

Step 7. Intel, whose flawless execution is unequaled by AMD, sells > 1 Million quads, cementing itself as the undisputed leader. Where's Waldo [AMD?].

Step 8. AMD whines about Intel unfair a monopolistic tactics. Journalists tired of the rhetoric: "AMD: you got no one to blame but yourself"

Step 9. AMD realizes their mistake. In attempting to be just like Intel, they have managed to whittle away shareholder value and bleed the company into a financial morass - their debt relegated to junk.

Step 10. AMD forced to change its business model because they cratered themselves. Plant closures, layoffs, deferal of capital investment lead to AMD falling behind.

Step 11: AMD files for Chapter 7 protection. Perhaps this is where they want to go so they can do away with their debt one and for all. Dillusion of shares, the end follows.

happy AND THEY ALL LIVED HAPPILY EVER AFTER happy
0 Votes
+ -
TechExec2 aka AMD guerilla marketeer
thetruthhurts 4th Aug 2007
Back at you.

Randy Allen, AMD VP Server and Workstation Division

"Barcelona to completely blow away the existing Clovertown product in every dimension"

http://youtube.com/watch?v=pSw2GTLHtQo

Nice dive AMD stock has taken lately. Keep losing money in GFX, keep burning cash, and keep bankrolling those empty studies to distract the public from the real train wreck that AMD has become:

AMD study concludes what it was paid to conclude concludehttp://news.com.com/8300-10784_3-7.html?authorId=140

AMD's War with Intel Becomes a Street Brawl
http://www.betanews.com/article/AMDs_War_with_Intel_Becomes_a_Street_Brawl/1186183278

Good luck stopping the bleeding and good luck with the antitrust lawsuit against you in GFX and with the patent infringement suits against you that you can't afford to litigate.
0 Votes
+ -
Substance...
Sensi-Kul 4th Aug 2007
"Nice dive AMD stock has taken lately. Keep losing money in GFX, keep burning cash, and keep bankrolling those empty studies to distract the public from the real train wreck that AMD has become:"

It won't be the first or the last time stocks take a dive and cash is burned. Anything substantive on the subject matter?

I can dig up articles about Intel's past and how it was being jostled around (and sometime embarrassed) in terms of performance and innovation in years past. I can also access current Intel stories concerning what is increasingly becoming internationally known as unfair monopolistic (anti-innovative) practices. However, it won't contribute to the subject matter. So I'll refrain.

I know, you're in the midst of a rebuttal to TechExec2... but, in the process the information you're bringing to light is unfortunately not pertinent. Yes, those are interesting stories to read, but AMD is nowhere likely to go anywhere for a long, long time. Hence, we're posting about "Getting to the bottom of the 3.0 GHz AMD quad-core demos" which WILL hit the market and WILL continue to place pressure on Intel (hence, the rush to 45nm, rush to lower prices and get rid of inventory and undermine the introduction of Barcelona). As an Intel faithful, I can only imagine you'll benefit from this these two companies' rivalry -- both from the standpoint of advancements in technology & innovation and reduction in price.

In actuality, the "good luck" should be directed toward Intel, who (if you haven't read lately) may face some serious consequences regarding their "monopolistic" practices... not just in fines, but in entire markets, who increasingly favor a partnership (whose rug won't be pulled from underneath it, unannounced) and an open platform.

The game is still on and alive and well... George Ou blogging about it (and you posting messages to it) is a testament to this fact.
0 Votes
+ -
Substance... which you lack
thetruthhurts 4th Aug 2007
It won't be the first or the last time stocks take a dive and cash is burned. Anything substantive on the subject matter?

Yes, goes to the heart of the issue. Showing 3.0GHz CPUs in an attempt to show the world+dog AMD is 'still in the game.' Unnfortunately, the reality is more grim for AMD (and I happen to know people that *work* there...do you?).

Hence, we're posting about "Getting to the bottom of the 3.0 GHz AMD quad-core demos" which WILL hit the market and WILL continue to place pressure on Intel (hence, the rush to 45nm, rush to lower prices and get rid of inventory and undermine the introduction of Barcelona).

Where on the roadmap is this 3.0 GHz AMD quad-core Phenom demoed that WILL hit the market, according to you? Show us the goods, won't you?

In actuality, the "good luck" should be directed toward Intel, who (if you haven't read lately) may face some serious consequences regarding their "monopolistic" practices... not just in fines, but in entire markets, who increasingly favor a partnership (whose rug won't be pulled from underneath it, unannounced) and an open platform.

I suggest you get educated. With respect to the EU: AMD's whinny ways are starting to irritate. The latest PR gimmick backfired.

AMD's openness is a necessary evil, not something they are doing out of benevolence and "customer centric innovation" or any other favorite AMD marketing lines. Spare me, oh AMD apologists with the "monopoly" blabery already.

There is no game. The game is over for AMD. RIP.
0 Votes
+ -
Prophet...
Sensi-Kul 4th Aug 2007
?Where on the roadmap is this 3.0 GHz AMD quad-core Phenom demoed that WILL hit the market, according to you? Show us the goods, won't you??

I was actually stating that AMD quad-cores will be released to the market and continue to place pressure on Intel. (To think I was stating otherwise would be to also think I was stating the demos will hit the market, too ? if you solely were focusing on what I had placed in quotations, the blog title.)

Nonetheless, back to your comment, ?Where on the roadmap is this 3.0 GHz AMD quad-core?. Is a roadmap the only guideline (or guarantee) of products to come in 6-12 months? Nobody knew of an x2 6000+ until late September (and those were unofficial and leaked) ? 5 months prior to its February launch. At the time, there were already roadmaps past Barcelona. I?m not even here to preach 3.0 GHz is absolutely going to come to fruition (although it?s seems inevitable, considering AMD?s history of progressive process refinement and the fact they?ve achieved this frequency at 90nm, even though nobody knew how far X2 would scale when it was released ? not even a roadmap). However, you stating it won?t happen because it?s not on today?s roadmaps is a bit myopic.

?(and I happen to know people that *work* there...do you?).?

Ok. You got me there. Because you know people who work there, it means you have more information about Barcelona than all of us. Now that we?ve established this fact, I will have all of my questions concerning K10 directed toward you.

How much of an IPC advantage (if any) will K10 generally have over C2D and K8 in INT, FP, in single- and multi-threaded workloads, encoding, 3D rendering, and single-thread gaming?
On what exact date can I expect to purchase a Phenom?
Can you send me detailed R800 specs?
What are Barcelona?s Memory Bandwidth, Processor Multi-Media, Processor Arithmetic, Cache and Memory, and Memory Bandwidth scores in SiSoftware Sandra @ 2Ghz?
What is the absolute highest AMD has been able to clock the very best processor from their current batch, what was the required voltage, and what was its TDP?
What was the exact yield percentage from last week?s production Barcelona wafers?

If you knowing ?people that *work* there? means anything, you can nail this and hundreds (if not thousands) of other related discussions shut. You can end all of the debate and speculation here and now.

We?re waiting.
0 Votes
+ -
FYI, AMD was STILL passing those bogus benchmarks two weeks ago in Europe and they got caught. They slammed me for accusing them of doing this two weeks ago and that it wasn't them but some Russian partner, but they finally had to admit they did in fact hand out those bogus scores AGAIN even after they yanked those benchmarks.

http://www.techarp.com/showarticle.aspx?artno=424&pgno=6
"George also pointed out that AMD removed the "simulated" benchmark results from their website on July 6th, and promised to post proper benchmark results of their Barcelona processor. However, we received the very same benchmark results just last week. If these results have already been officially junked by AMD, why are we still being served the same poo?"

So this business of poor old AMD is getting old. They did the deed after all.
0 Votes
+ -
One Trick Pony working on a second
nucrash 1st Aug 2007
That's why they did all the legal maneuvering not so long ago. Something to strike at Intel while they had them down. Now AMD is on the ropes and can't seem to get a handle on getting things together.

As much as we keep hoping, they keep disappointing.

I wonder if they would have argued if I loaded the folding SMP client on those two Phenom Demos. "Not a Benchmark" So I would be curious to see how long they would have taken before the starting causing troubles.
.
Reminder: You have been debunked

I have already thoroughly debunked your rabid blog-a-thon against AMD here (1)(2). I see nothing wrong with AMD distributing estimated benchmarks for an unreleased processor so long as they are clearly labeled as such (and they were).



Slam AMD like this...But beware

If you want to slam AMD (and you do), just remind people that AMD publishing estimated benchmarks shows the weak position they are currently in, and leave it at that. That is enough humiliation.

Of course, AMD's fortunes can quickly change with Barcelona, just like Intel's did last year with "Core 2", and AMD's did before that with "Hammer" and "ClawHammer".



Estimated Barcelona benchmarks are fine

Contrary to what "Tech ARP" says, there is nothing surprising about AMD not having official SPEC benchmarks for an unreleased processor. As an Intel partisan, you should know this. Intel did not have official SPEC benchmarks for the XEON Quad 3.0 GHz that they "released in April" until June 12, 2007 (1).



This is a George Ou Production

You (George Ou) are at the center of this controversy, not AMD. You're a one man anti-AMD crusade.

** You wrote the four-day rabid blog-a-thon attack on AMD filled with proven falsehoods and things you just made up to demonize AMD.

** You contacted "Tech ARP" and propagated your own false controversy. You cannot cite a "Tech ARP" story that you fed to them! laugh Only our mindless Intel partisan readers will fall for that! laugh



A far far more important story: Market manipulation by Intel

Intel's market manipulation and antitrust litigation in Japan (Intel lost), the European Union, and the United States is a far far more important story. I've known it for many years. And, you know it too.



And another thing

I hear crow tastes like chicken. Does it? laugh laugh laugh



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

(1) Debunking George Ou's rabid attack on AMD
http://talkback.zdnet.com/5208-10533-0.html?forumID=1&threadID=35931&messageID=661568&start=-1

(2) Baloney! This post FURTHER PROVES George Ou is unfair to AMD!
http://talkback.zdnet.com/5208-10533-0.html?forumID=1&threadID=36193&messageID=665742&start=-1
0 Votes
+ -
Define "bogus"
Robert Crocker 2nd Aug 2007
The fact that AMD was quite upfront in pointing out that the benchmark numbers provided were in fact simulated then where do you keep getting off slinging around the word "bogus"?
0 Votes
+ -
You just don't get it!
ajgarc 2nd Aug 2007
" The real challenge for AMD in the next six months is to start delivering some 2.6 GHz parts on time and even then it will have to face the onslaught of 45nm 3.33 GHz Penryn quad core processors."

This is about getting the best performance for a given energy
expenditure. The Xeon vs opteron was a good example. Guess who won? For me and maybe for those who care about energy use, this is what is important. Until we have the benchmarks, it would be nice to lay off this pissing contest angle and stick to the news which was to demonstrate that a 3.0 Ghz barcelona is doable. A 2.0GHz with the the same opteron energy requirement for now is good and 3 Ghz would be great.
No benchmarks released 1 month before launch is very telling. AMD knows they can't match Intel on single-threaded applications when memory bandwidth has no significant effect. When you take memory bandwidth out of the equation (for desktop applications), you're left with the Execution Engine and Cache advantage of the Core 2 architecture and Core 2 kills K8 or K10 at single threaded applications. You also have the SSE4 factor for video encoding advantage for Intel Penryn.

This is precisely why AMD sells an Athlon X2 3.0 GHz cheaper than an Intel 2.66 GHz Core 2 Duo because they know they do worse clock-for-clock on desktop applications. So when you're looking at 3.33 GHz Penryn versus 2.6 GHz Barcelona on the DESKTOP, the gap is even wider than the numbers suggest.
0 Votes
+ -
Interesting...
Sensi-Kul 2nd Aug 2007
1. Please provide the source which proves Core 2 kills K10 in single-threaded applications. You mentioned there are no benchmarks released. Do you have a K10 at your desk?

2. Athlon X2 had a ridiculous IPC (instruction per clock) advantage over all Pentium processors ever released. Core 2 reversed this and now has the IPC crown. Now you've compared Core 2 2.66 with an Athlon X2 3.0. We already know the IPC disparity between Core 2 and K8 (a 4-year-old architecture). That was established over a year ago.

Why is it difficult for you to believe that a new architecture will an IPC advantage over Core 2? We don?t have benchmarks against Core 2 yet, however AMD has made it clear that we can expect approximately an 80% increase in IPC over K8, per core. This translates, quite plainly, that K10 will be faster than Core 2 at the same clock. The question solely lies in how far K10 will ramp up in terms of clock frequency.

3. K10 will actually feature many of the key SSE4 instructions which help to accelerate video encoding (in select applications). This will not be an advancement exclusive to Intel.
AMD won't improve single-threaded applications till Bulldozer and that's likely due in early 2009 with a possible ship date in late 2008. The Intel Core 2 Execution Engine for non-HPC applications is known to be superior to AMD K10 until it becomes bandwidth constrained for highly-threaded applications that can saturate all 8 cores. So Barcelona K10 doesn't start beating Core 2 "Clovertown" until we get to 8 threads and 2.6 GHz. At 2 GHz and 8 threads, Clovertown actually beats AMD K10 at Integer performance because memory bandwidth isn't hurting Intel's Clovertown enough yet. The reason for this is that Intel's Execution Engine and Caching is superior at fewer thread count or fewer GHz. The problem for AMD is that they won't initially scale server-based Barcelona's to 2.6 GHz until probably Q2 2008 and by then Intel will be much further ahead.
0 Votes
+ -
Perspective...
Sensi-Kul 2nd Aug 2007
To state there will be absolutely zero single-threaded improvements K10 vs. K8 (and that we??ll have to await Bulldozer to witness single-threaded improvements) is false.

Here are a few (of many) thing we know about K10:

2 execute, 2 load, and 2 store 128-bit operation per clock (versus C2D??s one)
?? Double wide (bandwidth-wise) L2
?? Shared L3
?? Significantly more efficient memory controller (with a write buffer, so burst writes can be made ?V negating the performance hits of bus turn-arounds). Also DRAM now has prefetchers working in conjunction with the write buffer.
?? More accurate branch prediction

To suggest that a single-threaded application at 2.0 GHz on K8 will perform identically to the same single-threaded application at 2.0 GHz on a K10 is unfounded.

If anything, it might be sager to state K10 may not have an advantage in single-threaded applications over C2D. But, to state C2D will kill K10 in this respect is simply fuel for controversy on your blog (which seems to be the point of this thread). It is clear there will be noteworthy single-threaded improvements, K10 over K8, closing the gap with Intel or leveling the playing field (or perhaps better).

However, with K10??s absolute FP, multi-core scaling, and IMC advantages, even a level single-thread playing field does not speak in favor of Intel, considering the wide array (and increasing development and use of multi-threaded) applications being run on a workstation.
0 Votes
+ -
Don't put words in my mouth
georgeou 2nd Aug 2007
"To state there will be absolutely zero single-threaded improvements K10 vs. K8 (and that we??ll have to await Bulldozer to witness single-threaded improvements) is false"

I never said that. I said AMD doesn't compare to Intel on single threaded performance and they won't really make any major changes on single threaded performance until Bulldozer. Yes it's better with K10 but not enough to overtake Core 2 on single threaded applications. Sure, K10 will own HPC by a mile, but that's a different subject.
0 Votes
+ -
Chewing on words...
Sensi-Kul 2nd Aug 2007
"AMD won't improve single-threaded applications till Bulldozer"

If the above statement does not mean zero improvements, please enlighten me. You typed those words. Whether or not YOU chose to put them in your mouth is for you to share with us.

I understand that you have, since, clarified this statement: "they won't really make any major changes on single threaded performance until Bulldozer". However, part of the process of strengthening credibility is stating what you mean and meaning what you state, the FIRST time.

You also commenced this train of thought with: "the Execution Engine and Cache advantage of the Core 2 architecture and Core 2 kills K8 or K10 at single threaded applications".

Again, you can speak for K8 (as it is a released and benchmarked product), but can you account for K10? There is no information, other than between some walls at AMD, that can prove or disprove whether or not C2D will have any advantage, at all, at the same clock. Contrarily, almost all of the (limited) information we now have suggests AMD will have an absolute lead clock-for-clock. Even AMD Barcelona naysayers accept this, and instead focus on the fact AMD's low clock frequencies may not be enough to make up with their large IPC advantage alone. Do you believe a 2 GHz K10 will offer less performance than a 2 GHz C2D (in either a single- or multi-threaded application)?
0 Votes
+ -
Your 80% claim is laughable
georgeou 2nd Aug 2007
That "80%" improvement of K10 over K8 you cited is four K10 cores versus two K8 cores.
.
Sensi-Kul, I appreciate your effort. But, George Ou essentially NEVER provides a direct answer to a direct question. He never admits the truth even when you present it in unequivocal language right in front of him. He's worse than a used car salesman. I don't trust anything he says.

You're dealing with someone (George Ou) who refuses to play it straight with anyone about anything.

Advice: Just debunk him, and show your proof, and leave it at that. Doing otherwise will only frustrate you and waste your time.

All the best to you...
"however AMD has made it clear that we can expect approximately an 80% increase in IPC over K8"

That "80%" improvement is 4 K10 cores versus 2 K8 cores, lol.
0 Votes
+ -
Look again...
Sensi-Kul 2nd Aug 2007
Only your resopnse was laughable.

From your very own site, review:
http://www.zdnetasia.com/news/hardware/0,39042972,61984216,00.htm

"We expect across a wide variety of workloads for Barcelona to outperform Clovertown by 40 percent," Allen said. The quad-core chip also will outperform AMD's current dual-core Opterons on "floating point" mathematical calculations by a factor of 3.6 at the same clock rate, he said.

Also:

"A faster floating-point engine performs mathematical calculations--long an Opteron strong suit, though not as important a part of the chip as that for integer operations. At a given clock frequency, a Barcelona core outperforms a current Opteron core by a factor of 1.8. By going quad-core, a Barcelona chip overall will provide a boost factor of 3.6, Allen said."

(And, yes, I realize this refers to FP and not INT operations)
0 Votes
+ -
1.8x boost for integer using twice the cores isn't my idea of a massive leap in per-core performance or single core performance.
0 Votes
+ -
Can you read?
CFKane 3rd Aug 2007
Neither Sensi-Kul nor the article he was quoting were talking about a 1.8x boost for integer using twice the cores.

The article mentions a 1.8x boost per core for floating-point[/p] performance. That is a valid counter-point to your claim of insignificant advantages in IPC, which you couldn't back up anyway.
0 Votes
+ -
Reading 101...
Sensi-Kul 3rd Aug 2007
CFKane,

Yes, I can read.

And if you can, too, you will recall George Ou stated:

"That "80%" improvement is 4 K10 cores versus 2 K8 cores, lol"

And that was what I was clarifying, CFKane.

Unlike you (CFKane), who has posted sound information (stating it has "insignificant advantages in IPC", with which I agree), the article illustrates how George Ou has, again, posted invalid information.

http://www.zdnetasia.com/news/hardware/0,39042972,61984216,00.htm
0 Votes
+ -
Math?
Sensi-Kul 3rd Aug 2007
?1.8x boost for integer using twice the cores?

How have you derived (or calculated) 1.8x using twice the cores (INT)?
0 Votes
+ -
I was replying to George's answer
CFKane 3rd Aug 2007
He's the one with reading problems. He reiterated his "twice the cores" claim while the article clearly talks about a factor of 1.8 per core (3.6 for twice the number of cores) and additionally confused integer and floating point performance. I do not consider that 80% increase (if it holds true) insignificant, btw.

It is of course an early statement by an AMD official and should be taken with a grain of salt for that reason alone. The raw throughput of floating point operations is also not that important in many desktop and server applications, but that all doesn't matter. George Ou's statement of insignificant IPC advantages was just unfounded and again another example for his "in dubio pro Intel" attitude.
0 Votes
+ -
Interesting...but you are not.
thetruthhurts 4th Aug 2007
1. Do you have a K10 at your desk? Please share.

2. Really? I happen to have a Core 2 Pentium under my desk and it decimates your X2 box by 25-30%

You don't have benchmarks against Core 2? Well, you all had SIMULATED benchmarks? Is that not enough for you? LOL

3. K10 will not feature anything even close to Intel's SSE4. You are a liar.
0 Votes
+ -
.
Warning: George Ou does not have ANY performance data for the K10 (Barcelona) processors. He is making this up.



"...You also have the SSE4 factor for video encoding advantage for Intel Penryn..."

It is very gratifying to see George Ou having to resort to talking about UNRELEASED INTEL PRODUCTS in order to make a favorable argument for Intel (and he doesn't have data for THAT either!). Bravo! Outstanding! laugh laugh laugh



"...This is precisely why AMD sells an Athlon X2 3.0 GHz cheaper than an Intel 2.66 GHz Core 2 Duo because they know they do worse clock-for-clock on desktop applications..."

Wrong!

The AMD Athlon 64 X2 6000+ (3.0 GHz, $170) delivers about the same performance as the Intel Core 2 Duo E6700 (2.66 GHz, $317) in desktop applications. The PC WorldBench benchmark of desktop applications shows this quite clearly.

The market price of the AMD and Intel CPUs is based on what customers are willing to pay. It has NOTHING to do with what "AMD knows" or "Intel knows".

Customers get their information from lots of sources, including misleading ones with an agenda like this idiotic blog.

Flaming Intel Partisans(tm) like George Ou help to mislead customers that "Intel is faster than AMD". It is a lie! It is absolutely not true! The only thing that IS true is that Intel's very fastest CPU (about $1000) is currently faster than AMD's very fastest CPU (about $170, an INCREDIBLE bargain).

But, AMD's Barcelona is going to change that. That Barcelona quad-core 3.0 GHz that AMD demonstrated last week will soon be shown to be the fastest publicly-shown X86 CPU on the planet.

Poor George! We'll soon find him wandering the streets muttering: Penryn!...SSE4!...Henri bad!...Henri hypocrite!...Mommy! laugh laugh laugh



And another thing...

I hear there is a job opening at the National Enquirer alien abduction reporting desk. George Ou ought to check it out! He is a perfect fit! laugh laugh laugh



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

(1) AMD Athlon 64 X2 6000+, 3.0 GHz, $170
http://www.newegg.com/Product/Product.aspx?Item=N82E16819103773

(2) Intel Core 2 Duo E6700, 2.66 GHz, $317
http://www.newegg.com/Product/Product.aspx?Item=N82E16819115002

(3) Intel Core 2 Duo E6750, 2.66 GHz, $220
http://www.newegg.com/Product/Product.aspx?Item=N82E16819115029
0 Votes
+ -
Hey, TechExec2...
thetruthhurts 3rd Aug 2007
TechExec2 wrote:

"...You also have the SSE4 factor for video encoding advantage for Intel Penryn..."

It is very gratifying to see George Ou having to resort to talking about UNRELEASED INTEL PRODUCTS in order to make a favorable argument for Intel (and he doesn't have data for THAT either!). Bravo! Outstanding!


Are Barcelona/Phenom UNRELEASED AMD PRODUCTS?

Aren't these the products you've been thoroughly DEFENDING, arguing about simulated results this and that?

*** for tat
No, wait, it is too funny to see you with your pants down.

I totally agree that nobody at Intel even raised an eyebrow about the 3GHz Phenom demo - they probably know much more about Barcelona than AMD would show publically. But I also have little doubt that you were among those close to a heart-attack when the news about that demo surfaced.

You couldn't keep yourself from commenting about it on your blog even though the ratings of your recent AMD-related posts clearly show that your view on those matters is not what people are interested in. And now you're blabbering inconsistent nonsense with next to no relation to the message you replied to. Who the hell would care about the single-threaded performance of a quad-core processor, even if there would be enough evidence to justify such a comparison at this point?

I see George Ou getting hysterical in the face of a demoed possibility for a product which will not be shipping in the foreseeable future, but might make Intel's top-performers look second best, if it were so.
0 Votes
+ -
"Who the hell would care about the single-threaded performance of a quad-core processor"

Who cares about single threaded performance? Oh, about 90 plus percent of the desktop market. The development world hasn't caught up to multi-threaded development yet.
0 Votes
+ -
The fact that 90 plus percent of the desktop software is not (massively) multi-threaded is exactly the reason why 90 plus percent of the processor sales are not made up of quad-core parts. Somebody deciding to go for a quad-core cpu will surely run applications that take advantage of the four cores and will not focus on benchmarks in the single-threaded area.

Stop grasping at straws here. A Phenom will definitely not be too slow to run MS Office productively and it will surely outperform any X2 or Core 2 Duo in Cinebench and Povray.
While benchmarks will show speeds and performance in artificial situations, Real world outcomes are a different thing and one has to get a fair opinion. I hope had provided that.

a dual core amd vs a dual core Xeon test like this one
(http://www.worlds-fastest.com/wfz995.html)
would be a better gauge of cpu technology.

In comparing two CPU technologies one have to put a leveled playing field. Just like formula 1 cars. Same engine displacement, car length, weight. No much different with CPU.
So if we compare a 65 nm 4 quad core 2.0, 2.6 or ever 3.0 GHz system we ask ourselves which would consume the least amount of energy consumption and provide the best processing output. As i have said before, no benchmarks are available so it would be wise to put the FUD out of one's "respectable" column.

Let the actual results, when they arrive, be the truth. I ask no less from the author given that he has the ability reach many around the globe.
0 Votes
+ -
Wrong again...
Sensi-Kul 2nd Aug 2007
Only your response was laughable.

From your very own site, review:
http://www.zdnetasia.com/news/hardware/0,39042972,61984216,00.htm

"We expect across a wide variety of workloads for Barcelona to outperform Clovertown by 40 percent," Allen said. The quad-core chip also will outperform AMD's current dual-core Opterons on "floating point" mathematical calculations by a factor of 3.6 at the same clock rate, he said.

Also:

"A faster floating-point engine performs mathematical calculations--long an Opteron strong suit, though not as important a part of the chip as that for integer operations. At a given clock frequency, a Barcelona core outperforms a current Opteron core by a factor of 1.8. By going quad-core, a Barcelona chip overall will provide a boost factor of 3.6, Allen said."

(And, yes, I realize this refers to FP and not INT operations)
0 Votes
+ -
Thanks for another proof
CFKane 5th Aug 2007
The people critizing George's biased reporting here are sticking to the topic, while you come up with a completely unrelated comment, which is just linked to the story through the "AMD vs. Intel" aspect. You couldn't have shown more clearly what your real motives are and the fact that you're also notoriously applauding George's blog posts with "Thanks for looking into this, George" and the like, tells a lot about what audience his perspective pleases.

Thank you.

Join the conversation!

Formatting +
BB Codes - Note: HTML is not supported in forums
  • [b] Bold [/b]
  • [i] Italic [/i]
  • [u] Underline [/u]
  • [s] Strikethrough [/s]
  • [q] "Quote" [/q]
  • [ol][*] 1. Ordered List [/ol]
  • [ul][*] · Unordered List [/ul]
  • [pre] Preformat [/pre]
  • [quote] "Blockquote" [/quote]
ie8 fix
Click Here
ie8 fix

The best of ZDNet, delivered

ZDNet Newsletters

Get the best of ZDNet delivered straight to your inbox

Facebook Activity

White Papers, Webcasts, & Resources
ie8 fix
ie8 fix