Tuesday, September 18, 2007

AMD should do a 1.5 core CPU

With K10's advanced technology of clocking core independently, what I proposed all along can be easily realized -- an asymmetric processor with two cores running at different speeds.

It will be very nice to have one core running at 3.6GHZ and the other at 2GHZ.

Intel has 1.5 core technology already. An Intel quad core is more like 1.5 core. 4 cores on a shared FSB is like 4 dogs breathing through a single straw -- a retarded situation.

47 Comments:

Blogger Intel Fanboi said...

Why not 1/2 a core?

4:02 PM, September 18, 2007  
Blogger Bubba said...

You just described a feature of Nehelam.

Not that it will matter.

Intel put AMD out of business today.

4:04 PM, September 18, 2007  
Blogger Hornet331 said...

@bubba
i think the mobile penryn are also capable of this.

5:20 PM, September 18, 2007  
Blogger Intel Fanboi said...

This comment has been removed by the author.

5:47 PM, September 18, 2007  
Blogger Intel Fanboi said...

Carried over from the previous topic:

Christian M. Howell wrote:

I've worked at Microsoft and make a high 5 figure salary. I guess I won't be an enterprise software architect either. Oh wait, I already am.


You are an enterprise software architect? Lets check the web:

Christian M. Howell is an analyst programmer for Accenture. Christian maintains and assists in developing websites for major pharmaceautical companies and other large companies. His personal specialty is in windows and web programming.

Ha ha ha! Some architect you turned out to be. Just another web hack who has trouble creating links in a blog. If that was my job I guess I would retreat into a fantasy world of being a "screenwriter trapped in a SW developer's cube". At least they pay you enough to keep you in silly hats.

5:48 PM, September 18, 2007  
Blogger Christian M. Howell said...

Christian M. Howell is an analyst programmer for Accenture. Christian maintains and assists in developing websites for major pharmaceautical companies and other large companies. His personal specialty is in windows and web programming.

Ha ha ha! Some architect you turned out to be. Just another web hack who has trouble creating links in a blog. If that was my job I guess I would retreat into a fantasy world of being a "screenwriter trapped in a SW developer's cube". At least they pay you enough to keep you in silly hats.


Wow, you found my Princeton Premier profile. I guess it was worth the money. Where's yours?
But then, why are you stalking me you perv?

What does that have to do with AMD or Intel?

I guess though that K10 CAN already do that with the right BIOS. It would only have to link to the HW clock.

Each core CAN be clocked independently. All it would need is to clock two cores higher. I was actually wondering if they might do that.

6:00 PM, September 18, 2007  
Blogger Evil_Merlin said...

Hrm, if you worked for Microsoft you musta been a contractor.

No proof that you have ever worked here before.

Whoops...

7:32 PM, September 18, 2007  
Blogger Scott said...

Christian works for New York City.


But anyway, enough about BaronFudrix.



I have a reason why AMD shouldn't do a 3.6 ghz core and a 2ghz core:

AMD can't get their current process above 2ghz.

So perhaps they should do a 2ghz core and a 5mhz core. That might be more doable.

7:38 PM, September 18, 2007  
Blogger howling2929 said...

Speaking of the triple-core announcement, Intel's Otellini got off a shot at AMD by responding to the Inquirer's query about whether Intel would ever consider making a triple-core part with this zinger: "We see a distinctive advantage in having all the cores on one die work."

Ouch.

Paul Otellini at IDF

Nuff Said!

7:38 PM, September 18, 2007  
Blogger Giant said...

Speaking of the triple-core announcement, Intel's Otellini got off a shot at AMD by responding to the Inquirer's query about whether Intel would ever consider making a triple-core part with this zinger: "We see a distinctive advantage in having all the cores on one die work."

But by having low yields AMD is "innovating" and giving people choices by giving people a 3 core CPU!

7:46 PM, September 18, 2007  
Blogger Giant said...

Some images of the Nehalem demo:

http://www.fudzilla.com/index.php?option=com_content&task=view&id=3122&Itemid=51

Notice they were actually running some 3D stuff. This is in stark contrast to AMD's task manager demo!

7:52 PM, September 18, 2007  
Blogger Intel Fanboi said...

Why not a 1/4 core?

7:54 PM, September 18, 2007  
Blogger netrama said...

Speaking of the triple-core announcement, Intel's Otellini got off a shot at AMD by responding to the Inquirer's query about whether Intel would ever consider making a triple-core part with this zinger: "We see a distinctive advantage in having all the cores on one die work."


Paul O has no clue on the engineering side,, a better answer would have been ...'we just have to worry about making 2/2 cores work instead of 4/4 in our double cheese burger design.' ... or better yet...
'What AMD is making a Tri Core? ...we are gonna make a Pentium Core ...(mmmm I love that name Pentium)
Patty and Ratty can you see , if we can glue another core to our quad core ...wait will that be then 5 core or 6 core ..." at this point Paul O appears a bit confused..

lol lol lol

8:01 PM, September 18, 2007  
Blogger lex said...

Life is so funny these days.

Anandtech said it best

INTEL is on a Tick Tock

AMD has no schedule..

Paul had it right, what is AMD to do with a bunch of quadcores with only 3 cores working. Damm marketing genious lets convince people 3 cores are better then 2.

Only a Phony could think of something as stupid as 1.5. We all know why he has to make up for is failures with his PhD title.

Its the first day of IDF and you see Penry launch, wiggling Nehalem.

Damm can't wait till day 2

8:04 PM, September 18, 2007  
Blogger Giant said...

You just can't make it up: Barcelona vs. Xeon:

http://www.hardware.info/en-UK/articles/amdnZWppZGWa/New_quad_core_server_CPUs_AMD_Barcelona_vs_Intel_Harpertown/1

Barcelona gets eaten alive by the 2.66Ghz quad core Xeon. Never mind the Harpertown CPU they tested as well! AMD might have to sue Intel again for making CPUs just too fast!

8:11 PM, September 18, 2007  
Blogger Giant said...

Paul O has no clue on the engineering side,, a better answer would have been ...'we just have to worry about making 2/2 cores work instead of 4/4 in our double cheese burger design.' ... or better yet...

Just a bit upset that Intel can supply all the quad core CPUs it needs on time at speeds of up to 3Ghz are you? Meanwhile AMD paper launches Barcelona with no availability until November.

You do know that if AMD's yields were above an abysmal 30% they wouldn't be making these tri-core CPUs don't you?

Finally, what do Cheeseburgers and CPUs have to do with each other?

8:24 PM, September 18, 2007  
Blogger oneexpert said...

This comment has been removed by the author.

8:45 PM, September 18, 2007  
Blogger Giant said...

IMAGINARY PAPER LAUNCH PRODUCTS DONT EXSIST....

I agree totally oneexpert. Paper launches like Barcelona for instance?

Now oneexpert, I've got a little task for you. Open this link:

http://www.hardware.info/en-UK/articles/amdnZWppZGWa/New_quad_core_server_CPUs_AMD_Barcelona_vs_Intel_Harpertown/1

Now go and have a look through, how many benchmarks does AMD win vs. Intel? Ignore the Harpertown results for now.

I'll give you the answer: Not too many. That's a 2.66Ghz Clovertown (shipping for over 10 months now) vs. a non available Barcelona that you can't buy until November. By then Intel will have launched Harpertown and extended this existing lead even further!

8:54 PM, September 18, 2007  
Blogger Giant said...

You can buy AMD barcelonas today...

Please provide proof of this.

BUY INTELhi performance, state of the art, energy saving, low cost cpus, platforms, and video solutions...
REAL PRODUCTS FROM REAL PEOPLE THAT
ACTUALLY EXSIST. UNLIKE AMD BARCELONA PAPER LAUNCHED CPUS NOT AVAILABLE UNTIL NOVEMBER.

9:02 PM, September 18, 2007  
Blogger oneexpert said...

This comment has been removed by the author.

9:53 PM, September 18, 2007  
Blogger Bubba said...

Well giant if you think all the companys offering barcelona systems are liars or maybe they are just fooling you.
HP
SUN
IBM
APPRO
CRAY

Please point me to a link where I can buy a Barcelona system today.

You can't, they don't exist.

On November the 12th AMD will learn what a hard launch is.

On November the 12th the final nail will be hammered into AMD's coffin.

It's pretty amazing that Intel is showing running chips of their second generation quad core, while AMD has yet to ship their first generation part.

10:07 PM, September 18, 2007  
Blogger Giant said...

They all might be liars giant but I cant find any harpertown parts or systems can you?????

Oh They do have Barcelona servers. They just can't sell them yet because AMD is still trying to produce enough 2Ghz Barcelona CPUs.

Since when did Intel say that Harpertown was released now? Since when did anyone claim to be sellinga a Hapertown server today? November 12 is a day that will go down in history as the death of AMD.



BUY Intel and NVIDIA AND VIA AND IBM AND SUN AND FREESCALE AND SONY hi performance, energy saving, low cost, cpus, platforms and video solutions.

INTEL CLOVERTOWN 142,000,000% faster than AMD Barcelona!

10:26 PM, September 18, 2007  
Blogger Only AMD said...

In the first public demonstration of the Nehalem processor, Otellini said it will deliver better performance-per-watt and better system performance through its QuickPath Interconnect system architecture, which will include an integrated memory controller and improved communication links between system components.

Notice the comment about "integrated memory controller" WOW! Need i say more!!!!

11:22 PM, September 18, 2007  
Blogger Alex said...

Only AMD:

Intel had a SOC in 2002 with internal memory controller, this is nothing new, at the time they stopped the project right before it went into production.
So Intel had it long time before AMD. I s that all you can come up with? After getting your fanboi behind nailed today?
I guess I should not ask you dorks when AMD will have its 32nm out should I?

11:46 PM, September 18, 2007  
Blogger Giant said...

WOW! AN INTEGRATED MEMORY CONTROLLER! Why didn't Intel think of that?! Wait, they did. The 486 had an integrated memory controller! They moved it off-die to the Northbridge starting with the Pentium CPU.

1:14 AM, September 19, 2007  
Blogger tech4life said...

giant said
Now oneexpert, I've got a little task for you. Open this link:

http://www.hardware.info/en-UK/articles/amdnZWppZGWa/New_quad_core_server_CPUs_AMD_Barcelona_vs_Intel_Harpertown/1

Now go and have a look through, how many benchmarks does AMD win vs. Intel? Ignore the Harpertown results for now.

I'll give you the answer: Not too many. That's a 2.66Ghz Clovertown (shipping for over 10 months now) vs. a non available Barcelona that you can't buy until November. By then Intel will have launched Harpertown and extended this existing lead even further!


We're talking about a 2.0 Ghz Barcelona against a 2.6 Ghz Clovertown and a 3.0 Ghz Harpertown. What is funny is that the Barcelona was actually able to beat the much higher clocked Intel cpus on some benchmarks. I don't see this as a reason to brag for Intel fanboys.

5:59 AM, September 19, 2007  
Blogger Evil_Merlin said...

Skulltrail doesn't FASN8, it does obliterate

IDF Fall 007 Seaburg, iceberg, we're all borged



By Nebojsa Novakovic: Tuesday 18 September 2007, 22:56

TOWARDS THE END of Kicking Pat's keynote, he showed a spread of new 45nm Penryn platforms, including the anticipated Skulltrail. The second generation V8 platform has two - unlocked supposedly - HarpertownDP chips similar to the ones in our, um, delayed review this morning.


Couple it with "optimised" - read streamline latency and bandwidth at the expense of some RAS features - Seaburg chipset, four OC FB-DIMM sockets (one module per channel) with T1 command rate, and top-notch VRM and board cooling. Then add cascaded dual phase-change - 80 degrees C freeze cooling, we're talking about a potential 5+ GHz, eight-core monster with dual FSB2000+ and (I assume liquid-cooled too) quad-channel DDR2-1000+ CL4 FB-DIMMs. Oh boy...



There are four PCI-E x16 v1 GPU slots on board, created by expanding the 32 PCI-E x16 v2 lanes of Seaburg chipset using twin PCI-E v2->v1 bridges into 64 PCI-E x16 v1 lanes, i.e. four slots. This would give you Quad CrossFire or Quad SLI capability at pretty much full GPU I/O speed, since most near-future GPUs are more than happy with the 16-lane wide PCI-E per GPU.

Intel's man behing the SkullTrail (and the famed Bad Axe D975XBX2 board), the ex-Alienware Eric Cubbage, also showed a bunch of early "Skulltrail friends", among the pioneers to do a part of the first sales units of the new platform from December onwards. These included the UK-based SCAN Computers and the high-end cooling from the very local Vigor Gaming from nearby City of Industry, CA.

Oh boy, how bad it is that there is no AMD FASN8 anytime soon now, at least we could have something to compare the Skulltrail to.

6:42 AM, September 19, 2007  
Blogger Christian M. Howell said...


http://www.fudzilla.com/index.php?option=com_content&task=view&id=3122&Itemid=51

Notice they were actually running some 3D stuff. This is in stark contrast to AMD's task manager demo!


If Intel is so great why can't they get equal load across all cores like AMD did with their Task Manager presentation?

BTW, there is no way to "simulate" load in Windows so it was probably running a low footprint stress test.

7:11 AM, September 19, 2007  
Blogger Christian M. Howell said...


Hrm, if you worked for Microsoft you musta been a contractor.

No proof that you have ever worked here before.

Whoops...



DO you know the difference between a Blue Badge and an Orange Badge?

How about the difference between an STE and SDET?

7:15 AM, September 19, 2007  
Blogger Christian M. Howell said...


Christian works for New York City.


But anyway, enough about BaronFudrix.



I have a reason why AMD shouldn't do a 3.6 ghz core and a 2ghz core:

AMD can't get their current process above 2ghz.

So perhaps they should do a 2ghz core and a 5mhz core. That might be more doable.



Yet another tragic tale of misspent youth. Aaaahh, America.

7:16 AM, September 19, 2007  
Blogger Evil_Merlin said...

Anyone with google can figure that out Christian.

it doesn't take a lot of work...

For those interested, blues are full time Microsoft employees. Orange Badgers are contractors.

The same goes for Software Test Engineers and Software Design Engineers.


Anything else you would like to know?

7:42 AM, September 19, 2007  
Blogger pointer said...

This comment has been removed by the author.

7:53 AM, September 19, 2007  
Blogger pointer said...

Christian M. Howell said...

http://www.fudzilla.com/index.php?option=com_content&task=view&id=3122&Itemid=51

Notice they were actually running some 3D stuff. This is in stark contrast to AMD's task manager demo!


If Intel is so great why can't they get equal load across all cores like AMD did with their Task Manager presentation?

BTW, there is no way to "simulate" load in Windows so it was probably running a low footprint stress test.


you said you are programmer ... nvm, try this batch file:

echo off
:loopstart
goto loopstart


run 2X instances of the above batch file for every cores that you have and observe task manager get 'stressed' while not work being done by the CPU. (Indeed most work involved is how the Task manager showed the load)

7:56 AM, September 19, 2007  
Blogger Christian M. Howell said...


Anyone with google can figure that out Christian.

it doesn't take a lot of work...

For those interested, blues are full time Microsoft employees. Orange Badgers are contractors.

The same goes for Software Test Engineers and Software Design Engineers.


Anything else you would like to know?



Don't call me Christian. My name is The BaronMatrix. Do a search on Tomz forum where I posted a pic of MY BLUE BADGE.

9:31 AM, September 19, 2007  
Blogger oneexpert said...

This comment has been removed by the author.

9:53 AM, September 19, 2007  
Blogger oneexpert said...

This comment has been removed by the author.

10:03 AM, September 19, 2007  
Blogger Evil_Merlin said...

Still on that Pentium III kick huh cocklick, whoops, I mean oneexpert.

Still no comment on how that supposedly glued together Pentium III is BEATING AMD's latest and greatest.

its already WELL proven in the marketplace that thus far Barcelona is a bust for AMD with even the middle bin Xeons available TODAY beating Barcelona, and its being proved handily at IDF that AMD doesn't stand a chance unless it can ramp Barcelona up fast. VERY fast.

Which doesn't seem likely...


Keep dreaming to yourself about AMD oneexpert...

You sound just like those blow hard Mac fans that are convinced Mac rule the world when in fact they are a very minor part of it but like to make the noise to think they are better.

10:57 AM, September 19, 2007  
Blogger Scott said...

Christian; our tax dollars hard at work, posting FUD on this forum.

The citizens of New York called, they want you to go back to work and program something for them rather than promote a private company with their tax dollars.

10:59 AM, September 19, 2007  
Blogger Giant said...

What does Intel think of Native Quad core in 65nm?!

http://www.fudzilla.com/index.php?option=com_content&task=view&id=3149&Itemid=51

I still haven't stopped laughing since I saw that!

Oh tech4life, who cares about the clockspeeds? It's not Intel's fault that AMD is stuck at 2GHz with 30% yields.

Oneexpert, no one here cares about a three core CPU. The only reason AMD is making them is because AMD's yields are so terrible. Intel will just use faster dual core CPUs to bridge the gap. Since AMD is stuck at 2Ghz this is easy to do. A 3Ghz dual core is faster than 2 Ghz tri core. Or they could just drop quad core prices.

INTEL ONLY. QUAD CORE TODAY. $279 on Newegg. Don't bother waiting until AMD can get yields above 30% to get a few quad core CPUs together for sale.

BUY Intel and NVIDIA AND VIA AND IBM AND SUN AND FREESCALE AND SONY hi performance, energy saving, low cost, cpus, platforms and video solutions.

INTEL CLOVERTOWN 142,000,000% faster than AMD Barcelona!

11:04 AM, September 19, 2007  
Blogger Mo said...

I just don't know if INTEL can be stopped right now.
The giant has awoken and well the rest is pretty much history.

11:21 AM, September 19, 2007  
Blogger Giant said...

I just don't know if INTEL can be stopped right now.
The giant has awoken and well the rest is pretty much history.


Intel has vast resources. In terms of engineers, financial resources ($$$) and manufacturing resources (fabs).

This is far beyond what AMD has at it's disposal. It's actually quite amazing that AMD kept up with Intel and had a great lead with K8 for a few years considering the difference in the size of the companies.

If AMD wants to get back into gear, the first thing that needs to happen is that most of the executives have to go. Hector Ruiz (market share regardless of cost!), Randy Allen and Bob Rivet etc.

Idiots like Sharikou and Oneexpert might claim that pricing dual core CPUs at $60 to gain back market share in the low end desktop space is a good strategy but it's not sustainable. How long do you think AMD can keep taking $600m/quarter losses? Sure, the price war has hurt Intel's profits by about a billion a quarter, but they're still making a ton of cash every quarter.

Intel have the resources to continue the price war indefinitely, AMD does not. It's that simple. AMD needs to be totally turned around. Ruiz is not the person to do that. The first step is firing Ruiz. Bring back Jerry Sanders or promote Dirk Meyer.

11:49 AM, September 19, 2007  
Blogger Christian M. Howell said...

You just can't make it up: Barcelona vs. Xeon:

http://www.hardware.info/en-UK/articles/amdnZWppZGWa/New_quad_core_server_CPUs_AMD_Barcelona_vs_Intel_Harpertown/1

Barcelona gets eaten alive by the 2.66Ghz quad core Xeon. Never mind the Harpertown CPU they tested as well! AMD might have to sue Intel again for making CPUs just too fast!



Wow, you forgot how to use an href tag already?

What was your opinion when Opteron was making Intel look like a kindergarten lego team?

Probably different. Wanna bet that as new revs come out at higher speeds AMD will beat Harpertown?


That would have been a little more realistic had they gotten a 2360 rather than a 2350.

But as Anand's testers say, the 2347 blows away ALL XEONS in PERF\WATT, the most important benchmark for data centers.

Maybe it's the 3rd grade FBDIMMs.

1:13 PM, September 19, 2007  
Blogger Christian M. Howell said...

I want someone to post a SMALL proof of AMD's yields.

There own slide says < .05cm^2. I could see if you were saying the yields at 3GHz are bad but you're saying the yields for every clock speed are bad and YOU HAVE NO WAY TO PROVE IT.

I guess a lot of Intel janitors do come here.

1:17 PM, September 19, 2007  
Blogger Andy said...

I want someone to post a SMALL proof of AMD's yields.

There own slide says < .05cm^2. I could see if you were saying the yields at 3GHz are bad but you're saying the yields for every clock speed are bad and YOU HAVE NO WAY TO PROVE IT.

I guess a lot of Intel janitors do come here.


Not learnt what a yield is yet then baron.

Faulty is faulty. Yield has nothing to do with clock speed. Defects break the chip, they don't just break it at certain clock speeds.

For the amount of time you have been around you should really know these things.

6:36 PM, September 19, 2007  
Blogger Giant said...

Wanna bet that as new revs come out at higher speeds AMD will beat Harpertown?

The Inqurier reckons AMD needs at least a 2.8Ghz Barcelona to compete:

Humour - or relief - aside, I believe, based on the Harpertown's initial results, AMD needs a 2.8GHz Barcelona part to have a strong, balanced competitive position against a 3.2GHz Harpertown.

http://www.theinquirer.net/?article=42423

You wouldn't bet one cent. As AMD raises clockspeeds Intel can do the same. They have a LOT more frequency headroom to work this thanks to the 45nm high-k process. I think we all need to remember that Intel is years ahead of anyone else with 45nm high-k.

For yields again I'll this for you:

It may be speculated that AMD is only yielding K10 at a jaw-dropping 30%. 30%!!

Based on AMD's published data, which claimed 0.5 defects density, and coupled with yield management chart, it can be concluded that AMD's K10 are yielding at no more than 35%.

AMD's Data:
http://www.iian.ibeam.com/events/thom001/22876/browser/slides/20070726084721294707/default_large/Slide158.JPG

Yield management by ICEC
http://smithsonianchips.si.edu/ice/cd/CEICM/SECTION3.pdf

If you scroll down to Figure 3-9 in Yield Management, you can see the chart. Follow the 0.5 density defect rate to 280, then....

9:17 PM, September 19, 2007  
Blogger Christian M. Howell said...


You wouldn't bet one cent. As AMD raises clockspeeds Intel can do the same. They have a LOT more frequency headroom to work this thanks to the 45nm high-k process. I think we all need to remember that Intel is years ahead of anyone else with 45nm high-k.



The problem isn't clock speed, it's TDP. Harpertown at 3.2GHz is not going to come in less than 100W.

Even with all of that AMD still OWNS intel in Power\Perf.

If they are so far ahead why is Power 6 running at 5GHz?

6:46 AM, September 20, 2007  
Blogger Andy said...

Power 6 is a totally different kettle of fish. The die is huge and it is only a dual core.

Have you seen the price of it too?

Apple switched to Intel because IBM couldn't deliver enough chips. It's great for apple. They are charging the same but saving a huge amount on the cpu costs.

9:40 AM, September 20, 2007  

Post a Comment

Links to this post:

Create a Link

<< Home