Sunday, August 27, 2006

"Dual core Sempron" frags Pentium XE 955

The Chinese tested Athlon 64 X2 3600+ (2GHZ, 2x256KB) against a Pentium D 915 (2.8GHZ, 2x2MB). The X2 3600+ frags Pentium D 915 by 30% in most tests (39% in HL2, 29.8% in Doom3, 31% in FarCry). For Pentium D to compete against the "dual core Sempron", it needs a clockspeed of at least 2.8GHZ*1.3 = 3.64GHZ.

Pentium XE 955 is only 3.46GHZ, so it will be fragged by X2 3600+ with a good margin.

95% of Intel's CPUs are below Pentium XE 955, which is priced at $999. Therefore, the dual core Sempron frags 95% of Intel's production, at a price of $125.

A lot of readers have asked me to ban those pure insulting comments. I think it's a good measure to make this place more enjoyable. So, from now on, unless a comment makes some prima facie showing of IQ, it will be disallowed.

48 Comments:

Anonymous Anonymous said...

Meanwhile, enthusiasts with Conroes overclocked to 3GHz+ on stock voltage are 'fragging' all AMD pretenders.

Hey Sharikou, what's this I hear with AMD having troubles with 65nm?

http://www.theinquirer.net/default.aspx?article=33964

TheINQ also thinks K8L will be delayed to Q1 2008. Hmmm, it is a little odd you don't mention either articles, considering you post a LOT of TheINQ material - when it favors your arguments towards AMD, of course. ;)

9:59 AM, August 27, 2006  
Anonymous Anonymous said...

While its true what you say, you have to remeber that the pentium 955 xe have a higher fsb of 1066 mhz. This would mean that the pentium 955xe should get close or perhaps beat the x2 3600 with a very small margin, since the fsb on a pentium d is one of the largest things holding it back. But when you see at price and the enourmus heat output you would need to be stupid to buy a 955xe.

So the X2 3600 will be a great successor and will probably force Intel to dump prices even more in order to sell of it inventory.Because of the price i guess this chip also would give the lower end conroes a good competition.

I think we will se a lot of X2 3600 in oem machines.This chip will sure be popular:)

10:14 AM, August 27, 2006  
Anonymous Anonymous said...

The images show x2 3600 overclocked from 2.0 to 2.8 ghz

When compared with 2.8 ghz pentium, it's clearly superior. But everyone already knows that.

At 2.0 ghz the chip is sometimes behind 2.8 Pentium.

11:08 AM, August 27, 2006  
Anonymous Anonymous said...

Good decision on filtering out useless & degrading comments. I think it's an absolute waste to allow those comments through. I don't agree with everything you write but I do enjoy your blog.

Ed

12:07 PM, August 27, 2006  
Anonymous bgt said...

Sharikou, thats a very good id to bann those people who can only insult. It was a nussance to read the comments. Keep up the good work.

12:08 PM, August 27, 2006  
Anonymous Anonymous said...

A lot of readers have asked me to ban those pure insulting comments. I think it's a good measure to make this place more enjoyable. So, from now on, unless a comment makes some prima facie showing of IQ, it will be disallowed.

If I may be so bold as to ask, will that include insulting comments by YOU, or will this be a one monkey show?

12:52 PM, August 27, 2006  
Blogger "Mad Mod" Mike said...

I think your analysis's (is that spelled right?) are great and give us the idea of what the truth is. Doing this break down shows just how effective AMD is with Cache and how even something w/ 1/8 the cache of a competitor, it can make performance exceed.

I've always said I wonder how good an AMD64 CPU would perform with only 64KB of L2 or none at all. AMD could make a low-voltage, 64KB L2 cached sempron and use it to compete with ULV Pentium M's for developing markets, the cost would be about $10 for a CPU, and many people would gobble those up like crazy.

It looks like those Chinese folks have taken the 3600+ to 2.8GHz, and at those clock speeds, not a single Pentium processor could touch it, even @ 4.6GHz.

4:26 PM, August 27, 2006  
Anonymous onef'dupdude said...

The 915 only has an 800MHz FSB, while the 955 has a 1066MHz FSB which in my experience makes a huge difference on the netburst cpu's.

However, the 3600 will still win a few benchmarks versus a 955.

4:27 PM, August 27, 2006  
Anonymous Wirmish said...

"So, from now on, unless a comment makes some prima facie showing of IQ, it will be disallowed."

It's like a breath of fresh air...

4:42 PM, August 27, 2006  
Anonymous Anonymous said...

This is true. Pentium D's are no contest to any dual core AMD has ever made. Why AMD was better for 3 years and still is. Intel can not compete in a 64-bit arena agenst AMD as well. Netburst is long dead!

4:50 PM, August 27, 2006  
Anonymous enumae said...

Back in the day, your claimed...

"AMD's 90nm X2 3800+ frags Intel's 65nm Presler Pentium XE965."

Saturday, July 29, 2006

Now you now have a new post in which you say the...

"The X2 3600+ frags Pentium D 915"

It would seem you are stuck on X2 beating P4, as this is widely accepted by people who understand computers it is not by the average consumer, that is where marketing takes over.

As we all know Intels is a huge marketing giant, and since most will never know this since AMD doesn't have a mainstream marketing voice, it would seem null and void.

General consumers will not know what they are getting, as they usually only see price, and as you may have noticed, there is liitle to no mention of C2D in any of Intels advertising, and probably won't be until inventory is cleared.

We all know that Intel is ramping, and it will take some time, but if Intel clears its inventory, AMD may be in trouble if an enthusiast is looking at price/performance.

Looking at benchmarks, it clearly shows the X2 5000+ as the closest competition to the E6400, except a few benchmarks which show advantages for AMD and Intel, performance/price wise a $242.00 Intel chip is equal to a $759.00 AMD chip.

Also while online I have noticed that AMD's 65nm parts only go up to 2.4GHz, so I am guessing it will be a little while before we see the 25% clock increase you speculated.

I have a question for everyone...

What happens when Intels line up consist of all C2D processors, and Netburst is gone, and if K8L will not be here in 2007 (as reported earlier this week), what is AMD going to do in 2007?

Thanks to anyone who answers.

5:05 PM, August 27, 2006  
Anonymous Anonymous said...

This is no secret. Athlon is faster than Pentium.

However, Core 2 Duo is faster than Athlon.

So, while it is true that AMD is offering more performance than Intel for many desktop market segments, this will reverse when Core 2 Duo becomes the bulk of Intel's volume.

AMD's small 256K L2 cache does a good job of feeding AMD's simpler execution model. Intel's large 2MB L2 cache does not do a very good job of feeding Intel's complex 30+ stage execution engine.

As for the AMD's X2 with 256K cache per core, it shows that AMD's design does not utilize their L2 cache very well (beyond 256K) as models with 512K and 1MB cache do not show any great performance improvements vs. 256K.

And it also shows that Intel does not utilize their L2 cache well either as 2MB L2 doesn't provide much of a performance boost over 1MB L2 for Intel.

What really may be going on is that all modern operating systems, compilers, and applications are not evolved enough to give CPU vendors good L2 performance.

Or perhaps the various OS companies and apps companies just don't do the rather expensive and time consuming tuning that is required to use the L2 cache most effectively.

It will be interesting to see what performance boosts are to be found with AMD's on-chip L3 cache that will be part of a forthcoming processor.

Intel may counter with a large off-chip L3 cache as they have done with various Xeon models in the past. The cost will obviously be higher, as will the L3 latency, but the performance may be reasonable.

So far 2006 looks like a good technical year for Intel with Conroe and Woodcrest, and perhaps Cloverleaf in 4Q06, but 2007 and 2008 will be big years for AMD.

6:13 PM, August 27, 2006  
Anonymous Anonymous said...

The Athlon 64 X2 3600+ (2GHZ, 2x256KB) chip is the best bang/buck processor out there. Considering this is where the volume market is, Intel is probably very worried; they just don't have a high enough yield on Conroe.

6:19 PM, August 27, 2006  
Blogger Tony said...

The reason he is comparing to pentium D models is because of the huge surplus of pentium D chips. Several billion dollars worth of inventory and growing. What he is showing here is that this inexpensive, widely available processor obsoletes all of the pentium D stock. He is not comparing to Core 2 Duo because of the price point difference.

At the end of the day, Intel has tons of pentium D's which they shouldn't be able to clear because of this single chip.

12:07 AM, August 28, 2006  
Anonymous Anonymous said...

"A lot of readers have asked me to ban those pure insulting comments. I think it's a good measure to make this place more enjoyable. So, from now on, unless a comment makes some prima facie showing of IQ, it will be disallowed."

What makes you better than dictators who censor any opinions against him? I think you are using this reason to not allow the opinions who propose a stronger argument against yours..

If this blog is a one party argument, you will lose a lot of bloggers over time and pretty soon, your 80 comments will decrease to 15..

Why can't you be more like rahul? He makes sense and intelligent people voice their opinions.. Do you notice they do not use insulting remarks at his remarks??? He has credibility and he does not even claim he is a Phd.

12:57 AM, August 28, 2006  
Anonymous Anonymous said...

"The Athlon 64 X2 3600+ (2GHZ, 2x256KB) chip is the best bang/buck processor out there. Considering this is where the volume market is, Intel is probably very worried; they just don't have a high enough yield on Conroe."

Intel is indeed worried about AMD. Every day AMD's manufacturing capability increases and Intel worries more.

With AMD on the rise, it means Intel has to get back to what made them Intel. Execution. Good design. Testing.

It will take Intel years and they may not make it. For all the people that built Intel are gone and Intel's culture has a bias against smart people.

History may show that Intel had their time of glory, got fat, arrogant, and sloppy, and yielded their position of dominance to AMD.

And of course only time will tell what really happens.

As for the "best bang for the buck" entry level dual core processor, that is probably a shootout between the x2 3800 (which costs the same as the x2 3600 now -- they are both $152 at Newegg) and the Conroe E6300 (at $193).

If it were my money, I'd spend a bit more for the E6400 ($243) and get performance that is roughly comparable to an FX-62 ($827). In the real "bang for the buck" contest, this chip, the E6400, is the winner.

While that is nearly $100 more than an entry level dual core, it is a great price/performance point. The Conroe will also run cooler and quieter than the FX-62. Never mind that you also get a motherboard and RAM for the price of the AMD chip...

1:56 AM, August 28, 2006  
Anonymous Anonymous said...

"I've always said I wonder how good an AMD64 CPU would perform with only 64KB of L2 or none at all. AMD could make a low-voltage, 64KB L2 cached sempron and use it to compete with ULV Pentium M's for developing markets, the cost would be about $10 for a CPU, and many people would gobble those up like crazy."

Yes, AMD with the IMC could do wonders. I think that removing all the L2 and its logic, going completely direct could be faster and even improve Ghz. (AMD as exclusive L1 and L2, I’m assuming there are wasted cycles here).
Also I don’t know if amd could do the shared cache very easily by going off the processor into the IMC. IF the IMC controls the cache than there was no need of dedicated cache on each processor.

3:04 AM, August 28, 2006  
Anonymous Anonymous said...

"What happens when Intels line up consist of all C2D processors, and Netburst is gone, and if K8L will not be here in 2007 (as reported earlier this week), what is AMD going to do in 2007?"

... the current AMD X2 would be dirt cheap, maybe around $70. I would still buys AMD X2 because it will be dirt cheap.

The old PentiumD would be $30 or less, mainly sold to the 3rd world country only, or as land fills.

3:10 AM, August 28, 2006  
Anonymous Anonymous said...

Here is another benchmark from hkepc as well.
http://www.hkepc.com/bbs/hwdb.php?tid=648852&tp=amd-x2-3600-2&rid=651569
Please note that its seems the X2 3600 use 1.200 volt and are delivered in a 65 watt power envelope:)

I have a question for you sharikou,
Do you know when the x2 3600 will be released?

3:37 AM, August 28, 2006  
Anonymous Anonymous said...

What makes you better than dictators who censor any opinions against him? I think you are using this reason to not allow the opinions who propose a stronger argument against yours..

In case you didn't notice, this is Sharikous blog, and noone forces you to visit it, so i don't understand why wou compare him with a dictator.

If this blog is a one party argument, you will lose a lot of bloggers over time and pretty soon, your 80 comments will decrease to 15..

I rather read 15 smart posts that 80 posts full of insults.

Why can't you be more like rahul? He makes sense and intelligent people voice their opinions.. Do you notice they do not use insulting remarks at his remarks??? He has credibility and he does not even claim he is a Phd.

Again, you are not forced to visit this blog. Don't like it -> don't read it. I personally enjoy reading it every day, and i'm very happy with his decision of not letting through all those empty and senseless posts.

4:38 AM, August 28, 2006  
Blogger "Mad Mod" Mike said...

"With AMD on the rise, it means Intel has to get back to what made them Intel. Execution. Good design. Testing."

What made them Intel was screwing over AMD in 1990, otherwise AMD would be 5x bigger than they are today.

6:42 AM, August 28, 2006  
Blogger Sharikou, Ph. D said...

What makes you better than dictators who censor any opinions against him?

In some states, such as Virginia, using dirty language in public is a misdemeanor. Freedom of speech doesn't mean you are free to say anything you want. Here, what we are trying to do is make the posts stay on topic. Pure insults, is, by definition, off topic.

8:18 AM, August 28, 2006  
Anonymous SaintGreg said...

"With AMD on the rise, it means Intel has to get back to what made them Intel. Execution. Good design. Testing."

I would really say that invention made Intel what they were. The 8088 was the bomb in its day. They need to continue their ceased innovation where AMD has left off if they want to really get back in the game down the road.

9:56 AM, August 28, 2006  
Anonymous Anonymous said...

"In some states, such as Virginia, using dirty language in public is a misdemeanor. Freedom of speech doesn't mean you are free to say anything you want. Here, what we are trying to do is make the posts stay on topic. Pure insults, is, by definition, off topic."

If you offer a balanced scientific non biased view, you won't get insults!

10:06 AM, August 28, 2006  
Anonymous Anonymous said...

Sharikou,
I'm not sure why this is even news? Are you just trying to stir up some positive AMD news in the face of the strong wave of Core 2 Duo momentum??

10:33 AM, August 28, 2006  
Blogger Sharikou, Ph. D said...

TheINQ also thinks K8L will be delayed to Q1 2008. Hmmm, it is a little odd you don't mention either articles, considering you post a LOT of TheINQ material

INQ folks sometimes are good, they dig out some good stuff -- mostly insider info. But in most cases which require some analysis, they are clueless. You have to be able to analyse their articles and found the credible ones. For instance, when they reported the AMD-DELL contract, I knew it was real -- it gace enough supporting information. When INQ argued that DELL wouldn't go AMD, I knew it was wrong. Their analysis was based on wrong data. When they reported that Woodcrest were crapping out everywhere, I knew it was right. It named companies and customers...

10:42 AM, August 28, 2006  
Blogger Sharikou, Ph. D said...

We all know that Intel is ramping, and it will take some time, but if Intel clears its inventory, AMD may be in trouble if an enthusiast is looking at price/performance.

That's a big IF. Intel has inventory of $8 billion legacy chips, and is producing more and more everyday. You have to understand 50% of Intel's production is still at 90nm. There is next zero chance that Intel can reach 50% conroe before K8L is out.

Conroe was prematurely delivered, the result will be fatal. Watch Intel BK.

10:52 AM, August 28, 2006  
Anonymous enumae said...

The Inq is not the only one to have said K8L is coming out in 2008...

Here are two links from Dailytech.

1. Gearing Up For AMD Revision G

2. AMD K8L: Quad Core Desktops And So Much More

Both of which were posted in May of this year.

10:53 AM, August 28, 2006  
Blogger Sharikou, Ph. D said...

The Inq is not the only one to have said K8L is coming out in 2008.

AMD said in its June 2006 tech analyst meeting the quadcore with HT3 and L3 will be launched in mid2007. There was no ambiguity there.

In a recent email to investors, AMD stated that there is no such thing as K8L. The name was created by media.

10:59 AM, August 28, 2006  
Blogger pointer said...

In a recent email to investors, AMD stated that there is no such thing as K8L. The name was created by media.

as the link you provided about the spy consipiracy ... the name was created by intel :). Even if there is no such thing, i seriously think that quite some AMD emplyee there was such thing in their company :)

btw, now you say there is no such thing, so, what about your previous consipiracy theory? :)

11:13 AM, August 28, 2006  
Anonymous Anonymous said...

"Conroe was prematurely delivered, the result will be fatal. Watch Intel BK."

Haven't you ever considered it odd that virtually no-one else is making this claim? If one of America's biggest companies is, as you say, not more than two years away from a bankruptcy which nobody and nothing can prevent, wouldn't it be a huge story in the media right now? Wouldn't the business journals be competing to write the best editorials about perhaps the biggest destruction of a company in the USA's history? Wouldn't the financial analysts be kicking themselves wondering why no-one saw this coming? Wouldn't the government be rallying around trying to find some way to prevent Intel from going down totally? Wouldn't the likes of IBM, Samsung and TI be in negotiations to buy Intel's soon-to-be-disused FABs and designs? Wouldn't Intel's prices be in the gutter as wholesalers tried to rid themselves of soon-to-be-useless merchandise? (And I'm talking REAL low prices, like a Core2 X6800 for $200 or so) Wouldn't the DoJ be trying to pin the blame for Intel's imminent demise on AMD, so that they can prevent a total AMD monopoly and actually look useful for a change?

11:13 AM, August 28, 2006  
Anonymous enumae said...

K8L name was made up, that is understood, but is K8L revision H?

If so, here is a quote from HKEPC...

"The first 4 cores K8L product, Deerhound, which is for Server use, will first appear in the second part of 2007. And then we have to wait till the upper part of 2008 for the Desktop use version, Greyhound, workstation version, Cadiz, and the other server user version with larger L3, Zamora."

Maybe I am misunderstanding things, but I had thought that this discussion was based on desktop K8L (Rev H), not server chips, since everyone has been saying Conroe is in trouble in the middle of 2007, which as of the recent news, seems they are not.

Please clarify.

Thanks

11:29 AM, August 28, 2006  
Blogger Sharikou, Ph. D said...

btw, now you say there is no such thing, so, what about your previous consipiracy theory? :)

That would indicate that the problem was even more severe. I heard that Intel has a group dedicated to studying AMD. It seems Intel has even created an internal naming system for AMD's future products, so they could have a common vocabulary when talking about AMD's next gen.

11:56 AM, August 28, 2006  
Blogger Sharikou, Ph. D said...

Maybe I am misunderstanding things, but I had thought that this discussion was based on desktop K8L (Rev H), not server chips, since everyone has been saying Conroe is in trouble in the middle of 2007, which as of the recent news, seems they are not.

Please go to www.amd.com and read Phil Hester's June 1 slides. both desktop and server will be moved to next gen core mid 2007.

12:07 PM, August 28, 2006  
Anonymous Anonymous said...

I heard that Intel has a group dedicated to studying AMD.

So what? Any company that doesn't spend time evaluating their competition, and their present and future threats would not only be foolish and naive, but negligent in protecting the shareholders' interests. Do you mean to tell us that AMD does not have a similar group studying Intel? Hell, do you think General Mills has a group studying Kellogg's and vice versa, or Pepsi and Coke, or are they all just whistling Dixie and trotting merrily down independent paths to success or disaster?

Thanks anyway for pointing out the obvious.

12:19 PM, August 28, 2006  
Anonymous enumae said...

I am not trying to nit pick, but the slide does not mention L3 cache for desktop, they mention it on servers and workstation.

Is that a feature for server and workstation only?

12:29 PM, August 28, 2006  
Anonymous Anonymous said...

Too late. You've already lost the respect of the technical community with your Dell 'dual explosions' supposition, and your response to those not agreeing with you as 'retards'.

Not Ph.D worthy on both counts, and not worth visiting here except for some free entertainment - most notibly through you.

How's my IQ?

2:19 PM, August 28, 2006  
Anonymous Graham said...

Some anonymous coward wrote:

I personally enjoy reading it every day, and i'm very happy with his decision of not letting through all those empty and senseless posts.

This new policy doesn't seem to be working... he let your post through.
Seriously though, a dual core sempron fragging a Pentium XE 955 is not going to matter as Pentium D will be dead sometime early next year. As in, not manufactured any longer. Intel will switch totally to C2D by Q1'07 and AMD will not be able to counter it until maybe the end of 2007 so contrary to an earlier poster, 2007 will be Intel's year on the desktop, mobile and 2S servers. Also, as appears to be the case from the Inquirer, AMD is not executing flawlessly on the 65nm transition as it needs to do so maybe 2008 won't be AMDs year either. They made hay while the sun shined but did they make enough to survive the winter? I don't think so. I'm predicting that AMD will BK in 5 quarters. Buh-bye AMD!

4:26 PM, August 28, 2006  
Anonymous N4CR @ work said...

http://www.theinquirer.net/default.aspx?article=33964

I remember hearing the same things about 90nm AMD/IBM SOI tech too... look where it is now.

5:53 PM, August 28, 2006  
Anonymous Anonymous said...

"I remember hearing the same things about 90nm AMD/IBM SOI tech too... look where it is now."

Glad you can point out that AMD by itself can not manufacture/design any good chips.. They really need IBM to breastfeed them..

6:28 PM, August 28, 2006  
Anonymous Anonymous said...

"Pure insults, is, by definition, off topic."

Do these isults include calling Intel Natzis and woodcrap...?

6:30 PM, August 28, 2006  
Anonymous Anonymous said...

What is wrong the Pretenders claim here?

He is comparing the entry level net bust processor from INTEL against AMD's best. There is nothing new here.

Ask why he isn't comparing the new Core2 against AMD's latest best?


A real PhD

6:55 PM, August 28, 2006  
Anonymous apollo said...

"Do these isults include calling Intel Natzis and woodcrap...?"

Sharikou is a New World Order stooge.

He attacks Intel for making a $300 million investment in Vietnam while he defends AMD's $3.2 BILLION investment in the home of the world murderers and global rapists.

The only thing to do is leave this blog to die. Otherwise, we are just giving this stooge a stage to spread his filthy lies.

1:27 AM, August 29, 2006  
Blogger N4CR said...

AMD said in its June 2006 tech analyst meeting the quadcore with HT3 and L3 will be launched in mid2007.

My mate (who ain't into computers as much as me) just confirmed that today, here in New Zealand. He was at an AMD tech meeting down under.. good to hear. I was very envious, pretty keen to get hold of the 3 hour slideshow and dissect it eh... my mate said it would come out and be released on AMD rewards or something like that, soon.

He also mentioned 14W or 5W (not sure of which one) opterons??? What the heck?

Also said it had benchmarks beating intel decently, not just fractions and small %ages but alot. But he is a bit foggy about what it was exactly they were comparing. Said one was only 7% from memory so I'm guessing either 'K8L' ain't too great or else it was some other 4 way step between now and 'K8L'.

I'm working on getting hold of the slides.. we will see. Very keen to!

Comments on this anyone.. correctness or what they might have heard?

5:38 AM, August 29, 2006  
Anonymous Anonymous said...

To the poster of the Dailytech links, you forgot the more recent one:

AMD Plans To Demonstrate Native Quad-Core This Year
Native quad-core en route
Jul 21, 2006 6:57 PM

http://www.dailytech.com/article.aspx?newsid=3448

"Earlier this year, AMD's Executive Vice President Henri Richard claimed this native-quad core processor would be called K8L.

Earlier AMD roadmaps have revealed that quad-core production CPUs would not utilize a native quad-core design until late 2007 or 2008."

6:14 AM, August 29, 2006  
Anonymous Your Daddy said...

Graham the Intel mega fanboy wrote:

"This new policy doesn't seem to be working... he let your post through."

Wow, he let your's through too, imagine that!

"Seriously though, a dual core sempron fragging a Pentium XE 955 is not going to matter as Pentium D will be dead sometime early next year."

Ummm how long do we have until early next year? What's that? 6 months??? Alot can happen in 6 months in the computer industry bub, especially with Intel trying to clear out mountains of outdated Pentium 4s that the 3600+ X2 makes obsolete and overpriced.

"As in, not manufactured any longer. Intel will switch totally to C2D by Q1'07 and AMD will not be able to counter it until maybe the end of 2007"

Haha, totally by Q1'07? We'll see. Right now, they can't keep enough Core2s to meet market demand, I suspect they will have to rely on Netburst a little while longer than they would like. Seems to me they aren't having the easiest time ramping Core 2.(i.e. hard to find high end Core 2 models and supply is spotty at times)

Once K8L is out in 1H 07' I suspect AMD will do a hard launch, AMD has learned from past mistakes and watched as Intel continually makes paper launches or botched hard launches. They also don't have much of a choice as Core 2 will provide stiff competition.(when they get out on the market in sufficent volume, i.e. all market segments)


"..so contrary to an earlier poster, 2007 will be Intel's year on the desktop, mobile and 2S servers."

Right up until AMD releases the K8L variant on desktop, Bulldozer on mobile with ATI's new mobile chipset and servers will be locked up from entry (thanks to 4x4) all the way up to enterprise levels.(4 way and up)

"Also, as appears to be the case from the Inquirer, AMD is not executing flawlessly on the 65nm transition as it needs to do so maybe 2008 won't be AMDs year either."

So they can't make the frequencies at the target voltage they wanted, big deal. 1.4v is still decent all things considered(my 3200+ Venice runs very cool, cool enough to work as a DTR for a laptop!) to get the frequencies they need to achieve. As long as they can keep parity or surpass Core2 in overall performance, it's not a problem. It will be resolved in time and they still have time until the actual chips are in production to solve the issue. We'll wait and see what ACTUALLY happens until then.

FWIW, in 2008 AMD will release it's TRUE next-gen chip, redesigned from the ground up.(K10 I believe) AMD reps are saying a 150% increase in it's performance over the current K8, so yea, 2008 IS going to rock for AMD.


"They made hay while the sun shined but did they make enough to survive the winter? I don't think so. I'm predicting that AMD will BK in 5 quarters. Buh-bye AMD!"

For a company that learned how to survive for about almost 2 years in the red is going to go BK in 5 quarters? HAHAHAHAHAHA!!!! That's rich man, tell me another joke, I need a good laugh this morning.

Seriously, as bold and far-fetched as Sharikou's claims are I do feel Intel is going to get hurt far more than AMD will with this price war they have initiated. AMD knows how to survive in a very hostile market, Intel only knows how to survive with it's monoplistic market stranglehold. While Intel may not go BK as Sharikou keeps saying,(I think the US Goverment will step in before that happens) they surely will become a mere shadow of it's former size and glory when the legal dust settles. The legal pressure AMD has put on Intel with the anti-trust suit is very evident and already has had great effect on AMD's recent successes. Now that Intel has no choice but to play fair, they don't know how because their monoplistic market mindset is so ingrained in that company from the execs at top on down to their engineers.

In short Intel sucks at a game it doesn't know how to play... an open and fair market. AMD's Hypertransport Consortium and various other organizations that cater to open standards and cooperation from partners in the industry is only going to bolster thier position in the market. Intel on the other hand is biting it's own hand that feeds it by alienating these same partners, closing it's architechture from it's partners and it will come back to haunt(hurt) them.

Well, that's enough "Graham's distorted reality" dispelled for one day, BUHBYE!

7:45 AM, August 29, 2006  
Blogger N4CR said...

So 'K8L' is rev g or h? Early next year? They are stepping from e-f now.. so to get to h means there is an intermediate release and not K8L... but some other part. Maybe a quad core current tech K8? Would they release another one between them? Usually ain't too soon. So it should be rev g.. any ideas?

I'm just trying to figure out if it's a K8L or not being released or just an intermediate refresh.

2:46 PM, August 29, 2006  
Anonymous Edward said...

"So 'K8L' is rev g or h? Early next year?"

Well... people have different info about K8L, because they have different definitions of K8L. A better question is, what is different between rev. G and H, and how is each different from rev. F?

First, on the term K8L -

We don't even know where the term K8L came from. Initially it seems just a rumor overheard by Charlie at the Inquirer, who later claims he heard it from Intel. Well, Charlie shamelessly contradicts himself in so many things that I find it extremely difficult to put any credibility on his words.

In any rate, it is safe to say that AMD did not publicly spoken of 'K8L', and if the term was actually used by AMD it's only used internally by a limited group of people.

Second, on the revisions -

It seems that the current rev. F will be the last single-focus core design. AMD revealed in its June 2006 Tech Analyst Day that it is going to diversify design directions to server, desktop, and mobile.

If you look at rev. F and previous K8 cores, Opteron, Athlon and Turion are almost identical in terms of the core design. There is really not much difference between a rev. E Opteron from a rev. E Turion, except one has faster/more HT links and one consumes less power.

Thus for previous K8's, the revision number across all types of processors is well-defined. This however is not going to be true for future AMD processors, whose server, desktop, and mobile variants are going to diverge. My guess is this divergence will start at rev. G and become the norm at rev. H.

Third, on quad core -

We now focus on the server/desktop line of future K8. It is probably safe to say that AMD's native & improved quad-core K8 in mid-2007 is rev. H. It's because according to AMD "internal rev. G samples were generated and production ready" by June 2006, on 65nm, which is scheduled for production shipment in 4Q 2006.

While rev. G won't bring to us the core improvements mentioned in the Tech Analyst Day, it will likely be on 65nm and get some performance improvement.

Another question is, will rev. G be quad-core as well? I don't think likely given how AMD talked about 4x4, nor would it bring much benefit to AMD due to quad-core's small market. But it's still possible if the earlier "quad core tape out" rumor was true and if Intel hypes its own MCM quad core well.

Conclusion -

1) K8L is just a stupid term.

2) Diverging server/desktop/mobile core design after rev. F.

3) For server/desktop, rev. H brings improved core + native quad core on 65nm in mid-2007.

4) Rev. G by the end of 2006 on 65nm with fewer core improvement, and likely dual-core only.

5) Innovative mobile core design???

1:45 AM, August 30, 2006  

Post a Comment

Links to this post:

Create a Link

<< Home