Monday, April 23, 2007

C2D's pony trick rediscovered

Cache is pretty much the only trick --- I told you that long long time ago, and told you again recently. Also, Core 2 temperatures are danerously high for normal operation. Scientia explained why Intel hasn't released anything faster.

The Israeli dudes aren't so smart after all, certainly no match to the grand masters at AMD, whose K10 design will frag Intel to death.

30 Comments:

Blogger rathor said...

K10 at 2.5 GHz beats Intel quad at 3.0+
------------------------------
Intel in trouble, code red !!!
------------------------------

The new K10 based processors will really make Intel run for its money at least till the end of the year. Just as Intel stole K8's mojo with Merom aka Core 2 Duo and quad, AMD is ready to finally fight back. It took them almost a year to prepare for the big clash.

Barcelona / Agena FX will beat Intel Core 2 Quad at 3.0 GHz+. Intel knows it and it is scared. Intel also knows that it can not go over 3.0 GHz with 65 nanometre at least not at 120 W TDP. Some reports already claims that QX6850 with FSB 1333 and 3 GHZ clock already pushing the envelope as intel's 120W TDP is actually 150W+ at least on AMD's TDP scale. TDP stands for Thermal Design Power.

Intel needs to go over 3.0 GHZ to 3.2 or more to match Barcelona / Agena parts at 2.5 GHz. Intel knows that it needs to go to 3.2+ GHz and it needs 45 nm Yorkfield core to match Agena 2.5 GHz.

Finally AMD should be able to make 2.7 if not even more with its 65 nm process. Intel has a good fighting chance with next generation Nehalem core in second half of 2008.

1:02 AM, April 23, 2007  
Blogger KingRichard said...

This comment has been removed by the author.

7:02 AM, April 23, 2007  
Blogger KingRichard said...

Now we know why Intel is releasing all kinds of useless Quad cores with massive price dumps...

Anything to hold off the Slaughter from AMD...

At least it will be a great Christmas for me!

7:04 AM, April 23, 2007  
Blogger R said...

“The Israeli dudes aren't so smart after all, certainly no match to the grand masters at AMD, whose K10 design will frag Intel to death.”

I would argue the Israeli engineers are very smart for coming up with a quick and easy fix that gave Intel a temporary lead and caused AMD to lose millions this past year.

7:26 AM, April 23, 2007  
Blogger Bubba said...

K8 is already worthless. Buy a flat panel duisplay, and Dell will through in a dual core AMD pc for FREE.

http://www.dell.com/content/topics/segtopic.aspx/odg/msn_module_jump?c=us&l=en&s=bsd&cs=04

7:35 AM, April 23, 2007  
Blogger Tanrack said...

rathor, source for all your claims?

7:39 AM, April 23, 2007  
Blogger R said...

“Tanrack said...
rathor, source for all your claims? “

The 3 ghz thermal ceiling is well documented for AMD as well, thus we now have more cores per CPU

8:10 AM, April 23, 2007  
Blogger PENIX said...

The P3 can't scale, which is why Intel moved to the worthless Netburst. Moving to 45nm is just another patch on a fundementally flawed platform. Intel needs to throw out all their junk and just start over from scratch.

8:20 AM, April 23, 2007  
Blogger middle_of_the_road said...

Sharikou said :-
Cache is pretty much the only trick --- I told you that long long time ago, and told you again recently.


You also said "4x4 will start well below $1000 and become a main stream platform, which will permanently pin Intel at half the performance.", so as you can imagine I'll take most (if not all) your claims with a large pinch of salt.

I suspect that Intel had to do a little more than just bolt cache onto a cpu to beat K8. K8 was a superb processor and even now still has it's place in the market.

8:27 AM, April 23, 2007  
Blogger netrama said...

"I suspect that Intel had to do a little more than just bolt cache onto a cpu to beat K8 "

Yes ofcourse very much , bribing Anand,Tom and other button pushing a**kissers. Even the dudes who work at Bestbuy and CC.
How did all this help - Lead people to the perception that Core2 is the best sh*t they can buy..
Wait did I forget those Full page Intel logo ads mercilessly printed on any magazine you would pick at a newsstand , be it Forbes or a woman's gossip magazine

9:06 AM, April 23, 2007  
Blogger Ho Ho said...

penix
"The P3 can't scale"

Perhaps so but it still scales better than K8 does. Also it seems as ancient P3 is beating the hell out of superadvanced K8.

9:19 AM, April 23, 2007  
Blogger Bubba said...

Wait did I forget those Full page Intel logo ads mercilessly printed on any magazine you would pick at a newsstand , be it Forbes or a woman's gossip magazine

Too bad AMD is so horrible at marketing. Maybe then they wouldn't be losing billions of dollars a year.

9:22 AM, April 23, 2007  
Blogger Poke said...

This comment has been removed by the author.

9:43 AM, April 23, 2007  
Blogger Giant said...

AMD is finished. Intel has ensured that AMD will bleed to death. Due to AMD's utter stupidity in buying Ati they will have a tough time raising any more cash. They have to raise double what they actually need since Morgan Stanley gets half. AMD's BK is inevitable.

Thanks to the revolutionary high-k process (Which is the biggest advance in transistors in the last forty or so years!) Intel can clock Penryn at super-high rates with no problems.

Re: Sharikou's stupid comments that C2Ds run hot, nothing could be further from the truth. I run my E6600 with a Zalman fan that you can barely hear and it idles at below 30C, and just reaches 43C at a full load with both cores. This is is stark contrast to AMD's 125W spaceheater 6000+ 90nm junkyard technology.

Penryn will increase Core 2's IPC lead of 20% by a minimum of 10%, in some cases the amount can be greatly higher. This, coupled with the ability to clock Penryn at huge clockspeeds, will ensure that AMD is finished once and for all.

What were those financial results again? AMD losing 600 million dollars?! Only got 1.2bn in the bank. AMD is out of cash in six months and then they will BK. They can't even sell anything big like a FAB to make money. Morgan Stanley would get all the money!

P.S. Some companies are rumored to be looking at buying AMD. http://www.theinquirer.net/default.aspx?article=39110
My advice is don't bother. AMD will BK and cost the company that acquired AMD billions of dollars in losses.

9:43 AM, April 23, 2007  
Blogger Randy Allen said...

Hah. The reason Intel hasn't released any faster dual core parts is because it doesn't need to. A 2.4Ghz is more than enough to frag AMD's 6000+ spaceheater garbage.

http://www.xbitlabs.com/images/cpu/dualcore-roundup/charts/average.png

Scientia is an AMD fan alright, he only just falls short of the standards set by Sharikou and Penix. Intel took their 5160 Woodcrest 3Ghz and doubled it: Quad Core 3Ghz. This presents far, far more of a thermal challenge than increasing the clock speed on a dual core by 333mhz. TechReport tested the world's fastest desktop processor, the Core 2 Extreme QX6800. In their power consumption tests the QX6800, under load, used less power than an AMD 6000+ and only two more watts than the previous champion, the QX6700.

http://techreport.com/reviews/2007q2/core2-qx6800/cine-power-peak.gif

Clearly Intel has plenty of headroom in it's 65nm process.

9:55 AM, April 23, 2007  
Blogger Roborat, Ph. D. said...

Sharikou said: "Core 2 temperatures are dangerously high"

Another ignorant comment linking Core2Duo absolute thermal limits with Core2Duo's DESIGN PACKAGE limits.
While the former is about the silicon's junction temperature, the latter is about the temperature measured at the heat spreader, which is a function of the heat spreader and thermal interface material design, but NOT the CPU's limitations. Intel can easily change the heat spreader if it wanted to. Any Package Design Engineer knows this. At least i now know you two aren't one.

9:58 AM, April 23, 2007  
Blogger middle_of_the_road said...

"Yes ofcourse very much , bribing Anand,Tom and other button pushing a**kissers. Even the dudes who work at Bestbuy and CC."

Of course :-

the 14 stage pipe line (down from 30 in the P4)
four-wide architecture (compared to 3 in P4)
Micro Ops
SSE engine improvements

have nothing to do with the performance improvement compared to Intel's previous generation.


Next you'll be arguing that Intel pushed AMD over and stole it's dinner money.

9:59 AM, April 23, 2007  
Blogger Tim said...

Lol. Are you that desperate that you'd repeat the same ignorant garbage over and over...you must be.

10:10 AM, April 23, 2007  
Blogger tech4life said...

Randy Allen said

...AMD's 6000+ spaceheater garbage.


Every time you say "spaceheater garbage" my mind automatically associates that with Pentium 4 and Pentium D. What a shameful 4+ years for Intel. And now that they've won the performance crown back finally they're just going to lose it again after a year.

10:59 AM, April 23, 2007  
Blogger Tim said...

Given AMD's release schedule - I seriously doubt Intel will lose anything anytime soon.

11:02 AM, April 23, 2007  
Blogger tech4life said...

tim said

Given AMD's release schedule - I seriously doubt Intel will lose anything anytime soon.

I guess we'll see in a few weeks when the Tunisian NDA's expire. Either AMD was lying about the 40-50% performance advantage over C2D or they were telling the truth. With Intel's recent price drops for C2D it seems that THEY believe AMD's claims.

11:34 AM, April 23, 2007  
Blogger rathor said...

@all Intel fans:

http://www.amd.com/us-en/Corporate/VirtualPressRoom/0,,51_104_543~117115,00.html

PS: More?

11:57 AM, April 23, 2007  
Blogger Sal said...

tech4life:

Price cuts don't just happen all of a sudden. They are planned in Quarters ahead.

12:02 PM, April 23, 2007  
Blogger tech4life said...

Sal said

Price cuts don't just happen all of a sudden. They are planned in Quarters ahead.


Actually price cuts are usually strategic. Like when AMD slashed prices on its X2 lineup right before the C2D launch last year.

12:18 PM, April 23, 2007  
Blogger Sal said...

tech4life:

Yeah, but AMD was desperate. As when they AGAIN slashed their prices on the 9th of April, before Intels 22nd April cuts.

We knew about that cut in february, and they could'nt have known how Barcelona would perform then.

We see the same situation now, April has unofficially said new cuts in Q3, and introduction of new CPUs.

12:43 PM, April 23, 2007  
Blogger Sal said...

oh my god, i ment "Intel has unofficially...". Not April :)

12:44 PM, April 23, 2007  
Blogger Bubba said...

rather,

I assume your (broken) link is to the AMD press release that says "up to 50% faster per clock".

Up to? Marketing speak for the absolute best case scenario we can build. We're up to 100% worse in a real world case.

And we won't even get into the completly meaningless IPC metric.

12:53 PM, April 23, 2007  
Blogger symbiansn said...

At least, Intel can drop in more cache... to such an extent which they may remove their outdated and junky FSB completely as well.

1:17 PM, April 23, 2007  
Blogger lex said...

"Ph"ony "D"octorate I hate to break it to you.

Barcelona is too little too laaaaaaaaaaaaaaaaaaaaaaaate!

As nice as them Barcelonda numbers look it ain't enough. Actually they are a bit dissapointing; 50% Fp and only 20% integer??? What I want to know is how the other benchmarks look like. I'm suprised at the tradoffs the AMD archtects took in balance FP vs integer. Can't wait to see how dissapointing the improvements look across the rest of the common benchmarks. Is it no wonder they aren't getting released.

Lastly, Barcelona will have a window of 9 months before Nehalem pushes them into the under 200 buck space. Manufacturing this 281mm^2 die on 65nm from one 300mm factory compared to a Penrym at 107mm^2. Even if you times it by two its only 214mm^2. Also rememeber INTEL only has to yield "pairs" of core while AMD will be scrapping and downbinning anything that doesn't have 4 perfect cores. Also INTEL has advantage that a larger portion of the area is cache that has redundancy. AMD double penalized, big die, 65nm, one factory, 4 cores on a single piece of silicon, less cache and redundancy.

AMD will be scaling back capacity expands that will further limit Barcelona ramp and yield learning. This is a horrible cycle where AMD won't recover.

Recall when Opteron came out AMD had only one factory too, but INTEL was shooting itself in the foot with Netbust Prescott.

INTEL ain't shooting itself anymore. Instead its regained focus and the cross hair has moved from its own foot to AMD's bankaccount. AMD will soon be bankrupt. Hector really blew his wadd here on ATI when he should have been building factories and pulling in 65nm 4 years ago.

6:57 PM, April 23, 2007  
Blogger lex said...

What a trick heah, "Ph"ony "D"octorate!

Got 4 300mm 65nm factories competing against a single 90nm AMD factory.

Will have 4 45nm factories competing aginast a single half empty 65nm 300mm AMD factory.

That gives you flexibility to bolt on a huge amount of cache to cover a bandwidth bottleneck. Damm the covered it so well with the C2D architecture they are able to win all benchmarks that people seem to care about.

Who cares its at the termal limit? The fundamental architecture can overclock a ton and that tells you the headroom for Penrym on 45nm wiht lower gate leakage. Damm, think about that one trick pony on 45nm with increasing the cache and making it even smarter on Penrym with 6M.

What a trick heah? One that turned a 1.6 billion profit this past quarter.

What a trick heah? Forced AMD to drop prices greater then 50% on their line and causing their 2nd 600 million loss quarter.

Phoney do you realize that after this past quarter, if you sum up AMD's sum profit/loss they haven't made a damm dollar since their founding. Is that amazing, AMD hasn't made a dollar in its lifetime. Sure they got some fancy designs, a outdated process running in Dresden, but no a dollar.

POOR AMD, POOR "Ph"ony "D"octorate, that is one damm good trick heah?

I'll take that trick over AMD anyday. You see what BOise State did with all them "trick" plays? Pss, phony, they WON the game, just like INTEL. WInning is all that matters and INTEL is winning.

6:58 PM, April 23, 2007  

Post a Comment

Links to this post:

Create a Link

<< Home