Intel IDF Day 1 - Quad Core, Santa Rosa And More

MojoKid writes "From demos of the new Alan Wake game engine on a 3.73GHz overclocked Quad-Core QX6700 to design showcases with a wafer of 80-core teraflop capable chips, Intel's IDF opening day was brimming with tech-wonder from the company affectionately known as Chipzilla. Paul Otellini also showed pics of upcoming fab facilities in Arizona (Fab 32) and Israel (Fab 28). In total, Intel will have three 45nm fabs by the end of next year at an investment of about $9B, all targeted 45nm manufacturing processes. Finally, a bevy of Quad-Core Kentsfield-based systems are shown here, with Dell and Voodoo's offering looking especially swank."

Comments

Grr

[in2mind]

Paul Otellini, Intel's president and CEO, kicked off this season's IDF by coining the phrase "It's what's inside that counts", and spoke about why processing power matters again.

All these years we all thought whats outside the processor that matters?

Re:Grr

[Apocalypse111]

...spoke about why processing power matters again.

When did it stop mattering, and why wasn't I told?

Re:Grr

[TheRaven64]

It stopped mattering to Intel Marketing while NetBurst was being completely spanked by the K8. Now the Core microarchitecture is (finally) giving competitive performance/money and performance/power numbers, it matters (to Intel Marketing) again.

Re:Grr

[wwiiol_toofless]

Yep, I was thinking more neon case-lights, and the latest Intel sticker more metafloppers per parsec. No really, this time around, it's not about marketing, it's about *chokes up* what's inside... No really.

Re:Grr

[Red Flayer]

Not only that, but he didn't even coin the phrase. Hell, even the Enquirer used that phrase in a headline in 2000. Of course, they didn't put a 80-teraflop chip into the puppy in the article in order to get it to perform aide tasks -- now that would have been innovative.

Re:Grr

[Apocalypse111]

Well, good for Marketing. When it stops mattering to someone who matters, let me know.

Re:Grr

[Anonymous Coward]

It's not just Marketing. The important part is that performance matters to Intel *MANAGEMENT* again.

Re:Grr

[LuxMaker]

"All these years we all thought whats outside the processor that matters?"

Apparently he has not pulled his head out of his CPU long enough to tell what really matters.

Core 2 Challenge

[oskard]

http://www.hothardware.com/image_popup.cfm?image=b ig_keynote_139.JPG&articleid=877&t=a

The image says:

ANNOUNCING THE
$1,000,000
Intel Core 2 Challenge

What is this challenge they speak of? I want a million dollars...

Re:Core 2 Challenge

[TheRaven64]

Unless they are running two challenges, it is to design a Mac Mini-like box to be marketed under the ViiV brand. The prize money is (as I recall) $600,000 for product development, and $400,000 for marketing. I don't know what this will do for the Apple-Intel relationship; paying people to compete harder with your customers isn't exactly the kind of thing that makes suppliers popular.

Re:Core 2 Challenge

[NatasRevol]

I wonder if Apple/Steve Jobs could just walk in to Paul Otellini's office and ask for the $1M check?

Re:Core 2 Challenge

[frankie]

Yes, that certainly sounds funny, but I bet it's actually true. If Steve Jobs called Otellini and said he was willing to slap "Intel(r) Core(tm)" and "Intel(r) Viiv(tm)" logos on the mini, Intel would be all over that with a big fat grin. Apple is already foregoing piles of marketing moolah from Intel for not using the logos on Mac, so it would be a huge score if they gave in.

Re:Core 2 Challenge

[Amouth]

i wonder if i could get it buy. buying a macmini / spray paint/ and rebrand it.

Re:Core 2 Challenge

[el_womble]

I know Apple arn't the richest company in the world, but would they really sell out for a measly $1m? The corporation that pimps me out makes $3.5m a day (thats for 3500 staff) and we're no where near as big as Apple.

Re:Core 2 Challenge

[Slider821]

The challenge is, without using AMD technology, create an intel based chip that you CANNOT also use as a barbeque.

Catchup again's Sun's 8 core UltraSparc T1

[dananderson]

Sun's UltraSparc T1 has 8 cores, 32 threads. So, will Intel catch up anytime soon?

Re:Catchup again's Sun's 8 core UltraSparc T1

[TheRaven64]

Apples to oranges. The T1 is a superb chip for some workloads, and an appalling one for others. The T2, which has an FPU for each core (unlike the shared one in the T1) should do a bit better, but there are still a lot of workloads where the T1 does very badly. This is why Sun still sell UltraSPARC IV+ chips as well, and these are only dual-core.

Re:Catchup again's Sun's 8 core UltraSparc T1

[DrDitto]

Sun's UltraSparc T1 has 8 cores, 32 threads. So, will Intel catch up anytime soon?

And my single-threaded program runs 10x slower on a T1 than it does on a P4. I measured this myself.

Re:Catchup again's Sun's 8 core UltraSparc T1

[Anonymous Coward]

Intel will have 8-cores in 2008-2009 that will still be fast on single-threaded workloads. Niagara is sometimes a pain to work on because you have to go back to year 1999-2000 in single-thread speed on those machines. Niagara are good on special workloads, in the majority of cases they are not fun to work with.

Re:Catchup again's Sun's 8 core UltraSparc T1

[TheLink]

The T1 was probably behind the very day it was launched.

And by now it's way behind as AMD and Intel battle it out furiously.

Just look at Intel's new CPUs (which beat AMD's chips is many situations by 40-50% - AMDs chips beat Intel's old CPUs by a similar margin too ).

Show me a real world benchmark done by an independent party where the T1 does better.

Re:Catchup again's Sun's 8 core UltraSparc T1

[Programmer_Errant]

My impression is that Intel's more worried about the applications keeping up with the cores. If today's apps could utilize all the cores you could throw at them, processors like Sun's T1 and T2 could be a problem for Intel. Intel and Sun need a killer app to promote their multi-cores. To do this, you're going to need to get multi-core machines into the hands of a lot more developers. I think Intel has a better chance of this and consequently the killer app will be more suited to Intel's processors than Sun's processors.

Re:Catchup again's Sun's 8 core UltraSparc T1

[raftpeople]

IMO, all non-x86 are too small market share to make a long term difference (probably). Whether Sun's procs are good or bad, I just don't see a server proc becoming a mass market standard like x86. Consumer's and volume is where it's at. IBM probably has the best chance at staying relevant due to cell and the fact it is targetted at high volume consumer electronics.

Re:Catchup again's Sun's 8 core UltraSparc T1

[TheRaven64]

IMO, all non-x86 are too small market share to make a long term difference

I disagree. x86 is a tiny, tiny market. PowerPC and ARM are both much bigger. Not sure about SPARC, but it might be. Have you looked at mobile 'phones recently? They're out-selling x86 CPUs, and most have a PowerPC or ARM chip in them. Take a look at a modern BMW; it has 25 PowerPC-based computers on-board. The three next-generation consoles? All of them are PowerPC based.

We are coming to the end of the desktop computing era, and x86 isn't likely to outlast it by very long. Thank God.

Re:Catchup again's Sun's 8 core UltraSparc T1

[samkass]

We are coming to the end of the desktop computing era

Um, no. You might... MIGHT... be able to make that statement if there's ever a year in which the year-over-year growth of desktop PC sales (let alone usable installed base) decreases. For now, we appear to still be in the initial geometric expansion phase of desktop computing.

They're not quad-core though

[tomstdenis]

they're dual-die. There is a difference. First, the dual-die process takes more power. Second, it costs more. Boo.

Core2Duo is neato, you can overclock them like mad and the ALU/FPU is very efficient. But let's not kid ourselves. dual-die is not the same thing as quad-core.

Tom

Re:They're not quad-core though

[maynard]

Those dual die dual-core chips are fabbed at 65nm. The upcoming 45nm facilities should be able to manufacture quad-core chips on a single die.

Re:They're not quad-core though

[Anonymous Coward]

not quad-core? Hmmmm....

One physical chip; 4 microprocessor engines/cores/brains/etc....

So, is it ok to call it Quad-Engine? Quad-Core? Quad-Brain?

Or better, you mention dual-die, so is the new chip a dual-core-dual-die? In that case Intel only has a 2x2 platform which can't compete at all with AMD's 4x4.

Re:They're not quad-core though

[Wesley Felter]

There is a difference. First, the dual-die process takes more power. Second, it costs more.

Putting the same number of transistors on one die takes as much power as putting them on two dice. (Unless there is some law of semiconductors that I haven't heard of.) Also, yield is lower (thus cost is higher) on a larger die than on two smaller dice.

Don't you work for AMD? I guess your attitude is not a suprise.

Re:They're not quad-core though

[diegocgteleline.es]

they're dual-die

As Intel Ceo said in the press sesion: "The initial (four-core chips) are multi-chip, but so what?" he said. "I think you guys are misreading the market (if you think customers care about that)."

First, the dual-die process takes more power

I will wait for their quad-core before judguing it. The first dual core Intel CPU was hot and slow because it was based in the presscot platform, not because it was dual-die. This "dual-die" quad core is based on the Core 2, which is a great platform, IMNSHO.

In other words: I don't really care if it's dual-die or quad-core. If it's fast and eats few power and it's not expensive, then my money is for Intel. The Core 2 platform looks good enought to make possible a dual-die quead core processor that beats the alternative from AMD - not because the dual-die design is better, but because the Core 2 is better. Additionally, this is just their first quad-core CPU - the rest of quad-core CPus will be real quad-core beasts.

Re:They're not quad-core though

[yabos]

There's more to it than just the core. There's all the rest of the CPU which includes cache(usually shared for dual core while separate on dual CPUs) and the rest of the interconnects for connecting to the motherboard and whatnot.

Re:They're not quad-core though

[LWATCDR]

The real issue will be memory speed.
The Core2 is a nice chip but two of them is really going to be pushing the memory bus. AMDs Hypertransport scales better than Intel's FSB. Not only that but do the math. Intel's quad core system is going to run twice has hot and use twice the power of the Core2. Why? because it is two of them.
The speed for watt and the speed for heat ratio will be worse than the Core2 because it will not be twice as fast. AMDs true quad core may very well bet Intel's two die solution. It will probably kill it once you get past one CPU/ 4 cores. AMD should be much better in the 8 core+ server market than Intel's solution.

Re:They're not quad-core though

[diegocgteleline.es]

Intel's quad core system is going to run twice has hot and use twice the power of the Core2. Why? because it is two of them.

Applying your logic: "AMD's quad-core systems are going to run x4 hotter and use 4x the power of a single opteron core. Why? Because there're four of them".

Re:They're not quad-core though

[pdbaby]

Unfortunately the 2.66GHz Core 2 Quadro uses 167W while idle(as much as the Core 2 Extreme 2.93GHz at full load) and 260W at full load. With heat to match, of course... I think I might wait a wee while to upgrade my Core 2 Duo to a core 2 quadro, anyway!

Re:They're not quad-core though

[tomstdenis]

No, because AMD is 95W at 90nm, the quad-cores will be 65nm parts.

The quad-core from AMD will be a real 4 core in a single die designed to fit in the 95W power enveloppe. Unlike Intel who designed a 65W processor and just pasted two of them on there.

Tom

Re:They're not quad-core though

[Anonymous Coward]

Those dual die dual-core chips are fabbed at 65nm. The upcoming 45nm facilities should be able to manufacture quad-core chips on a single die.

I don't think that the 65nm versus 45nm has anything to do with whether the CPU is dual die quad core or single die quad core. The reason is that the soon to be current Intel quad core CPUs are not designed to be quad core, they are designed to be dual core, and they have essentially glued two dual core CPUs into a single package. That is little better than having two physical CPU sockets each with a dual core CPU.

The point of comparison would be AMDs quad core CPUs which were designed to be quad core from the ground up. The integration between all four cores is not only faster, it is more efficient. That, plus the fact that AMDs dual and quad core CPUs will be utilizing independent 2GHz point-to-point interfaces rather than sharing/flooding a single 1066- or 1333 MHz bus should translate into vastly better performance.

Now, for those people saying that if the transistor count is the same whether it's dual die or single die (both being quad cores), then it shouldn't be any more power efficient, keep in mind that the transistor count wouldn't be the same. With the dual-die method, you're basically taking two complete CPUs (including interfaces to the bus, etc) and putting them in the same package. Because of that, there will undoubtedly be redundant parts from each core that wouldn't exist if you had designed a single-die solution. Not only that, but there will be extra circuitry to make the connection between the CPUs. All of those extra transistors draw extra power, and generate extra heat. They also drive up the cost for no good reason.

Don't get me wrong, Intel's quad-core will be neat, but they're basically taking their existing parts and glueing them together and calling them something new. If they had re-designed them to be quad core they could undoubtedly come up with a design that is not only more efficient (both electrically and architecturally), but that might also include additional functionality (like the ability to dynamically adjust the speed and voltage of each core independentaly of the others, which AMD will have in their quad cores). The tradeoff of course is that it takes longer to get there, and Intel has been so far behind AMD that they felt that they just couldn't wait. The apt metaphor would be to compare a Pentium D (two single cores glued together) to an Athlon 64 X2 or Core 2 Duo (CPUs designed from the ground up to be dual core).

Now sure, some market droid at Intel has come out and said that the end-users don't care whether it's dual-die or single-die, as long as it's four cores. And on that point he is generally correct. But what the end-users will care about is overall performance, cost, power draw, and heat dissipation, and all of those things should be significantly better on a single-die than on a dual-die. The end users will end up caring about the benefits that come along with having a single-die, and it will make a difference. If Intel is lucky, they might just have a native quad core out by then, but I wouldn't bet on it.

Re:They're not quad-core though

[maynard]

The recent test batch, and first gen quad-core "chips" are all based on 65nm lithography processes. 45nm will allow Intel to cram all four cores onto a single die. So -- yes -- the lithographic process most certainly affects transistor count (which, in turn, affects how many cores Intel can fit onto a die).

Re:They're not quad-core though

[Anonymous Coward]

The recent test batch, and first gen quad-core "chips" are all based on 65nm lithography processes. 45nm will allow Intel to cram all four cores onto a single die. So -- yes -- the lithographic process most certainly affects transistor count (which, in turn, affects how many cores Intel can fit onto a die).

That's a red herring if there ever was one. Yes, the lithographic process will allow them to put more transistors in the same sized space. Yes, that means that they can make smaller dies. That's true of any CPU design, not just quad core designs. But it is the design of the CPU that affects whether it's a dual-die or single-die implementation of quad core, not the lithographic process.

Case in point, AMD is building it's quad core CPUs on a 65nm process, not 45nm. Both the Opteron X2 and the new dual core Xeons both have around 250 million transistors each (the Opteron has a smaller cache, but an onboard memory controller which makes up much of the difference). If Intel had wanted to design a single-die quad core CPU on 65nm, they certainly could have done so. They could have eliminated a lot of redundant transistors from each dual-core die and wound up with a single-die quad core CPU built on a 65nm process that was significantly smaller than it's current dual-die design and had fewer transistors than the sum of it's parts. It just would have taken more time to do the desgin, and Intel decided that they would rather glue two dies together in a single package than the design a single-die solution. Doing so means that they will be able to sell a "quad core" part 3-6 months before AMD does.

Re:They're not quad-core though

[maynard]

"If Intel had wanted to design a single-die quad core CPU on 65nm, they certainly could have done so. They could have eliminated a lot of redundant transistors from each dual-core die and wound up with a single-die quad core CPU built on a 65nm process that was significantly smaller than it's current dual-die design and had fewer transistors than the sum of it's parts."

Perhaps. Perhaps not. Being not an employee of Intel, nor privy to their private project roadmaps and internal engineering goals, I can't say whether they could or could not have done a quad-core design within those constraints. What I do know is that they have announced plans for a quad-core single die chip on 45nm. *shrug*

Re:They're not quad-core though

[dfghjk]

It's not the same to Intel but it shouldn't matter to you. All that matters is cost, heat and performance.

Re:They're not quad-core though

[dfghjk]

"The point of comparison would be AMDs quad core CPUs which were designed to be quad core from the ground up. The integration between all four cores is not only faster, it is more efficient. That, plus the fact that AMDs dual and quad core CPUs will be utilizing independent 2GHz point-to-point interfaces rather than sharing/flooding a single 1066- or 1333 MHz bus should translate into vastly better performance."

If any of this is true then it will be shown in benchmarks. Until then, the internal construction should only matter to those who make it.

"...All of those extra transistors draw extra power, and generate extra heat. They also drive up the cost for no good reason."

Those few extra transistors don't matter since they are dwarfed by other aspects of the part. They also don't drive up cost because their impact to die size is small. On the other hand, the design cuts die size in half resulting in higher yields and lower costs. Bringing in development of a true quad-core design on the older process adds additional cost as well. One would have to believe that Intel's decision factors in costs.

"Don't get me wrong, Intel's quad-core will be neat, but they're basically taking their existing parts and glueing them together and calling them something new."

They are something new.

"...and Intel has been so far behind AMD that they felt that they just couldn't wait."

That's speculation on your part. Intel doesn't have integrated memory controllers so their design tradeoffs are different.

"The apt metaphor would be to compare a Pentium D (two single cores glued together) to an Athlon 64 X2 or Core 2 Duo (CPUs designed from the ground up to be dual core)."

The approach is the same but the Pentium D cores were inferior to AMD's whereas the Core 2 cores are more competitive.

"Now sure, some market droid at Intel has come out and said that the end-users don't care whether it's dual-die or single-die, as long as it's four cores. And on that point he is generally correct."

Yes, he is.

"But what the end-users will care about is overall performance, cost, power draw, and heat dissipation, and all of those things should be significantly better on a single-die than on a dual-die."

That remains to be shown. What constitutes "significantly"? 5%? 10%? Won't be that different.

"The end users will end up caring about the benefits that come along with having a single-die, and it will make a difference."

No they won't. Early adopters will want the latest thing. The others won't buy it. One die or two doesn't really make that much difference.

Re:They're not quad-core though

[LWATCDR]

Nope because AMD is doing a die shrink to get all the cores on a single die. The die shrink should buy you a lower heat output. Until we get them in our hot little hand we don't know how well this will work.
Intel on the other hand is just sticking two dies in a package. Each die will produce roughly the same heat and draw the same power as a single Core 2 chip since that is what they are.
Now when Intel goes for a die shrink it will also have a lower heat output as well as a true quad core cpu. The problem than will be the FSB.
Frankly it is starting to look as Intel is going from a mindless clock speed race to a mindless core count race. It could be netbust all over again.

Re:They're not quad-core though

[Extide]

No you guys are missing the point. They are still quad core whether its one or two or even four pieces of silicon, but that doesnt matter. Dual die is cheaper to produce because if you have one big die and there is a defect in only one of the cores they must throw the whole thing away, where as this way they can pick 2 good cores, and throw less away. The main advantage of running a single die is the ability to share cache. Each of the 2 dual core dies has one shared cache between those 2 cores, thus if they need to talk to each other they can talk through that, where as if they want to talk to the other die, it must go over the FSB. This is a similar setup to AMD's 4X4, where as the two cores in each socket can talk directly, but to access the other 2 they need to go out the HT link, although I think the HT link is going to be a little bit faster than going over the FSB. The process 65 vs 45 nm has a lot to do with getting the four cores on one die -- because of the reasons stated above, they will be smaller, and less prone to a defect, and less thrown away when there is a defect. The only reason they are doing this is to get it out the door faster, and hey I guess this is better than nothing, and if you dont agree, then you can always get a Conroe or something from AMD...

Re:They're not quad-core though

[StikyPad]

Indeed. A small, single die. For perspective, the small gold dot in the corner is actually the top of a construction worker's hat as he proudly admires the finished product.

Re:I remeber this....

[thinkzinc]

I find the technology pretty exciting, but I get along just fine with my 2 year old laptop and 4 year old desktop.

The summary

[Mike89]

I hope the article makes more sense because the summary has me staring blankly at the screen

Re:The summary

[Zaatxe]

You are seeing it from the wrong angle...

Re:The summary

[Zaatxe]

Oh, nevermind... this message was supposed to be in other article. That's what happens when tabbed navigation falls in the wrong hands.

Seems 80 cores isn't enough to resist the /. effec

[victorhooi]

Lol..slashdotted already...

See...haha...pah, Mr. Intel, you fought vat your 80-cores could stand before me? Mwahaha.....behold, my invicible weapon, the slashdot....

Still, it'll be interesting to see how Intel markets this to the everyday Joe user. I mean, the whole HT thing was marketted as helping you to burn a CD while you watched a movie...wonder if they'll use the same line here.

"See, now you can burn *79 CDs* and watch 1 movie, all without breaking a sweat"...

Sure, if you pick up Process Explorer, you'd think there's hundreds of threads running, but the truth is, most of those are idling. And I always figured that you actually had to pre-plan how to split up a task parallelised, to really take advantage of this kind of thing.

Victor.

Moving fast now, eh?

[tygerstripes]

It seems the beast has awoken. They were so far behind in the chip-war with AMD for so damn long, it seemed their market share was getting nibbled to death.

Now they finally seem to have woken up and, by god, they are really moving now, aren't they. $9bn in 45nm fabs? A wafer of 80-core chips already? Speaking as a one-time AMD Fanboi, I have to say - the daddy is back.

(Let the flaming commence)

Re:Moving fast now, eh?

[Golias]

When Apple switched to Intel chips for the Mac, a lot of people were asking "why not AMD?"

At the time, the answer from Apple was "Intel showed us their future road-map, and we wanted on board."

Now we are starting to get a glimpse of what they were talking about.

Re:Moving fast now, eh?

[norman619]

Meh... I used to love Intel until AMD came around. CPU's that were as fast or faster for much less. I'm a 3D aminator and this news has me drooling but are they still gonna be asking an arm and a leg for their CPU's compared to AMD?

Re:Moving fast now, eh?

[turgid]

All other big companies (and even the small ones) that design CPUs are doing the same. Some even have a large head start (Sun with UltraSPARC T1). You need the interconnect too. Sun, IBM and AMD have that with Hypertransport (intel's work-a-like won't be out until 2008, a full 5 years behind AMD, and 12+ years behind Cray, Sun, IBM and SGI).

Don't forget software. The Solaris kernel has been there since Solaris 7 in the mid-late '90s (SGI Irix too). Linux is catching up on multithreding (2.4 kernels with NPTL). Don't mention Windows. It just doesn't scale at all.

The reason Apple chose intel was price, availability of large quantities of processors and intel processors always look good on certain synthetic benchmarks (small, tight loops that fit in L1 cache).

intel processors suck on servers. Always have, and will until 2008 when their new interconnect comes out. Unless you live in a very cold country and want an itanic for novelty value.

Re:Moving fast now, eh?

[Anonymous Coward]

Spoken like (the vast majority of Itanium critics) someone who has never even seen one, let alone tested it. You wouldn't buy whine about your CPU not being able to perform like a GPU would you? Itanium does very specific tasks better than anything else, period. On yourdesktop (porn, internet forums, IRC, more porn, slashdot, more porn, then searching for penile abrasion remedies) yes, it would suck.

Re:Moving fast now, eh?

[Anonymous Coward]

I understand, but ... I still feel dirty. Where's my PowerBook G5? ;-(

Re:Moving fast now, eh?

[Anonymous Coward]

Now they finally seem to have woken up and, by god, they are really moving now, aren't they. $9bn in 45nm fabs? A wafer of 80-core chips already? Speaking as a one-time AMD Fanboi, I have to say - the daddy is back.

Aside from mentioning that your attitude suggests that your fanboi-ism changes with the wind direction, we haven't even seen Intel 45 nm parts or Intel 80-core chips. Intel is successfully selling you on marketing hype only so that it can save its marketshare a$$. Chill out, dude. Peace.

Re:Moving fast now, eh?

[turgid]

Spoken like (the vast majority of Itanium critics) someone who has never even seen one, let alone tested it.

OK, I'll bite.

About 3 years ago at Linux Expo in London the Red Hat guys had an itanic workstation. It was turned off, because it had overheated. They'd put it away out of site and wouldn't let me see it. I wanted to try it out, to see how fast it was. No way.

The guy on the HP itanic stand was waxing lyrical about the "new" crossbar switch in the big iron machines. I pointed out that he was a decade behind the big iron Unix companies like Sun and IBM. Mind you, if a 4-way Compaq ProLiant running Windows 2k is the best server you've ever seen, you might have been impressed. I asked him about itanic's ability to run legacy code (I wondered if he'd mention the pathetic x86 hardware emulation that was 10% as fast as the Pentium at the same clock speed). He said, "16-bit code?" I choked.

You wouldn't buy whine about your CPU not being able to perform like a GPU would you?

No, but a decade ago, intel was touting its proprietary itanic design as the leviathan that would crush all "proprietary" (sic) RISCs. (Never mind that a lot of the big 64-bit RISCs are Open and some are now "Open Source" under the GPL).

This post-RISC all-things-to-all-men has been consistently over-budget, behind schedule and under performing. All of the benchmarks that they advertise and the few selected ones that actually show it in a good light (tight loops of numerical code). And they often compare against much older CPUs, and lower configurations, that other manufacturers. (A tactic straight out of IBM's book).

Now it's a niche player, relegated to a small handful of scientific applications. intel paid SGI to give NASA a 10240-CPU itanic supercomputer for free.

Never mind its horrendous power and cooling requirements.

I'll stick to my out=of-date, cheap and cheerful AMD processors, thanks very much.

I've got better things to fritter away my money on than another 20% gain on tight loops...

Oh, and I buy my porn on DVD from a reputable store and I have a high quality Cambridge Audio DVD player to watch it on.

ChipZilla

[Anonymous Coward]

Since when was ChipZilla an 'affectionate' nickname?

Re:ChipZilla

[Anonymous Coward]

I know. I always referred to them as "Pookey Bear." Much more affectionate.

Re:ChipZilla

[rsborg]

Since when was ChipZilla an 'affectionate' nickname?

Dunno, maybe it has something to do with naming similarity to the friendly Mozilla?

Re:Seems 80 cores isn't enough to resist the /. ef

[TheRaven64]

Sure, if you pick up Process Explorer, you'd think there's hundreds of threads running, but the truth is, most of those are idling.

Pretty much all desktop apps can be split into two categories:

1. The ones that contribute to the 5-20% load that your CPU generally sits under. (Web browsers, mail clients, music players, etc).
2. The ones that cause the load to spike at 100% for extended periods. (Audio/Video encoders, compilers, typesetting engines, etc).

Applications in category 1 will not see any benefit from a CPU that's twice as fast. That 5-20% load may drop to 2.5-10%, but no one cares. Those on category 2 will complete in half the time (assuming that they are CPU-limited and linearly scalable). As CPUs get faster, more and more things fall into category 1. Once you run out of things in category 2, stop upgrading. This happened for a lot of people about five years ago.

I recently found out about an interesting experiment Intel did a few years back. They have a full-system simulator that allows them to test various things easily. They modified it so that all CPU operations took zero (simulated) time to complete. This gave about a 2.5x speed improvement for most tasks, i.e. an infinitely fast CPU only gave a 2.5x speed boost to most tasks. It doesn't take a huge speed increase before you run out of CPU-limited things and start hitting memory, disk, and network bottlenecks.

Santa Rosa

[baudbarf]

What's the relevance of Santa Rosa?

Re:Santa Rosa

[paskal]

They mention the upcoming Santa Rosa platform briefly on page three of the article:
http://www.hothardware.com/viewarticle.aspx?page=3 &articleid=877&cid=9

He spoke of the upcoming mobile "Santa Rosa" platform that will incorporate NAND storage, Core 2 Duo CPUs, 802.11n WiFi, and G965 chipset. Santa Rosa will be a big step up in performance, and will have better battery life through advancements in manufacturing and the use of NAND storage. NAND will be integrated onto the motherboard in Santa Rosa and we can see its obvious usefulness in a sort of scratch-pad application, allowing drives to spin down while data is cached in NAND memory perhaps. A NAND flash array should also allow for faster access to OS functions for enhanced instant-on capabilities. He then spoke of the progress with WiMAX and stated that WiMAX will be part of the Centrino platform in the future, as part of a total WiFi solution.

Re:Santa Rosa

[Com2Kid]

It's in north cali so it doesn't suck so much?

Only thing I can think of...

Re:Santa Rosa

[TheDarkener]

I live in Santa Rosa, and trust me, the *only* reasons they would name something after this city would be because it's:

1) Going to cost WAY too much
2) Going to drive an SUV 2 miles to work, and slow down to 2mph on right-turns
3) Going to drink wine and believe that it's "high-class"
4) Going to pay >$1M for a house that's about 2 inches away from the next >$1M house
5) Going to hide from its own shadow because of the terrorists downtown (people with spiked hair)

Re:Santa Rosa

[Anonymous Coward]

Maybe it's going to have delicious mexican food.

Re:Santa Rosa

[Com2Kid]

Well, in its defense, I have had good experiances with the areas surrounding Santa Rosa... :)

Wow...

[neoprog]

I don't really feel like dual-core has really even reached any kind of saturation point in terms of how many people own a dual-core processor. I can't believe (but I do thank) Intel for pushing ahead and making good tech affordable to everyone. I spent a small fortune building the box I'm typing this on (with an AMD x2 socket AM2 4600+). And I know I got a little burned by the am2, but I think the upgrade path will stay clear for a while. I wish I could have waited, but I needed a computer for college and this was top-end about 4 months ago. I loved AMD when they had the better stuff, and now I love Intel, cause the core2's kick butt. When AMD catches up, that's when I'll upgrade my 4600+. ;) prog

Re:Wow...

[oc255]

Please explain how you got burned because I'm considering an AM2 based mini-server for home. Burned as in going from 939->AM2 chipset isn't compatible? Price?

Re:Wow...

[Anonymous Coward]

I just built a new AM2 server because my old AMD SocketA 2600+ had died on me, and the price wasn't too bad. The GP might have built his machine when AM2 was new and before the major price drop on AMD chips. It cost about $600 including tax and shipping on NewEgg for the following:

Athlon 64 X2 3800+ AM2 ($150)
Asus M2N32-SLI Deluxe Wireless Edition Motherboard ($200)
OCZ Gold Series 2GB (2 x 1GB) 240-Pin DDR2 SDRAM DDR2 667 ($180)
Radeon X300SE 128MB PCI Express Video Card ($35)

This is for my server, so I wanted to get a video card with passive cooling and don't care about performance. Also the motherboard I got is probably way overkill, but it seemed to have the best scores in multiple online reviews, and also had passive cooling for the chipset. I already had the hard drives, case, and DVD drive, which made it really cheap, but even with all that you are looking at $1000 or less.

Also to answer your question, AM2 and 939 are not compatible, and you will need new memory since socket AM2 uses DDR2 memory instead of normal DDR now. Maybe there are a couple AM2 socket motherboards that support DDR? I have no idea, but most likely not.

Re:Wow...

[WuphonsReach]

None of the AM2 chips support DDR (the whole reason we needed a new socket was so that you couldn't mix up the DDR2 AMD chips with the DDR AMD chips). Plus all of the AM2 chips (except Sempron) support hardware virtualization (allowing virtual machines using unmodified guest OSs).

Otherwise you can compare the Asus product line and you'll see that the AM2 motherboards look a *lot* like the 939 motherboards. The only difference is hardware virt and the new type of memory.

(I've used both of the 6150 chipset microATX boards, 939 and AM2. We're using them for business workstations. Also own a SLI deluxe AM2 and the M2N-E AM2 motherboard. The M2N-E has better Linux support using the latest official BIOS, for the M2N32-SLI Deluxe you'll need to get ahold of the unofficial 0704 BIOS.)

I Was Just Starting To Like Intel Again!

[chromozone]

From the article:

"Paul Otellini, Intel's president and CEO, kicked off this season's IDF by coining the phrase "It's what's inside that counts", and spoke about why processing power matters again"

But then this in another article covering the same event:

"Otellini briefly responded to concerns that Intel's first quad-core packages are simply "glued-together" dual-core processors while AMD is working on a native, single-die quad-core chip. "So what?," said Otellini, adding, "The public doesn't care what's inside a processor."

http://www.tgdaily.com/2006/09/26/intel_core_2_qua d_announcement/

In yet another article in Ars Technica we read that Intel is look to an 80 core chip. I like the Core 2 Duo a lot but I hope the Intel megahertz fixation isn't just going to become a "core" fixation .

  http://arstechnica.com/news.ars/post/20060926-7840 .html

Robert Moses built a lot of bridges and roads around New york hoping to relieve congestion but it had the counter-intuitive effect of creating more traffic. I hope all the increases in size and power of computers doesnt just bring more garbage. With all the legacy code bloat, and things like video cards that get hot as toasters and power supplies that waste energy (the Google thing) I think computing could use a few reductions instead of increases. In that regard it's nice to see the Core 2 Duo bring down the wattage.

Re:I Was Just Starting To Like Intel Again!

[Jeff DeMaagd]

That is a good point, but I really don't think the Kentsfield being two Woodcrests "glued" together really is a big problem. People have been raising a big stink about it as simply being bad. The problem is that other than for gaming, the benchmarks seemed to show that there is a marked performance advantage over Kentsfield over Conroe, especially for media work, rendering and CAD.

Re:I Was Just Starting To Like Intel Again!

[DeadChobi]

Yes. I would like a computer which is just as fast and powerful as the one I have now, but that doesn't heat my entire apartment. Thanks to the Athlon 64 and my 450-watt power supply space heaters are a thing of the past.

Additionally, it would be nice if we could get past the "My specular lighting and bump mapping is slightly better than yours" pissing contest that causes all these hardware upgrades. I'm perfectly happy playing System Shock 2 at upwards of 200 frames per second because it's an engrossing game. Am I alone in saying that I don't really want a quad-core CPU? How about an offering that addresses power consumption and heat output exclusively.

Re:I Was Just Starting To Like Intel Again!

[chromozone]

Lol - my thoughts exactly. Just this week I was reading an article about video cards that were measured for temperature and plenty of cards were going to or over 200 degrees F. The first quad core will double the wattage of the Core 2 Duo and cost 1000 USD. But Intel people have said in interviews that dual core is the focus for now so even they arent pushing the news quads too hard at the mo. Upgrades for power just dont look appealing like a few years ago. Its beautiful to see Core 2 Duo do more with less. It would be nice to see that trend continue

"Core" fixation

[benhocking]

In yet another article in Ars Technica we read that Intel is look to an 80 core chip. I like the Core 2 Duo a lot but I hope the Intel megahertz fixation isn't just going to become a "core" fixation .

Speaking as someone who uses code that is very parallelizable - I hope it does! (Well, assuming that they also address memory bottlenecks and what not.)

Re:"Core" fixation

[OverlordQ]

No, you shouldn't hope it does. If Intel gets on a core fixation then they'll cram all the cores they can on there without a bus able to keep each core running full bore. So either an 80-core Intel chip that has bandwidth to keep maybe 20 going, or an AMD chip with 20 cores that has no problem with bandwidth.

Re:"Core" fixation

[Chazmyrr]

The 80-core chip is not a CPU. It is strictly for floating point. It could potenially be used for raytracing or scientific calculations that benefit from extreme parallelism.

I would guess that this will first be seen in some type of add-on board with it's own local memory. I don't think FSB speeds are likely to increase enough in the next few years to keep this monster fed otherwise.

Fat, dumb, and happy.

[Shivetya]

I was thinking the same thing. Very much like America before Japan forced official entry into the war. Very content with themselves and position in the world.

Just as it took America years to spin up properly, leveraging her resources, Intel now has come back to fight more prepared and ready for the long haul. The question that remains is, can AMD keep up with an Intel obviously aware of what the mission is?

I would hope that with AMDs recent acquisitions that they not only keep up but open some new areas as well. As mentioned on other threads, 80 core CPUs won't replace dedicated graphic cards but if all the cores are not the same then you can do about anything. Essentially bringing the "cell idea" to the x86 market.

Hey, if my PC can get down the size of a cell phone with my only needing to buy stand alone devices I am all for it. I would love to have nothing more than machines the size of Apple's Mac Mini with all the bells and whistles expected out of top end machines.

Re:Fat, dumb, and happy.

[dfghjk]

That's a pretty disrespectful analogy even if you find it accurate. There were estimated 62 million lives lost in WWII. No need to trivialize that by comparing it to a technology competition.

intel chips sets

[Joe The Dragon]

Intel chips sets are still behind Nvidia and Ati. The best that Intel has is x8 x8 or x1 x16 while Nvidia has x16 x16 one 590 board even has x16 x8 x16 and most 590 boards have with duel gig-e with ip/tpc offload. These are for amd 590 as the Intel ones are not out yet.
This also shows up in the workstation / severs chips sets as well. Aka the power Mac g5 has more pci-e lanes then the Mac pro and it has less bandwidth in the chip set to chip set link.
Also looking at Motherboards form super micro

Dual AMD® Opteron(TM) 2000 Series (Socket F) Support, 1000 MHz HyperTransport Link
2 (x16), 1 (x8) & 1 (x4) PCI-e, 1 64-bit 133MHz PCI-X, 1 64-bit 100MHz PCI-X
      Up to 16GB DDR2 667 SDRAM (or)
        Up to 16GB DDR2 533 SDRAM (or)
        Up to 32GB DDR2 400 SDRAM
  Dual-port Gigabit LAN/Ethernet Controller
  2-Channel Ultra320 SCSI with Zero-Channel RAID support

  Dual Intel® 64-bit Xeon® Support (667 / 1066 / 1333MHz FSB)
  Up to 32GB DDR2 667 & 533MHz FB-DIMM
  1(x16) & 1(x4 in x16) PCI-E, 2 64-bit 133MHz & 1 64-bit 100MHz
      PCI-X, 1 32-bit PCI
  Dual-Channel Ultra320 SCSI & Zero-Channel RAID Support
  Dual-port Gbit LAN

The amd board has a lot pci-e slots

Tyan

  (2) AMD Opteron(TM) (Rev.F) 2000 series processor support (1207-pin)
  (8) DDR2 DIMMs sockets; up to 32GB reg. DDR2 400/533/667 mem.
  Supports ECC memory moduels; dual channel memory bus
(4) PCI-E x16 slots
      - (1) x16 signal from IO55
      - (1) x16 from MCP55
      - (1) x16 from MCP55 with x8 signal
      - (1) x16 from IO55 with x8 signal
    (2) PCI-X 100MHz slots from NEC nPD720404
          or (1) PCI-X 133MHz slot if 133MHz card is used
    (1) PCI v2.3 32-bit 33MHz slot
    (7) Expansion slots total
  (6) SATA2 ports (3.0Gb/s), (8) SAS ports (opt.), and (2) GbE LAN ports
  (1) 1394a FireWire port and integrated audio
  SSI/Extended ATX footprint (13" x 12"; 330.2mm x 304.8mm)

Tyan does not any xeon workstation board with a full x16 slot.

To running something like this http://www.nvidia.com/page/quadroplex.html at the max you need 2 full x16 links.
http://www.nvidia.com/page/nforce_pro_workstation. html

intel is missing out on the high end workstation graphics market.

Re:intel chips sets

[julesh]

Intel chips sets are still behind Nvidia and Ati. The best that Intel has is x8 x8 or x1 x16 while Nvidia has x16 x16 one 590 board even has x16 x8 x16 and most 590 boards have with duel gig-e with ip/tpc offload. These are for amd 590 as the Intel ones are not out yet.

My understanding is that the nvidia chipsets only have 20 pcie channels, so when you have two x16 cards in there, it behaves as though those slots were x8s.

Still:
  (1) at least the slots are there, which they aren't on intel based boards;
  (2) 20 channels is better than the 16 that the best intel boards offer; and
  (3) 8 pcie channels gives you an equivalent bandwidth (32 gb/s) to the FSB on current processors (~33 gb/s for a core 2 duo), so why would you want more than 8 anyway?

Re:intel chips sets

[Joe The Dragon]

intel 590 is 48 lanes

NVIDIA nForce Professional 3600 and 3050 56 lanes 12 links Flexible
NVIDIA nForce Professional 3600 28 lanes 6 links Flexible
NVIDIA nForce Professional 3400 28 lanes 6 links Fixed
NVIDIA nForce Professional 2200 and 2050 40 lanes 8 links Flexible
NVIDIA nForce Professional 2200 20 lanes 4 links Flexible

Re:intel chips sets

[WuphonsReach]

I think the newer nForce 590 has more PCIe lanes. IIRC, the one motherboard I have with nForce 590 is a 16+16+4+1 for a total of 37 lanes (another note says 46 lanes)

http://www.sudhian.com/

nForce 550 - 20 lanes
nForce 570 - 20 lanes
nForce 570 SLI - 28 lanes
nForce 590 SLI - 46 lanes

Re:I remeber this....

[jank1887]

me too. I don't expect to need more power from my next computer (well, Vista aside). BUT I wouldn't mind if it USED less power to do the same amount of work, and the smaller transistors can help with that.

WAIT a minute

[BrookHarty]

1 youtube video can max out a cpu? Sheesh, this is the thing people are talking about, how can you have a multitasking system when every application spikes the cpu. Seems Hyperthreading was a nice limp along until dual core, now we need more.

You need a core just for IO! I'm using a dual core, and even its being pushed to its limits, I cant wait for a quad or dual quad core to actually make a system multi-tasking friendly.

BTW, This isnt a windows bashing comment, this happens in linux too...

Re:WAIT a minute

[Gothmolly]

Try BeOS then, its so threaded and slick, you'll think you're on a multicore machine, even with a single core.

+not a BeOS user, just tried it once
++Slashies!

Re:WAIT a minute

[Anonymous Coward]

Oh,I take it as a Windows-bashing comment. Or, Media Player at any rate. WMP sucks balls and sucks CPU power. Mplayer on a Gentoo box will play Youtube clips at like 10% CPU usage on a 1.5ghz Celeron M. If you do run into CPU usage problems under Windows with videos, try mplayer.. the Windows version seems to be roughly as efficient as the Linux version.. like under 1/2 the CPU usage as Media Player for the same videos. Plus, mplayer doesn't randomly crap out on some videos or play them without sound 8-). For that matter, Media Player 6.4, now available online as "Media Player Classic", also uses way less CPU power than WMP 7,8,9, or 10.

Re:WAIT a minute

[DeadChobi]

Fullscreen DivX video, 640x480, on an Intel Celeron 333. 100% CPU use, but it still gets all the frames out.

Go IDF!

[mi]

What's the relevance of Israeli Defense Forces?

Re:Seems 80 cores isn't enough to resist the /. ef

[MikeBabcock]

You need to move into the modern era.

#2 includes web browsers now (mine spikes 100% frequently for Flash / embedded video / javascript on AJAX sites).

Also it includes chat programs (which include real time voice/video communciation), and others. Many people are finding their PCs too slow to do 'new' things with because those 'new' things are hard on the CPU.

The what now?

[MagicM]

a 3.73GHz overclocked Quad-Core QX6700

mmmm...

a wafer of 80-core teraflop capable chips

ooohhhhhh...

three 45nm fabs

aaahhhhh...

swank

WTF!?

Re:The what now?

[Anonymous Coward]

Umm... hello? English must be a second language I guess? From dictionary.com...

Swank -
dashing smartness, as in dress or appearance; style.

In other words, the Voodoo PC bajillion core system they commented on was, in other words, dashing smartness, as in dress or appearance; style...

So that would be in-fact, WTF they were talking about, I assume.

Re:The what now?

[MagicM]

Actually, I was commenting on the nerdcore/1337/whatever-ness of most of the submission, followed by the are-we-back-in-the-70s-ness of the word "swank".

I guess I was too subtile.

Re:Seems 80 cores isn't enough to resist the /. ef

[FireFury03]

As CPUs get faster, more and more things fall into category 1. Once you run out of things in category 2, stop upgrading.

Your arguement is based on a faulty assumption - that is that the limiting factors will not change. Whilest it's true that something like a word processor is basically limited by the typing speed of the user and throwing CPU at it isn't going to do much, the idea that applications limited by, for example, memory bandwidth will end up in category 1 and stay there is flawed. Memory is getting faster all the time. Yes, it's currently being outpaced by CPU performance improvements, but if you stop upgrading the CPU you'll see a bunch of category 1 applications migrating back into category 2 as memory, network, etc. get faster.

That said, I've thought for the past few years that there is no reason for most users upgrading because most users aren't doing anything more intensive than web browsing, word processing, etc. Of course there will always be a few users who will max out whatever system you give them. Although I'm sure the number of people maxing out their systems will dramatically increase once they start installing the heavyweight lump of crap known as "Vista" :)

IDF?????

[sebol]

When I first saw IDF at the topic, the first thing come into my mind is "Isrealli Defence Force"

Correct me if I'm wrong

[matt328]

But those look like intel's 975XBX mainboards in the pictures of the quad core machines. I thought I read earlier that intel was hyping the 965G chipset and not the 975X for their quad core architecture? Unless I'm totally missing something and intel has started using black pcb and the blue flame heatsinks on a 965 board.

"Only" 20

[benhocking]

If, within 5 years, (as Intel is promising) I'll be happy if it only has the bandwith to keep 20 going. I do understand the problem with them not keeping other bottlenecks up to the task of 80 cores, but, if I can only get 20, that's better than 2 or 4.

MCM isn't always so simple

[amh131]

Well, the pads/pad drivers do consume a very significant quantity of the total die power. A dual-die solution has more pads that run at the same speed as a single-die and so should consume more power. In *theory*, it should be possible to leverage those additional I/O for increased performance since you do have more bandwidth. In practise, it might be damn difficult.

All of this is implementation dependent. Did they do a true MCM with reduced driver sizes? What sort of package are they using? How is the interconnect between the two dies managed? You'd have to see it and benchmark it to know for sure -- everything else is fairly idle speculation. Including this post.

more cores vs memory bandwidth

[tomz16]

Unfortunately just putting more cores on a chip will eventually reach a breakeven point where memory bandwidth becomes the limiting factor. If you can't fit it on the core's cache, you have to communicate with the outside world.

As it stands, a Core2 is only marginally faster than a single core for the multithreaded scientific computations I work on. I would bet that a quad core would show little to no advantage over a dual core.

-Tom

Re:more cores vs memory bandwidth

[julesh]

Note that the quad cores have a faster FSB, so you probably would see some improvement with them.

100% from a Youtube video?!?!

[ender-]

100% just to play a Youtube video? Man, Intel procs must really suck!

Seriously though, I think this is just something that Intel marketing decided to spout out in the hopes that nobody would actually check.

I've got a laptop with a Turion ML-40 [2.2Ghz] (32bit WinXP), and I just tested this. Even with the CPU throttled down to 800Mhz, the youtube videos only used 50% CPU, not even enough to bump the processor up to the next speed step.

If Intel CPU's really pegged at 100% just to play a flash video that an 800Mhz-throttled AMD can handle, they'd be having serious problems right now.

I'm sure this isn't the case. But it shows how careful you have to be when dealing with anything Intel is saying about it's future products.

/Really want a Core 2 Duo machine. :)
//Still love AMD & hope they come out swinging.

Headline good for a chuckle

[amliebsch]

The article is boring, but I read "IDF" as "Intel Defensive Forces" and imagined a division of Intel-powered robo-jews militarily occupying Santa Rosa, an amusing thought.

"Swank"

[Anonymous Coward]

Umm... hello? English must be a second language I guess? From dictionary.com...

Swank -
dashing smartness, as in dress or appearance; style.

In other words, the Voodoo PC bajillion core system they commented on was, in other words, dashing smartness, as in dress or appearance; style...

So that would be in-fact, WTF they were talking about, I assume.

Re:"Swank"

[julesh]

You just need a better dictionary.

swank

informal
[...]
      noun - behaviour, talk, or display intended to impress others.

(www.askoxford.com)

Re:"Swank"

[Anonymous Coward]

PUHLEEZ... get a life

Re:"Swank"

[julesh]

And you think reading a discussion board and randomly insulting people who have posted there is somehow better than pointing out where somebody's criticism of an article is based on an incorrect assumption?

Frankly, I think you're the one who needs to "get a life".

Good for quantum system engineering

[sidles]

Our Quantum System Engineering (QSE) Group has immediate need of 80-core teraflop/terabyte processing ... it's just what we need to compute the real-time dynamics of imaged biomolecular structures ... on the desktop.

Intel is batting for the fences

[chipace]

80 cores on a chip and optical interconnects sounds great...but I wish they would talk more about the end application goals (i.e. a system that does 1080p ray-tracing and has 100% speech recognition). It's great that they are pushing the design limits as they are, but without clear vision of how the technology is to be used, it's likely that it will miss the mark.

I was hoping to hear about a single die with cpu/northbridge/southbridge/gpu all integrated (and for mobile use)... that would certainly turn the computer market on it's head. Nvidia knows this already, and has everything but the cpu integrated. Intel not buying Nvidia is the stupidest thing I have ever seen.

Not 80 Cores! Just 80 FPUs.

[Anonymous Coward]

In yet another article in Ars Technica we read that Intel is look to an 80 core chip.

Otellini talked about an 45nm chip with 80 FPUs. (Experimental, intended for image and speech recognition and such.) One FPU is way smaller and simpler than an entire CPU core.

I'm very interested what they plan for interconnect between those FPUs and the workload management. Very very parallel tasks (loooong FP vectors) probably, not much dynamic scheduling. This is a dedicated co-processor.

Not really a contradiction

[nobodyman]

"Paul Otellini, Intel's president and CEO, kicked off this season's IDF by coining the phrase "It's what's inside that counts", and spoke about why processing power matters again"

But then this in another article covering the same event:
"Otellini briefly responded to concerns that Intel's first quad-core packages are simply "glued-together" dual-core processors while AMD is working on a native, single-die quad-core chip. "So what?," said Otellini, adding, "The public doesn't care what's inside a processor."

Personally, I think it's a silly catchphrase, but that aside I don't necessarily see these two quotes as a contradiction given the contexts he gave them in. I think Otellini was saying was that raw cpu performance is still relevant. That said, if Intel can create a quad-core that performs very well and doesn't get too hot, I could care less if it's four cores on one die, 2 glued dual-core, or four single-core dies glued together, or a legion of microscopic gremlins that are really good at math.

That AMD is going for a single die approach is no surprise: If you glued two high-end rev.f opterons 240 watt monster. They pretty much are forced to do a redesign. By contrast, the "glued" Intel quad core uses 120watts (still alot, but it's conceivable that you could cool the thing). They could be more efficient by going on a single die, but this approach allows them to get to market sooner, which is fine by me.

Re:Not really a contradiction

[chromozone]

"That said, if Intel can create a quad-core that performs very well and doesn't get too hot, I could care less if it's four cores on one die, 2 glued dual-core, or four single-core dies glued together, or a legion of microscopic gremlins that are really good at math."

True of course. When I first wrote post I was mindful of a recent Tom's Hardware review that had the quad core doubling up in watts used (compared to Core 2 Duo)and that will sell for around 1000 USD.

Core 2 Duo was such an exciting thing to see come around. Looking back on the Pentiums and all the heat and reduced efficiency for that sake of selling the MHz (and then the whooping AMD dished out on Intel) and you have to wonder what the heck went wrong. A Pentium 4 1.8 was the last cpu I would buy from Intel. A long time ago people were urging Intel to see the Pentium M as a way to the future.

I just hope Intel doesn't go wild pushing cores for the sake of pushing cores along with the the video card makers making bigger and more power hungry toasters. To be fair though, Intel people have said dual core is where the action will be for a while and pretty much de-emphaiszed the first quad cores.

linky on the 45nm quad core

[maynard]

Here's what Tom's Hardware had to say about quad-core single die at 45nm:

The climax of Intel's move to 45 nm will obviously be processors with as many as eight cores on a die. Technically, the Yorkfield and Harpertown cores are pretty similar, while Yorkfield steps into the desktop space and Harpertown harpoons the enterprise customer. Both will be 45 nm parts, with four cores and as much as 12 MB L2 cache. We can't say how the geometry of this chip is going to be and we suppose Intel is not entirely sure yet either.

I don't have a dog in this fight. I'll buy whatever has the best price/performance.

Hyper Transport

[Joe The Dragon]

that sound like a good idea for a htx card / socket add on but intel does not use Hyper Transport

Re:Seems 80 cores isn't enough to resist the /. ef

[toddestan]

Actually, I find when loading a web page, my web browser works more on the model of:

1. Wait until it has enough data to do something.
2. Do it as quickly as possible.
3. Go back to step 1 until the web page is completely rendered.

The net effect is that if the web browser has the CPU to itself, it will be constantly spiking the CPU to 100% for short amounts of time. If you are doing something like requires a bit of CPU usage, like playing a game, you'll notice it skip and jitter when a webpage is loading. Most CPU monitoring programs smooth out these bumps (including Windows task manager), so it looks like the web browser never grabs more than 20% or so.

amd/ati

[CaptainNerdCave]

i haven't been paying close attention to any news about amd in the last couple of weeks, but i am expecting that they may be looking to the portable market in the near future.

why should amd be concerned with intel building excellent workstation cpu's when they can (economically) build a portable, all-inclusive board with a high-end igp and swift dual-core cpu?

how much longer until we see dual-core gpu's that aren't two cards glued together?

if intel can't get well-coordinated with nvidia and amd offers its best gpu's as a integrated packages (assuming they maintain quality), amd may continue holding a place of honor with gamers. if that were the case, perhaps we could expect the next gen of gaming consoles to have at least one offering made solely by amd. not only that, but as long as they improve the opteron to keep up their share of the server market, they could grab a big piece of the portable computing market and be sitting comfortably.

They must be idiots!

[jack_n_jill]

Spending big money to build a fab is Israel in idiotic! In the next war the Arabs will likely have longer range and more accurate missiles. The fab will make an inviting target. Israel is making no serious efforts at a just peace; giving the Palestinians equal rights and stopping the theft of Palestinian land. Another war is inevitable.

Silicon of the learned elders of Zion!

[Anonymous Coward]

One must wonder if the new middle-east manufacturing facility will also be built on occupied palestinian land like the existing fab is? That would make Intel a legitimate target of arab/muslim resistance.

The tremendous amount of waste water (used to purify air for fab cleanrooms) is currently disposed targetedly to contaminate water resources available to palestinians so they become ill and their agricultural product becomes un-marketable. Intel is party to a crime of ethic oppression.

'Coining' the phrase?

[Junta]

It says he started by coining the phrase it's what's inside that counts. I coulda sworn *I* coined the prhase trying to get girls to go out with me back in high school!

How the hell can someone claim a phrase is being coined when it's such a generic phrase used everywhere?