Intel Nehalem Working at 3.2GHz Pictured

[Jacky]

Fear, Uncertainty and Doubt. But who's spreading that S***, people just want hardware news, fair and square... that BS makes me sad. If you are right, shintai, someone should get charlie d. in jail for that ongoing defamation/libel.

So the 3.2ghz system would have been a uniprocessor machine showing as 8 physical processors (?) and not skulltrail (2*4cores)? And the other one 2*Nehalem@2.13ghz?

[gOJDO]

Why do you even bother with Charlie and the rest of the clowns from the_INQ.

[Shintai]

So the 3.2ghz system would have been a uniprocessor machine showing as 8 physical processors (?) and not skulltrail (2*4cores)? And the other one 2*Nehalem@2.13ghz?

Yes, one is single Quad with SMT on. The other is a dual Quad with SMT on.

So one would be a Xeon/Skulltrail and the other one a desktop so to say.

[Zytek_Fan]

They just saw a different system. DP system. Like the 3.2Ghz was the only one...Of you note their "edvidence" is a 16 logical CPU system, the other is an 8.

Theinq just showing their dumbass stupidity as usual in the material incarnation of the retard Charlie.

Charlie @ TheInq is well known to be an AMD fanboy


Though it should be a rule of thumb to take anything from TheInq with a grain of salt.

[Donnie27]

Charlie @ TheInq is well known to be an AMD fanboy


Though it should be a rule of thumb to take anything from TheInq with a grain of salt.

More like a factory full of salt LOL! Yes, he's been known as an AMD fanboy for years.

[Zytek_Fan]

More like a factory full of salt LOL! Yes, he's been known as an AMD fanboy for years.

A grain of varying sizes


I believe that Nehalem was shown working at 3.2ghz. Why? Because Intel has been working on it for quite some time, and they showed WORKING silicon back at IDF in the fall.

[_Lone_Wolf_]

Can anyone link to evidence to demonstrated Nehalem clockrates rather than personal attacks on Charlie (funny he receives the same treatment as an Intel shill at AMDZone). I'm browsing Intels pressroom at the moment but can't find anything. Did anyone from here actually attend the IDF and see for themselves?

What I find odd are his suggested clockrates. With CSI running at initially running 2.4GHz (4.8GTs) I would have expected a similar arrangement to hypertransport with a 200 or 400 MHz base clock and the cores running off a multiplier from that, which doesn't mesh well with 2.13 or 2.53GHz(?) - But then I suppose a 266 base clock would take care of that...

[Shintai]

Can anyone link to evidence to demonstrated Nehalem clockrates rather than personal attacks on Charlie (funny he receives the same treatment as an Intel shill at AMDZone). I'm browsing Intels pressroom at the moment but can't find anything. Did anyone from here actually attend the IDF and see for themselves?

What I find odd are his suggested clockrates. With CSI running at initially running 2.4GHz (4.8GTs) I would have expected a similar arrangement to hypertransport with a 200 or 400 MHz base clock and the cores running off a multiplier from that, which doesn't mesh well with 2.13 or 2.53GHz(?) - But then I suppose a 266 base clock would take care of that...

With all respect, did you even bother at all to check the first page?

[_Lone_Wolf_]

With all respect, did you even bother at all to check the first page?

Which is being disputed.

[Shintai]

Which is being disputed.

By what? The completely DIFFERENT system Charlie posted and linked to?

So disputed? No...just some clueless comparision.

[_Lone_Wolf_]

I'm not having an endless circular diversionary argument about Charlie D.

What prompts me to keep a healthy dose of skepticism is Nehalem will have a third more logic transistors relative to a quad Penryn which will endure higher sustained utilisation under heavy multithreading workloads. It seems unavoidable to having a higher thermal density. The 3 MB cache deficit contributes very little to saved heat output.

To acheive an equal clock to the highest current official speed bin with what is claimed A1 revision silicon under an equivilent TDP would be astonishing. I'll reserve judgement till Intel disclose figures.

(I'd also love to read JumpingJacks thoughts on the subject)

[gOJDO]

I'm not having an endless circular diversionary argument about Charlie D.

What prompts me to keep a healthy dose of skepticism is Nehalem will have a third more logic transistors relative to a quad Penryn which will endure higher sustained utilisation under heavy multithreading workloads. It seems unavoidable to having a higher thermal density. The 3 MB cache deficit contributes very little to saved heat output.

To acheive an equal clock to the highest current official speed bin with what is claimed A1 revision silicon under an equivilent TDP would be astonishing. I'll reserve judgement till Intel disclose figures.
(I'd also love to read JumpingJacks thoughts on the subject)

The TDP is not an issue since Penryn has rated TDP much higher than it should have. That's why we see a 65W TDP Wolfdale wasting 2/3 of the energy wasted by a 65W TDP Conroe. That's why we see a 4.5GHz Wolfdale's on air, which is a mission impossible with a Conroe.

The high rated TDP of Penryn is kind of a backup which Intel is going to (ab)use for the next generation of CPUs. So, Intel are going to introduce same clocked Nehalem CPUs with the same TDP as Penryn and like AMD, Intel are going to claim xx% performance improvement in the same power envelope.
Of course, Nehalem is going to waste more energy because it has much more logic than Penryn, but it has TDP headroom and it has a larger package(larger surface) which will improve the heat transfer from the CPU to the CPU cooler.

The speed of the CPU depends of other, more important factors. One of the factors is the pipeline length. The more stages the CPU has means that each stage will have less complex logic. The less complex logic per stage means faster clocks. Penryn has 14 stages, while Nehalem has 16 or 17. Furthermore, another factor is the desing on a given process. Nehalem is designed up from the ground for the new 45nm process. Unlike Nahelm, Core2 was designed for the 65nm and Penryn as an improved shrink is not optimized for the new process.

IMO, Nehalem will have at least the Penryn frequencies and chances are(again IMO) that it will have slightly higher clocks. For example 15% higher. But we'll have to wait a year before we can see the high clocked parts of Nehalem. Intel will need a decent amount of highly binned Nehalem CPUs for release. Before the release of Nehalem I'm also expecting faster Penryns, ex: 3.6GHz Wolfdale and 3.4GHz Yorkfield/Harpertown.

[Donnie27]

@ donnie27

shure, but that doesnt mean nehalem can reach that, too

i for instance still havent seen a peryn windows stable @ 8 ghz+(even on one core), which can (of course) be made by some pentium 4s

but it looks promising, should be a real monster if it can reach 4.5-5ghz @ LN2

Where'd 8GHz come from!? If Intel based this on what AMD or their old processors were doing/performing, then they'd leave Nehalem at 3GHz and it'd still be the fastest.

Please go back and watch the video again?

http://youtube.com/watch?v=5k19pa970Is

That's a Dual Socket system.

Check this one out!

http://youtube.com/watch?v=uNXBmdLR8QI&feature=related

[GoldenTiger]

What ever happened to the 8-core (physical... 16 SMT in Task Manager) Nehalem? Am I misremembering it? Is it a different CPU, and Nehalem means the 4-core one with SMT (8 in task manager, 4 physical)? Did it never exist?

[Donnie27]

What ever happened to the 8-core (physical... 16 SMT in Task Manager) Nehalem? Am I misremembering it? Is it a different CPU, and Nehalem means the 4-core one with SMT (8 in task manager, 4 physical)? Did it never exist?

They showed a Dual socket Nehalem and that's why you saw 16 threads. There was a Demo of an 8 core talked about I don't remember if it was shown or not. The 8 core one is the 32nm though.

[GoldenTiger]

They showed a Dual socket Nehalem and that's why you saw 16 threads. There was a Demo of an 8 core talked about I don't remember if it was shown or not. The 8 core one is the 32nm though.

Gotcha, thanks . I kept associating Nehalem with octo-core for some reason... , guess not for awhile yet then.

[Donnie27]

Gotcha, thanks . I kept associating Nehalem with octo-core for some reason... , guess not for awhile yet then.

Here's the kind of stuff that was messing up folks. Let me say right up front before one of the Experts jumps me. There will be an 8 or Octo-Core at 45nm, it will not be meant for DeskTops.

http://forums.slizone.com/index.php?showtopic=18953

It's called Nehalem EX (Beckton). As you can see from the QPI links it is a big-time Server Processor meant for 4 or more Sockets. I was only talking about 8 core processors for the next version of Skulltrail will be at 32nm from what I've seen.

http://www.engadget.com/tag/octo-core/

[

B]Apple Mac Pros: now with 8-cores[/B]
Posted Apr 4th 2007 8:41AM by Thomas Ricker
Filed under: Desktops

So long Woodcrest, hello Covertown! Rumored since October, Apple finally got around to loosing an octo-core Mac Pro -- that's 8-cores of Intel processing power spread across 2 of Intel's latest Clovertown quad-core chips. Arriving just in time for Adobe's Intel optimized Creative Suite 3.0. Quad-core models are still available with processor speeds maxing out at 3GHz. Prices start at $2,499 for 2.66GHz quad-core action or tack on an additional $1,498 for a total of 8-cores running at 3GHz. Shipping now -- hoozah!

8 cores yes, Octo-Core, no. Many sites had news like "Intel's new Octo-core platform, the Skulltrail. It's very sexy stuff, ...". That's very misleading. What's Sexy is Skulltrail #2 with two Westmere's doing 24 threads or running two OS's at full speed multitasking in each LOL!

[Zytek_Fan]

8 cores yes, Octo-Core, no. Many sites had news like "Intel's new Octo-core platform, the Skulltrail. It's very sexy stuff, ...". That's very misleading. What's Sexy is Skulltrail #2 with two Westmere's doing 24 threads or running two OS's at full speed multitasking in each LOL!

Ah yes, the Smackover platform?

I still stand by my opinion that it should be called Smack Down

[XS Janus]

SmackDown FTW!

[Blauhung]

Here's the kind of stuff that was messing up folks. Let me say right up front before one of the Experts jumps me. There will be an 8 or Octo-Core at 45nm, it will not be meant for DeskTops.

http://forums.slizone.com/index.php?showtopic=18953

It's called Nehalem EX (Beckton). As you can see from the QPI links it is a big-time Server Processor meant for 4 or more Sockets. I was only talking about 8 core processors for the next version of Skulltrail will be at 32nm from what I've seen.

http://www.engadget.com/tag/octo-core/

8 cores yes, Octo-Core, no. Many sites had news like "Intel's new Octo-core platform, the Skulltrail. It's very sexy stuff, ...". That's very misleading. What's Sexy is Skulltrail #2 with two Westmere's doing 24 threads or running two OS's at full speed multitasking in each LOL!

westmere desktop and dp server will probably 6 core

Ah yes, the Smackover platform?

I still stand by my opinion that it should be called Smack Down

Smackover is the succesor to bonetrail not skulltrail.

[Donnie27]

westmere desktop and dp server will probably 6 core

Smackover is the succesor to bonetrail not skulltrail.

OK, that's what I meant. That's why I said 24 threads. That's 12 cores and 12 logical cores.

[Donnie27]

Ah yes, the Smackover platform?

I still stand by my opinion that it should be called Smack Down

Yepp, wrong name but correct result, it will "Smackdown" the competition. I'm thinking about the one after Smackover.

http://www.dvhardware.net/article23691.html

Intel is working on a new high-performance platform called Smackover. This platform will replace Skulltrail in Q4 2008:
It's name: Smackover. This server motherboard, rebranded as a halo desktop product, will only compete at the ultra-high-end enthusiast segment -- a market AMD recently exited.

Also like Skulltrail, Smackover is essentially a server motherboard with overclocking abilities for the upcoming Nehalem processor, also slated for release in Q4 2008. The board will use the Tylersburg server chipset with ICH10R.

Intel designates its product codenames off various points of interest on U.S. maps. Smackover, as it would happen, is actually a small town in Arkansas. It had to be somewhere I suppose.

I was thinking of the one for 32nm but I'm not sure if it will have the same socket or not.

[duploxxx]

The TDP is not an issue since Penryn has rated TDP much higher than it should have. That's why we see a 65W TDP Wolfdale wasting 2/3 of the energy wasted by a 65W TDP Conroe. That's why we see a 4.5GHz Wolfdale's on air, which is a mission impossible with a Conroe.

The high rated TDP of Penryn is kind of a backup which Intel is going to (ab)use for the next generation of CPUs. So, Intel are going to introduce same clocked Nehalem CPUs with the same TDP as Penryn and like AMD, Intel are going to claim xx% performance improvement in the same power envelope.
Of course, Nehalem is going to waste more energy because it has much more logic than Penryn, but it has TDP headroom and it has a larger package(larger surface) which will improve the heat transfer from the CPU to the CPU cooler.

The speed of the CPU depends of other, more important factors. One of the factors is the pipeline length. The more stages the CPU has means that each stage will have less complex logic. The less complex logic per stage means faster clocks. Penryn has 14 stages, while Nehalem has 16 or 17. Furthermore, another factor is the desing on a given process. Nehalem is designed up from the ground for the new 45nm process. Unlike Nahelm, Core2 was designed for the 65nm and Penryn as an improved shrink is not optimized for the new process.

IMO, Nehalem will have at least the Penryn frequencies and chances are(again IMO) that it will have slightly higher clocks. For example 15% higher. But we'll have to wait a year before we can see the high clocked parts of Nehalem. Intel will need a decent amount of highly binned Nehalem CPUs for release. Before the release of Nehalem I'm also expecting faster Penryns, ex: 3.6GHz Wolfdale and 3.4GHz Yorkfield/Harpertown.

nice to see that you already already create some hard facts on nehalem while no decent NDA is talking about nehalem speed (wait a year for higher binnen nehalem you are talking about 2010 ), they are even still wondering about full availability date

i'll give some known facts and maybe better to several here who are already dreaming about there next configuration.

nehalem birth is slated for november 2008.....on paper and some that really want to spend to much money, real volume is only q1 2009 this only on 1p and 2p sockets, 4p and higher is even more then 6months later.

there is no speed-bump nor octa core foreseen in the intel roadmap unless the competitor comes with something unforeseen the only thing they talk about are some specific 1600fsb quads.

Now lets say that this is a rather strange move from Intel, the forecast on 1p and 2p is that they won't loose that much market share now that barcelona is available. but the market share gained with 4P tigertown is at very high risk now with release of 4p barcalona, even at a 2,3 GHZ speed they are able to keep performance (but way lower price and power) with the high-end 2,93ghz tigertown on TPC-sap benches and even Virtualization. not to mention the higher speed parts soon.

So intel tackles its own market exactly the same thing they did long time ago from p4-c2d to regain a fast market share but put allot of old stuff into the bin, they did it again from 65-45nm with 45nm not able to deliver and will do it again to new arch....they don't need AMD to loose share or profit .

last but not least, its nice to see those statements that nehalem will be much faster then penryn/harpert/tigert or whatever, lets see first some results... for sure if they calculate on bandwidth they will gain allot and performance for 4p and higher will also increase allot, other parts? still to be seen.

[gallag]

nice to see that you already already create some hard facts on nehalem while no decent NDA is talking about nehalem speed (wait a year for higher binnen nehalem you are talking about 2010 ), they are even still wondering about full availability date

i'll give some known facts and maybe better to several here who are already dreaming about there next configuration.

nehalem birth is slated for november 2008.....on paper and some that really want to spend to much money, real volume is only q1 2009 this only on 1p and 2p sockets, 4p and higher is even more then 6months later.

there is no speed-bump nor octa core foreseen in the intel roadmap unless the competitor comes with something unforeseen the only thing they talk about are some specific 1600fsb quads.

Now lets say that this is a rather strange move from Intel, the forecast on 1p and 2p is that they won't loose that much market share now that barcelona is available. but the market share gained with 4P tigertown is at very high risk now with release of 4p barcalona, even at a 2,3 GHZ speed they are able to keep performance (but way lower price and power) with the high-end 2,93ghz tigertown on TPC-sap benches and even Virtualization. not to mention the higher speed parts soon.

So intel tackles its own market exactly the same thing they did long time ago from p4-c2d to regain a fast market share but put allot of old stuff into the bin, they did it again from 65-45nm with 45nm not able to deliver and will do it again to new arch....they don't need AMD to loose share or profit .

last but not least, its nice to see those statements that nehalem will be much faster then penryn/harpert/tigert or whatever, lets see first some results... for sure if they calculate on bandwidth they will gain allot and performance for 4p and higher will also increase allot, other parts? still to be seen.

can we not have one Intel thread without you coming in bashing it?
(living in a glass house and throwing stones?)

[gOJDO]

nice to see that you already already create some hard facts on nehalem while no decent NDA is talking about nehalem speed (wait a year for higher binnen nehalem you are talking about 2010 ), they are even still wondering about full availability date

I've created nothing. I only interpreted what's already floating all over the internet and I expressed my opinion. The facts are:
1) Nehalem is based on Core2, but has 2 more pipeline stages than Core2.
http://www.realworldtech.com/page.cf...WT040208182719
2) Longer pipeline yields higher clocks.
http://www.microarch.org/micro36/htm...timumPower.pdf
http://systems.cs.colorado.edu/ISCA2...line_color.pdf
3) Intel already demonstrated a 3.2GHz Nehalem ES running.

http://www.hexus.net/content/item.php?item=12538
4) Penryn based CPUs have higher rated TDP than they should have.
Both QX6850 and QX9650 have rated TDP of 130W, but they have different power consumption/heat dissipation:

http://www.anandtech.com/cpuchipsets...spx?i=3137&p=6
5) Intel can release 3.4GHz Yorkfield and 3.6GHz Wolfdale any day.
http://www.xtremesystems.org/forums/...splay.php?f=56
Do you agree with the above?

i'll give some known facts and maybe better to several here who are already dreaming about there next configuration.

Please enlight us with the "facts", but please provide any kind of a proof.

nehalem birth is slated for november 2008.....on paper and some that really want to spend to much money, real volume is only q1 2009 this only on 1p and 2p sockets, 4p and higher is even more then 6months later.

Cool but I don't care a bout servers. I'm interested about the desktop non-"Extreme" variant with the highest performance/price ratio.

there is no speed-bump nor octa core foreseen in the intel roadmap unless the competitor comes with something unforeseen the only thing they talk about are some specific 1600fsb quads.

Nice, but what kind of quads? Core2 or Nehalem based? 1600FSB quad doesn't sound like an improvement over my 1066FSB Q6600, because I love OC-ing.

Now lets say that this is a rather strange move from Intel, the forecast on 1p and 2p is that they won't loose that much market share now that barcelona is available. but the market share gained with 4P tigertown is at very high risk now with release of 4p barcalona, even at a 2,3 GHZ speed they are able to keep performance (but way lower price and power) with the high-end 2,93ghz tigertown on TPC-sap benches and even Virtualization. not to mention the higher speed parts soon.
So intel tackles its own market exactly the same thing they did long time ago from p4-c2d to regain a fast market share but put allot of old stuff into the bin, they did it again from 65-45nm with 45nm not able to deliver and will do it again to new arch....they don't need AMD to loose share or profit .

The topic was about Nehalem, not about Tigertron vs Barcelona performance in TPC, Virtualization or whatever. If you have nothing constructive to say about Nehalem, don't turn this thread into another AMD vs Intel BS.

last but not least, its nice to see those statements that nehalem will be much faster then penryn/harpert/tigert or whatever, lets see first some results... for sure if they calculate on bandwidth they will gain allot and performance for 4p and higher will also increase allot, other parts? still to be seen.

Of course. But while there are no hard numbers nobody can stop us speculating, right?
So, what do you think about it?