Intel's new strategy.

Cookie Monster

Diamond Member
May 7, 2005
5,161
32
86
Link

In continuation to our previous news reports we would like to make a few comments about Intel?s strategy concerning the introduction of new processors and architectures. As you already know, Core (Conroe) architecture will live as is at least until 2008. Then a new architecture will come to replace it, and although its name has been kept secret ever since, we knew it would be a 0.045micron process.

So, Intel?s chief executives admitted to their shareholders that they would from now on introduce new processor architecture every two years. Of course, the innovations we are talking about here will not be as revolutionary as the changes that we all witness during the transition from NetBurst to Core. However, they promised to have a good reason to use the ?new architecture? expression in 2008 already. All platforms released after Core will have one thing in common: they will maintain optimal balance between performance and power consumption.

The slide shows that 2008 will bring us Nehalem platform, which name comes from Intel?s earlier developments that have never really come out. In other words, it will be a completely different Nehalem than the one they were planning a few years ago :) The platform to come out in 2010 will be called Gesher. The processors developed within this platform concept will be manufactured with finer 0.032micron technological process.

The next slide explains Intel?s priorities in the platforms development. First of all, there will be unified architecture for all three major market segments: desktop, mobile and server. This way they can reduce the R&D costs and take less time to design new processors. A great example of this unified architecture is Intel Core that is coming out this summer.

To speed up the development of new platforms, Intel will combine the efforts of several engineering teams working in parallel. Moreover, the new architecture will not be tied up to the introduction schedule for the finer production technologies. They will use modified processors based on the previous-generation architecture to run in the new production norms. Once the new production technology has been polished off enough, they will begin the transition for the new processors. In fact, this evolution plan is already working:

* Presler turned into a ?finer? version of the Smithfield with separate cores when it moved to 0.065micron manufacturing process. A year later Conroe processors based on totally new architecture start coming off the 0.065micron production lines.
* The 0.045micron Conroe successor aka Penryn will be announced in 2007. Only in 2008 the new processors on Nehalem architecture will start using 0.045micron technology.
* 0.032micron Nehalem model aka Nehalem-C is scheduled to come in 2009. But only in 2010 we will see 0.032micron Gesher processors with new architecture.

We hope that this plan will help Intel reduce the time it usually takes them to prepare for the new product launch. It would be real great if the chipsets could remain compatible with at least one more generation of processors before phasing out. This way you won?t need to replace the mainboard every year when you upgrade your CPU: once every two years will do just fine.

By the way, our colleagues from The Inquirer site have also mentioned the name of the next CPU to come out after the desktop quad-core Kentsfield. This will be a 0.045micron Bloomsfield processor. It will have single-die design, i.e. all four cores and 8MB of cache memory will be pout into a single piece of silicon. The predecessor, Kentsfield processor, will consist of two halves, each of them being an analog to Conroe processor with 4MB L2 cache. Bloomsfield processor will features a shared L2 cache, and each core will be able to access the data stored there.

Discuss.
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
What is to discuss???

I mean it is similar to vid card methodology, but even they say what will be considered new archiotecture will not be as revolutionary as most of us think of new architecture....ie like conroe

Until they define better what will constitute a "new architecture" there is pretty much not much to talk about.....
 

Fox5

Diamond Member
Jan 31, 2005
5,957
7
81
I'm guessing the new architectures will be more like the shift from Pentium 3 to Pentium M to Core Duo to Conroe or from Northwood to Prescott (which was fairly significant, even if it didn't amount to much due to a design failure).
 

dmens

Platinum Member
Mar 18, 2005
2,275
965
136
fyi, "new architecture" means a complete rework of the overall architecture, as opposed to tweaks and steppings. In this context, even a relatively small change, i.e. dothan to yonah is considered a new architecture, along with complete reworks, such as the previous P3 to P4. I'm pretty excited about gesher, and not just because it has a funny name.
 

Praxis1452

Platinum Member
Jan 31, 2006
2,197
0
0
Originally posted by: Corporate Thug
Originally posted by: xtreme26
it would suck if we would have to change the motherboard everytime a new cpu comes out

i wouldnt mind changing every two years...

I sure as hell would! -_-
 

Griswold

Senior member
Dec 24, 2004
630
0
0
Reminds me of the GPU wars, when they announced "new goodies" every 6(?) months and failed to live up to the expectations of the crowd around here. Reality seems more like 12 months.

I believe it when I see it. Though, 2 years would fit perfectly into my upgrade habit.
 

bunnyfubbles

Lifer
Sep 3, 2001
12,248
3
0
Motherboards get better just as quickly as every other component, why people complain about it is beyond me, they aren't even the more expensive components like the CPU and GPU
 

BrownTown

Diamond Member
Dec 1, 2005
5,314
1
0
we'll its pretty well known that Nehalem was the new architecture that Intel wanted, but after the Presscott failed so miserably Intels hands were forced and they had to take the best they had (Pentium M), and get a new core out quick, so now we have Conroe, but its important to note that Conroe is NOT the arcitecture of the future for Intel, that is Nehalem, Conroe is designed only to fill the gap between Netburst and Nehalem. So, I would expect the Conroe to Nehalem jump to be pretty signifigant, and noboy knows jack about Gesher, so I can't really comment on how radical that will be. Also, Intels 32nm (or possibly a revised 45nm) will include metal interconnects, and high-k gate dielectrics which have a leakage of 1/1000 the current amount. This was origionally planned for 45nm, but was dropped so they could meet 45nm deadlines. So, leakage power concerns should be lessened, IF Intel can deliver.

The biggest thing I see in this news story is that Intel is still over a year away from releasing a single die quad core. Having all 4 cores on a single die is a very good thing for performance.
 

Maximilian

Lifer
Feb 8, 2004
12,604
15
81
Originally posted by: Duvie
What is to discuss???

I mean it is similar to vid card methodology, but even they say what will be considered new archiotecture will not be as revolutionary as most of us think of new architecture....ie like conroe

Until they define better what will constitute a "new architecture" there is pretty much not much to talk about.....

Yea, "new architecture" really does need a decent definition. The K8 is apparently a new architecture, but its really a K7 with lots of extra goodies, the pentium m architecture is a beefed up pentium III. It depends on how its defined, personally i think if somthing has a certain amount of changes to it and these changes add up to a certain increase in performance at the same clock speed as the old architecture then it should be considered "new".

Of course that definition falters with netburst CPU's, they would suck at the same clock speed as other architecture cpu's, so its a difficult thing to define.
 

Lonyo

Lifer
Aug 10, 2002
21,938
6
81
Originally posted by: bunnyfubbles
Motherboards get better just as quickly as every other component, why people complain about it is beyond me, they aren't even the more expensive components like the CPU and GPU

Yup.
I went from XP1900+ on a VIA chipset mobo (KT333 maybe), then I upgraded to a 2500+ Barton, which may have worked on my VIA board, but I went with a better mobo so I could overclock more (NF7).
I will possibly do the same again if I upgrade my current proc (3200+ on MSI K8N).
Also it means you can keep the existing proc/mobo combination and use it to build a secondary PC, since most people have RAM and gfx cards lying around as well.
 

Avalon

Diamond Member
Jul 16, 2001
7,571
178
106
Originally posted by: dmens
fyi, "new architecture" means a complete rework of the overall architecture, as opposed to tweaks and steppings. In this context, even a relatively small change, i.e. dothan to yonah is considered a new architecture, along with complete reworks, such as the previous P3 to P4. I'm pretty excited about gesher, and not just because it has a funny name.

What does Gesher entail, that you know of so far?
 

Cooler

Diamond Member
Mar 31, 2005
3,835
0
0
Originally posted by: Avalon
Originally posted by: dmens
fyi, "new architecture" means a complete rework of the overall architecture, as opposed to tweaks and steppings. In this context, even a relatively small change, i.e. dothan to yonah is considered a new architecture, along with complete reworks, such as the previous P3 to P4. I'm pretty excited about gesher, and not just because it has a funny name.

What does Gesher entail, that you know of so far?

My guess is Nehalem will have a memory controller as Intel could not fit it in time for core architecture.
Gesher may Have built in controller for PCI-E I know amd was talking about that as next thing to add.
 

Viditor

Diamond Member
Oct 25, 1999
3,290
0
0
Originally posted by: Avalon
Originally posted by: dmens
fyi, "new architecture" means a complete rework of the overall architecture, as opposed to tweaks and steppings. In this context, even a relatively small change, i.e. dothan to yonah is considered a new architecture, along with complete reworks, such as the previous P3 to P4. I'm pretty excited about gesher, and not just because it has a funny name.

What does Gesher entail, that you know of so far?

I assume that we are talking about 2009 and Intel's CSI for Nehelam (which would mean on-die mem as well). It should also give them a hardware based IOMMU to finally boost their 64bit speed up to AMD standards...
I have a feeling that the extra FPUs on K8L will bring AMD in line with the Conroe core, so it will be interesting to see what AMD can bring with K10. It better be substantial because I've a feeling that BT is correct about Nehelam being Intel's real goal!

As to Gesher, probably only dmens knows...and I doubt he's willing to risk his job to tell us. ;)
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
Originally posted by: Praxis1452
Originally posted by: Corporate Thug
Originally posted by: xtreme26
it would suck if we would have to change the motherboard everytime a new cpu comes out

i wouldnt mind changing every two years...

I sure as hell would! -_-



Better then what Intel did to users over the last 2-3 years...maybe 3 mobo/chipset changes...
 

Cooler

Diamond Member
Mar 31, 2005
3,835
0
0
Originally posted by: Duvie
Originally posted by: Praxis1452
Originally posted by: Corporate Thug
Originally posted by: xtreme26
it would suck if we would have to change the motherboard everytime a new cpu comes out

i wouldnt mind changing every two years...

I sure as hell would! -_-



Better then what Intel did to users over the last 2-3 years...maybe 3 mobo/chipset changes...

I think it was 4 or 5 .
 

dmens

Platinum Member
Mar 18, 2005
2,275
965
136
merom isn't any kind of halfway gap in between P4 and nehalem, it is a fully revamped architecture in its own right. Just because it met schedule does not mean features were skimped. Even with the FSB merom is faster (and more flexible) than AMD's solutions from laptop to DP. And once AMD moves to 65nm and depending on how the knobs were tweaked for K8L, merom should still be able to hold its own, imo.

as for nehalem being intel's real goal, that is something that needs to be defined, but if real goal means continuous performance leadership, then sure why not heh. Depends what K10 is and when it gets into the market. If TheInq is right about K10 it might end up matching against 32nm nehalems.
 

Kakumba

Senior member
Mar 13, 2006
610
0
0
Hey, every 2 years seems perfect to me. a full upgrade every 2 years is what I consider to be about right for a desktop. anyways, if this will help clear up chipset support, which is something that has confused me about Intel CPUs of late (I havent the time to work out what mobo I have to buy to support which CPU). Oh well, wait and see time.
 

aka1nas

Diamond Member
Aug 30, 2001
4,335
1
0
The new Nehalem sounds like it will be interesting. Remember when it was going to be a 10Ghz Netburst chip just a few years ago.:p
 

bunnyfubbles

Lifer
Sep 3, 2001
12,248
3
0
Originally posted by: Lonyo
Originally posted by: bunnyfubbles
Motherboards get better just as quickly as every other component, why people complain about it is beyond me, they aren't even the more expensive components like the CPU and GPU

Yup.
I went from XP1900+ on a VIA chipset mobo (KT333 maybe), then I upgraded to a 2500+ Barton, which may have worked on my VIA board, but I went with a better mobo so I could overclock more (NF7).
I will possibly do the same again if I upgrade my current proc (3200+ on MSI K8N).
Also it means you can keep the existing proc/mobo combination and use it to build a secondary PC, since most people have RAM and gfx cards lying around as well.

Yeah, I remember going from KT333 to nF2 Ultra as well (NF7-S). Using the same ram and same CPU, there was a noticeable difference in performance ? then I was able to swap in DDR 400 on top of it and had many more new overclocking options (as well as support for faster CPUs), not to forget soundstorm and decent onboard lan, even SATA...
 

Viditor

Diamond Member
Oct 25, 1999
3,290
0
0
Originally posted by: dmens
merom isn't any kind of halfway gap in between P4 and nehalem, it is a fully revamped architecture in its own right. Just because it met schedule does not mean features were skimped. Even with the FSB merom is faster (and more flexible) than AMD's solutions from laptop to DP. And once AMD moves to 65nm and depending on how the knobs were tweaked for K8L, merom should still be able to hold its own, imo.

as for nehalem being intel's real goal, that is something that needs to be defined, but if real goal means continuous performance leadership, then sure why not heh. Depends what K10 is and when it gets into the market. If TheInq is right about K10 it might end up matching against 32nm nehalems.

I think by real goal he means that Nehelam was to have all of what Merom has, plus the CSI.
As to Merom being faster than Turion X2, I'm sure it's possible and even probable...but we really don't know yet so I won't let you get away with that one m8. :)
Servers are the "weakest link" for Intel...1P and 2P should be faster in 32 bit, but I am going to bet that 64 bit still goes to AMD. 4P and higher is AMD hands down...
 

dmens

Platinum Member
Mar 18, 2005
2,275
965
136
Originally posted by: Viditor
I think by real goal he means that Nehelam was to have all of what Merom has, plus the CSI.
As to Merom being faster than Turion X2, I'm sure it's possible and even probable...but we really don't know yet so I won't let you get away with that one m8. :)
Servers are the "weakest link" for Intel...1P and 2P should be faster in 32 bit, but I am going to bet that 64 bit still goes to AMD. 4P and higher is AMD hands down...

well, nehalem is somewhat more than merom with csi, but ill let that sit. In regards to turion X2 vs intel cores, that's just a engineering estimate based on the knob-tweaking we've seen with the 90nm K8. If dual turions debut on 65nm, then it is a different story, but there's only so much you can do with a transistor these days... plus the layout restrictions and metals just get worse. Yech.

As for 64-bit code performance, guess you'll have to wait and see... hehehe.
 

Accord99

Platinum Member
Jul 2, 2001
2,259
172
106
Originally posted by: Viditor
I think by real goal he means that Nehelam was to have all of what Merom has, plus the CSI.
As to Merom being faster than Turion X2, I'm sure it's possible and even probable...but we really don't know yet so I won't let you get away with that one m8. :)
Mobile processors have a certain power limitation. Right now, a Yonah 2.16GHz uses less power than a low-voltage single-core Turon MT 2.2GHz under load. It's no wonder AMD will have to resort to 1.075v to try to be competitive.

http://www.silentpcreview.com/article313-page5.html

Servers are the "weakest link" for Intel...1P and 2P should be faster in 32 bit, but I am going to bet that 64 bit still goes to AMD. 4P and higher is AMD hands down...
Higher than 4S goes to Intel by default, since there are no remotely decent 8S Opteron product and none from HP or Sun at all. Meanwhile IBM's X3 based servers using Xeon MPs goes to 32S and scales well. Fujitsu-Siemens and Unisys also produce >4S servers using Xeon MPs.

IBM also has the fastest 4S x86 server in the all-important TPC-C benchmark and is considerably faster in all aspects than Intel's Truland chipset. Just because Intel makes a mediocre 4S chipset doesn't mean somebody else can't improve on it.