Forget Huron River, 22nm Ivy Bridge on the Way

Code named "Chief River", this mobile platform will debut at CES 2012 featuring:

• 3rd Generation (and final) "Core i" Processors (before new architecture Haswell in 2013)
• Should be a compatible upgrade for Sandy Bridge notebook owners
• Shrink of Sandy Bridge to 22nm
• Focus on efficiency
• Thunderbolt integration
• 3rd Gen High-k Metal Gate Technology
• USB 3.0
• DirectX 11 enabled gpu
• GPU has 24 Shaders
• 3D Transistors

BTW, Haswell follows in 2013, and 16nm Rockwell in 2014.

 

I always look more forward to the ticks than the tocks anyways. Ivy Bridge should be a great refinement of Sandy Bridge.

 

you made me forget sandy bridge already!

 

22nm is when I get new hardware. But I'll be buying on the "Tock", rather than the tick. 65nm-->45nm-->22nm I saw no real benefit of upgrading to 32nm, as most of my 45nm C2D's were quite efficient at saving power, and have really good performance.

 

Damn!! I was going to pull my trigger on a sandy bridge laptop in the first half of 2011!! Now newer hardware will be available in the second half of 2011, bah, I guess the waiting game never stops

 

As long as the Sandy Bridge powered notebook you buy gives you access to the cpu, you should be able to upgrade to an Ivy Bridge once they come out (also if a bios update supporting it is released).

 

So Ivy Bridge and Sandy Bridge have the same socket? They changed that on the desktop side of things AGAIN if I have my facts straight...

 

Yes, Nehalem to Sandy Bridge is a new socket, but Sandy Bridge to Ivy Bridge is just a shrink, so they should be compatible. Just like when Core 2 went from 65nm to the first gen 45nm, they were still socket compatible.

 

So forget Arrandale, and now Huron River. I want to punch you.

 

LOL. 10char.

That is good news for some then. The ones who buy into SB can (hopefully) upgrade to IB.

 

Forget Breakfast, Lunch on the Way...

Forget Lunch, Dinner on the Way...

Forget Dinner, Breakfast on the Way...

 

No. Volume production starts in late 2011, which means the hardware won't actually be available until 2012. It should be quite good though: they'll be taking an already power efficient architecture and shrinking it down. We haven't seen such a thing since Penryn since Nehalem needed the die shrink just to be able to fit in mainstream laptops and its ultra-portable versions are still not as efficient as Penryn.

Also, if they can keep their promise of doubling GPU performance, Ivy Bridge should actually be competitive with mid-range GPUs.

 

lol that is exactly how I feel...

must have new technology!!!

 

I would think tocks will be better than ticks considering major architectural improvements. Nehalem did not have full tock/tick for notebooks. There is 45nm quad cores( Clarksfield) and 32nm dual cores( Arrandales). At least sandy bridge should bring dual/quads in both 32/22nm flavors.

I think 2720qm with 2.2ghz base clock and aggressive turbo should be good enough with fairly decent gfx. Hopefully we see thin/light laptops with quad cores. dual core sandy bridge is a disappointment considering we have arrandale at 2.8ghz while fastest SB will be at 2.7ghz. I guess intel wants to push dual core notebooks only to low end while quad cores should take over mid range/high end market.

 

I too wouldn't bet my shirt on an actual availability of ivy bridge within less than a year after SB is released... but, if in the end I'm proved to be mistaking, I further doubt mobile CPUs would be released so soon. Correct me if I'm wrong but I think the situation with Sandy Bridge is already not quite typical (for mobile chips not lagging behind desktop for release).

 

It depends on whether the chip is fundamentally a server chip or a client chip. Nehalem was a server chip -- they had to do a lot of binning and trimming to push the TDP down to where it can fit in notebooks. Sandy Bridge is a client chip so the server version (Sandy Bridge E*) won't be out until Q3 while the mobile version gets released at the same time as the desktop one. Ivy Bridge is just a die shrink of Sandy so it'll probably come out simultaneously again.

 

The reason nehalem didn't come out for laptops for so long was that they only produced the 45nm processor in a quad core configuration. Core 2 was deemed good-enough for low-end and mainstream applications because there was no competition from AMD. There weren't any power issues relative to the core 2. Intel only came out with the dual core nehalem with 32nm, and even then, they continued making their 45nm quad and no 32nm quad! This way they made a lot more money out of their investments in 45nm and core 2 production, all thanks to their superior developmental position vs AMD. If there was any sort of power envelope issue we wouldn't have the same sort of 45nm quads they released two years ago as the high end mobile processor.

 

I read that 45nm Auburndale (dual core Nehalem) was skipped due to the economy, and Westmere pushed up to January.

 

Well yeah. What I was trying to say was that there was no technical limitation to it; it was just a financial matter.

 

Yes, there were. The reason Intel only made quad cores at 45nm is because a laptop quad core could reasonably be expected to have a TDP of 45W+ while a dual core could not. Note that when all 4 cores are used, the Clarksfield CPUs run at extremely low frequencies and even when restricted to two cores, the frequencies are still quite low given the TDP. Intel could have made a 45nm dual core, but if they wanted it to compete with Core 2 Duo at the same TDP, they would need to reduce the clock speed to the point where performance would be comparable (Nehalem is better clock for clock, but not by that much).

Let me put it another way: the TDP of the more powerful Arrandales is 35W and this is with a 32nm process. What do you think the TDP of the Core i7 620M, Core i5 520M, Core i5 450M, etc. would be if they were made at 45nm instead? Nehalem would not work for mainstream laptops without the die shrink.

This I agree with. They did not need to make a 32nm quad and so they kept it 45nm.

 

22nm On the Way, thanks to upgraded factories.

DailyTech

Thinq

 

I just wanted to say I always enjoy these "Forget x, y is on the way" every few months. I await the day when the thread begins "Forget Moore's Law, ..." and I'm sure NBR will have all sorts of stuff on things beyond our current silicon fab methods.

 

the thread about sandy bridge is not even over that we get this

 

Sorry, but I am always looking ahead.

 

I'll be sad if Sandy Bridge doesn't last a year at least.

 

lol, Yeah i noticed that too... every single thread's just been "Forget this, ____ on the way"

 

45nm to 32nm and now 22nm... reaching the 16nm limit soon.

 

What 16nm limit:

Xbitlabs

 

What is the the point of waiting for the next gen. Once we get to dec 2011 people will simply say forget ivy bridge here comes 16nm!

Just buy now hahaha

 

But dood, why buy something now when it will just be obsolete a year from now when you wish you would have just kept on waitin'.

 

so i heard intel is releasing some sick stuff year 2030?

 

Haha, that's how technology rolls. You have to live in the moment once in awhile.

I'm not going to say it's completely wrong but I think that chart is a decent extrapolation that might be outdated.
A few more die shrinks beyond 16nm and we begin to bridge over to nanotechnology because of quantum tunneling (current chipmaking doesn't technically count as nanotechnology because it doesn't rely on quantum mechanics so much as classical electrical mechanics). Another problem we will eventually face is what's called a chip "blackout" where the wires are too small to safely handle the required current to power the whole chip, which means we'll have roughly the same number of transistors but on smaller dies, unable to us more than a certain power. Apart from that, yields become harder to improve since each fab shrink becomes harder to implement, something TSMC struggled with in its latest shrinks, causing delays for GPUs.

Source: I'm a senior EE taking a microelectronic device fab class and a VLSI design course. We mention the ITRS every so often.

Anyone interested in last year's roadmap and the near-future of lithography should look to here until December:
http://www.itrs.net/Links/2009ITRS/2009Chapters_2009Tables/2009_Litho.pdf

EDIT: Thanks Phil for deleting the multiple posts!

 

Mainstream quad-core notebooks... That just sounds great. Unfortunately, I won't be nuying a new laptop until at least Haswell, but it's good to know that Intel will have quad-core notebooks in my budget by then

 

Hmm. If quad cores will be the norm as dual cores are the norm currently, what will be the "enthusiast" variation of mobile CPUS?

Six cores? Eight?

 

lol moore's law... the guys at intel think they're gods but i don't think we'll get anything smaller than 16nm..

 

I think 11nm is generally regarded as the current barrier before quantum mechanics steps in. Then again, they once said we couldn't scale past 60nm, and here we are at 20.

 

Haha, I'd noticed the pattern here too, the recurrent "Forget whatever, something better is coming!" threads. Definitely the technological treadmill. And somehow it seems Jay is always the one who posts these threads. It's become part of the NBR environment in a way.

Yeah, that's what I figure. Why waste money on something that'll be obsolete soon? Sure, I've fallen off the far end of the roadmap (Santa Rosa 65nm for me still), but I'm sure eventually they'll hit some fundamental limits of physics and I'll upgrade then. Or if my computer gives out, I'll upgrade to whatever's available then. Might not be able to run whatever the current version of Notepad is if my computer lasts too long, but at least I'll be saving money!

Realistically though somewhere in the Sandy Bridge to Haswell timeframe is probably when I'll upgrade. Still not too thrilled about quad cores since I still pretty much use single-threaded applications, but already the Core i series has shown big architectural improvements and hopefully that'll continue. Other perks like USB 3.0 are nice, too (although really I'm surprised it didn't catch on sooner, I suppose we have Intel and somewhat AMD to thank for that).

 

Let's get to Sandy Bridge first before we start forgetting about Huron River.

 

Forget you. Sandy bridge is playing it old school.

 

Okay guys, let's try to focus on the topic at hand, shall we? Debating whether quantum mechanics is part of physics and engineering (and it surely is, as I can say as a bioengineering major ) is discussion meant for the OT subforum. Multiple posts haven been deleted.

 

I remember when thye thought we couldn't pass the 1u (1 micron) mark.

 

Just for sean473


See:
Riding a Rabi cycle past the limit of Moore's Law


Don't laugh too hard at moores law... where there's a will, theres a way.

 

Here are some interesting rumours about Ivy Bridge by Fudzilla:

Bye-bye, low-end graphic solutions by AMD or nVidia!

 

I doubt AMD or nVidia integrated graphics will end up weaker than Intel's in the near future...

 

NVidia's probably not even going to have any IGPs after Sandy Bridge releases unless they do something BIG to impress Apple.

Though, he's talking about entry-level dedicated cards, which will definitely be extinct by 2013.

 

The Ivy Bridge mobile platform is code named "Chief River".

 

Some comparison chart of the new Intel HD graphics only puts it to like a 256MB 9800PRO...so newest HD graphics is only equivalent to high end 7 year old graphics..

 

dayum, and I payed $500 for my 9800 pro when it came out too, helluva card. I'll probably be saying the same thing to my next purchase in 7 years time.

 

I believe the current trend is graphics power doubles every 18 months.

 

I couldn't afford it back then. I went from a RAGE 3D 16MB to a 9200 PRO and that was 189 dollars at CompUSA. The 9800 PRO was like 350 lol