Print 48 comment(s) - last by blandge.. on Apr 26 at 12:22 AM

  (Source: Clubic)
IGP is faster than Sandy Bridge, but well behind AMD; power savings to be key amid higher prices

Today Intel Corp. (INTC) is reportedly launching (according to reports by Electronista and BBC News) its third-generation of Core i-Series processors, code-named Ivy Bridge.  The launch comes just over a year and four months after the introduction of Sandy Bridge, which was launched at the 2011 Consumer Electronics Show.

I. Third-Generation Core i-Series Chips Announced a Bit Earlier Than Expected

Ivy Bridge brings a new integrated graphics processor (IGP), a 22 nm die-shrink, new power-saving 3D FinFET transistors, and long-overdue native USB 3.0 support.

The arrival of Ivy Bridge is a bit of a pleasant surprise.  While it was reportedly complete and ready in January, low demand from OEMs reportedly pushed the launch date backwards.  Rumor had it that the launch would be pushed all the way to June, but here we are with 13 Ivy Bridge chips reportedly ready to ship.

There's been no official press releases from Intel, thus far, and prices/models have not gone live on, Inc. (AMZN) or  It is not immediately clear when these parts will be available.

Ivy Bridge
A colorized die-shot of Intel's third generation Core i-Series CPU. [Image Source: BBC News]

Intel's PC business chief, Kirk Skaugen, told the BBC, "The momentum around the system design is pretty astonishing.  There are more than 300 mobile products in development and more than 270 different desktops, many of which are all-in-one designs.  This is the world's first 22 nanometre product and we'll be delivering about 20% more processor performance using 20% less average power."

Intel reportedly has three factories fully upgraded to the 22 nm process and churning out Ivy Bridge chips, with a fourth coming online later this year.  States Mr. Skaugen,  "This is Intel's fastest ramp ever.  There will be 50% more supply than we had early in the product cycle of our last generation, Sandy Bridge, a year ago. And we're still constrained based on the amount of demand we're seeing in the marketplace."

The world's largest chipmaker's marketing of the chip -- aimed at its much touted "ultrabooks" -- revolves around the power-saving benefits of the die shrink and new 3D transistor gate design.  Lauds Mr. Skaugen, "A lot of people had thought that Moore's law was coming to an end.  What Intel has been able to do is instead of just shrinking the transistor in two dimensions, we have been able to create a three-dimensional transistor for the first time.  For the user, that means the benefits of better performance and energy use will continue for as far as Intel sees on the road map."

II. More Questions Than Answers

But there are many unanswered questions about Ivy Bridge.  First, there's the question of its supposed strength -- power consumption.  Competitor Advanced Micro Device, Inc. (AMD) just unleashed a power-sipping pair of system-on-a-chip designs dubbed Trinity and Brazos 2.0.  While the AMD chips are built on a more power hungry 32 nm process, they are also expected to feature less-powerful CPU cores.  Who comes out on top power-wise is still up in the air -- Intel clearly has the process lead, but AMD's slimmer cores may close that gap, while delivering superior pricing.

And that's not to mention the waiting ranks of ARM chipmakers like Qualcomm, Inc. (QCOM) who are preparing their chips, such as the Snapdragon 4 series, for Windows 8 laptops.  Like AMD, they are built on a bigger feature size, but include more lightweight cores.  Additionally, ARM chips enjoy certain architectural advantages from a power perspective versus Intel and AMD's x86 chips.

Who comes out on top of the power battle will likely not be decided until October or November, when the ARM designs launch in mass, and Intel/AMD's respective lineups have had time to flesh out and be thoroughly benchmarked.

The other major question is graphics.  While the HD 3000 IGP inside Sandy Bridge was a step forward for Intel, it was also nearly twice as slow as the IGP found in AMD's Llano chips.  Intel is touting Ivy Bridge's HD 4000 IGP as a major advance performance-wise.  In some talks it's bragged of 2 or more times increase in performance -- but other reports indicate a more conservative 60 percent (on average) performance bump.

Early reviews [example] are not looking good for Intel on the IGP front, as they show the chip unable to beat AMD's last generation design, and posting 50 percent (less than promised) gains in real world games and 80 percent gains in synthetic benchmarks.  While there's a need for more careful analysis (Anandtech's thorough review is on the way -- teaser here.), it appears that Ivy Bridge may be DirectX 11 compatible, but will need to be paired with a discrete GPU (unlike AMD's design) to have acceptable gaming.

With AMD's Trinity boasting a reportedly much-improved hybrid design incorporating elements of AMD's HD 6000 and HD 7000 series of discrete graphics chips, Intel's hopes of winning the IGP war this round are all but over.

III. Can Intel Sell Customers on Twice-as-Expensive Ultrabook Designs?

Likewise, it's anyone's guess how the IGPs inside high-end laptop-ready ARM chips will stack up to Intel's offerings in Windows 8 benchmarks.

If there's one safe bet, it will be that the new Intel chips will be more-powerful CPU-wise on a per-core basis than its competitors.  The compelling question, though is whether that will matter enough to consumers to overlook the chips' relatively high price point.

The initial launch is expected to feature 13 pricier quad-core models, according to BBC News.  These chips will fall under the Core i5 and Core i7 monikers.  The lineup will only be fleshed out with single-core and dual-core (Core i3, i5) models later this spring.

Intel Ultrabooks are expected to be $800 to $1,000 USD.  By contrast, AMD is promising sub-$500 USD ultrathins.  ARM notebooks are expected to debut at a similar price -- or even lower.  The price difference reportedly stems largely from the higher unit prices for Intel's chip + chipset solution and the potentially weaker IGP, which necessitates a low-end discrete GPU, in many cases.  In other words, don't necessarily expect a $900 USD Intel Ultrabook to be any higher build quality than a $500 AMD ultrathin.

If AMD can beat Intel in graphics-enabled applications such as Adobe Systems Inc.'s (ADBE) Photoshop CS6 and various games, the question is what if any good Intel's much more powerful CPU cores are for laptop users.  Likely only browsers (think intense multi-tab sessions) and a handful of other apps might perform better on Intel's more powerful CPUs.  But the question is whether winning in "some cases" is enough to convince customers to pay nearly twice as much.

Trinity in the wild 

Trinity in the wild
Intel is facing fierce competition from AMD, whose Trinity (center). is expected to outperform Ivy Bridge graphically, while being featured in laptops that are nearly half the cost of Ivy Bridge ultrabooks. [Image Source: Jason Mick/DailyTech LLC]

For that reason it's little wonder that Intel is hyping power consumption so much.  Power is a must-win scenario for Intel in the laptop space.  Utility in mobile devices is still heavily dictated by battery life.  If Intel can beat AMD from a power perspective, it will make a compelling case for its pricier chips.

Intel is already looking ahead to its architectural makeover in the 2013 Haswell, promising even more power advances.  States Mr. Skaugen, "We are targeting 20 times better battery life on standby - always on, always connected.  So you can get all your files and emails downloaded onto your PC while it's in your bag, and still get more than 10 days of standby and all-day battery life."

The veteran will need to deliver big on those promises in the face of intense competition, or it could face extinction in the mobile space it has long dominated.

Sources: BBC News, Electronista

Comments     Threshold

This article is over a month old, voting and posting comments is disabled

RE: Sensational Mick
By Goty on 4/23/2012 8:13:21 PM , Rating: 2
Taken in isolation, IVB is great, but the performance gains just aren't there for anyone who owns pretty much any processor from the previous two generations (heck, I'm still rocking my 920 D0 @ 4.2 GHz and don't feel the slightest bit of desire to upgrade). Take this with the less-than-promised (well, hyped, anyhow) power consumption gains and disappointing overclocking and IVB is definitely a let-down.

RE: Sensational Mick
By Reclaimer77 on 4/23/2012 10:17:55 PM , Rating: 2
These aren't the low power models though. These are desktop chips. How is a desktop i7 pulling just 45w at max load at "Turbo" clocks not impressive as hell?

Overclocking? To make this chip a better overclocked you would be giving up WAY more than you get. Explained here:

And honestly who really needs to overclock anymore? The off-the-shelf performance of these chips are just outrageous. Even you just said two generations ago are "good" enough, so why do you also put emphasis on overclocking as a negative? That's a contradiction.

I guess if you're putting unrealistic expectations on it, then sure, IVB is a "let down". But in reality, given that this is the first iteration of 22nm Ivy Bridge, you cannot be serious. Let down compared to what? Nothing out there is even close.

RE: Sensational Mick
By Goty on 4/23/2012 11:40:11 PM , Rating: 2
"Unrealistic" is a relative term. Everyone and their brother were hyping finfets/3d transistors as being the next coming/the best thing since sliced bread, when at best it is an incremental improvement that doesn't even really match up to the move to SOI (for AMD) or HKMG (for both Intel and AMD).

RE: Sensational Mick
By Reclaimer77 on 4/24/12, Rating: -1
RE: Sensational Mick
By Goty on 4/24/2012 1:05:32 AM , Rating: 3
It is. It's pushed back Moores Law. If you can't understand that, I don't know what to say.

I don't know what to say. (See what I did there?)

And SOI? Where are you getting this stuff from? SOI is a whole lot more expensive to produce than building a finFET based processor currently. SOI has always been more expensive thus why Intel has never used SOI in their processors. Why don't you actually read about the problems that AMD has had with SOI and compare Bulldozer to Ivy Bridge. Notice something?

The two aren't even at odds anyway. SOI is a materials change while finFet is a structural improvement. In theory you could have finFET processors on SOI, which would pretty much rock.

I'm not sure what exactly you took offense to here (or why you're getting to riled up, outside of a little too much love for a name). SOI, HKMG and tri-gate transistors are all targeted at one thing: improving transistor performance, whatever the metric you choose to measure "performance". Also, I'm simply comparing the impact of the move to these different techniques; I don't care about ongoing difficulties or diminishing returns. Also, I never implied that SOI/HKMG/tri-gates were mutually exclusive technologies, that's just another case of a fanboy making up arguments that don't exist.

That's by design of course. Hello? This is the first Ivy Bridge chip.
You think this is the best they'll ever get? Or that Intel doesn't want to hold back? Of course they're holding back to maximize profits, they have ZERO competition from AMD. Why wouldn't they? It's like you're tunneling on the end result of this first chip and ignoring the potential of the technology entirely.

Oh, absolutely, Intel could have made a monster chip that's four times as fast as this IVB on a completely new process with a new transistor technology if they just worked at it!


That being said, I'm judging the technology as it is currently implemented, and the results are (again) an incremental improvement. Based on the hyperbole that was spread for most of the past year about the technology, I was expecting more. If things change down the road (as they most certainly will), so will my opinion.

You people can be Tri-Gate deniers if that makes you happy, whatever. It's willful ignorance in my opinion. It's like arguing that fuel injection isn't all that great, and carburetors are almost as good lol.

I'm not sure what a "Tri-Gate denier" is, but ok. I'll continue to be happy in my opinion (which seems to be somewhat more informed than your own), and you can continue to make up arguments and defend Intel against all opposition (real or imagined) to your last breath as long as that makes you happy.

RE: Sensational Mick
By Reclaimer77 on 4/24/2012 2:01:02 AM , Rating: 2
There's really no need for that much sarcasm and condescension. But there seems to be a concerted effort on some peoples part to insist that Tri-Gate is a disappointment and is no big deal. If you are not one of those people, I apologize.

SOI, HKMG and tri-gate transistors are all targeted at one thing: improving transistor performance, whatever the metric you choose to measure "performance". Also, I'm simply comparing the impact of the move to these different techniques

Well seeing as how tri-gate is the only technology that's actually incorporated into competitive CPU's for the desktop/server, I feel you only brought up these other processes to, again, minimize Intel's accomplishment here. You said it didn't "match up to the move to SOI" without even bothering to support your argument. That is far from a definitive statement.

That being said, I'm judging the technology as it is currently implemented, and the results are (again) an incremental improvement.

I'm just trying to understand your basis for comparison. They were able to increase CPU AND GPU power and still make a significant improvement in efficiency. Not sure what's so disappointing about that.

It's the most efficient clock for clock and most powerful quad core CPU you can currently buy. Hands down. You can dance around it all you want, but there it is.

RE: Sensational Mick
By bigboxes on 4/24/2012 8:18:27 AM , Rating: 2
Goty, you are spot on today. As another Intel i920 D0 owner I completely agree with what you are saying. I too have decided to wait to Haswell to update my desktop. I'm thinking that Trinty will be fine on the laptop and ARM may rule the tablet (for all practical purposes). I'm not all caught up in the spec wars. I just want what's best for me, the consumer.

RE: Sensational Mick
By BSMonitor on 4/24/12, Rating: -1
RE: Sensational Mick
By Reclaimer77 on 4/24/2012 5:24:56 PM , Rating: 1
Yeah BS I really don't get this guy. I just love how people work so hard to make the facts somehow fit their world view on something, but in this case Goty is really stretching reality.

It's the first released chip using a brand new process, OMG it doesn't instantly blow away a mature platform like Sandy Bridge? God it sucks!

RE: Sensational Mick
By someguy123 on 4/24/2012 12:46:34 AM , Rating: 2
FinFet exists to reduce leak and provide better scaling at lower voltages. I'm not sure why anyone would think that finfet would provide a huge bump in desktop performance, where the standard voltage fluctuates at around 1.2, which is pretty much in line with performance scaling graph intel released a while back showing gate delay improvements. As anandtech's testing notes, the chips they received could be undervolted down to about .9v while still overclocking to 3.9ghz and running stably. This sounds like a case where people were hyped due to misinformation.

RE: Sensational Mick
By Amiga500 on 4/24/2012 2:44:16 AM , Rating: 2
This sounds like a case where people were hyped due to misinformation.

No, people were hyped because they were stupid and unable to comprehend the information put in front of them.

RE: Sensational Mick
By TakinYourPoints on 4/24/2012 4:55:34 PM , Rating: 2
Exactly. People expecting decent overclocks from a CPU made off a new process versus a mature one were not paying attention or thinking critically.

RE: Sensational Mick
By blandge on 4/26/2012 12:22:34 AM , Rating: 2
Ivy Bridge was about where most people with technical knowledge expected it to be. All of the data Intel release showed 22nm finFET transistors having marginal performance gains at desktop voltages.

Consider that the FX 8150 has a 125W TDP. Compare that to IVB with 77W. That's a 48W difference in TDP with SIGNIFICANT performance advantage to IVB. That's almost unbelievable.

Still, the true advantage of 22nm finFETs is yet to be revealed in full. The performance per watt delta between 22nm finFET and 32nm planar is not linear. At lower power finFETs become more efficient and more powerful compared to planar. Wait until 22nm ULV (17W) and Atom (<1W - 10W) are released before you start making negative claims about finFET performance.

RE: Sensational Mick
By BSMonitor on 4/24/2012 9:26:59 AM , Rating: 1
LMAO, it's ONLY overhyped in YOUR mind.

RE: Sensational Mick
By tamalero on 4/24/2012 12:06:45 PM , Rating: 2
except anandtech are well known to swing for Intel every single time.. and again proven when they were saving intel, during the whole failing laptop graphics saga.

RE: Sensational Mick
By TakinYourPoints on 4/24/2012 4:53:28 PM , Rating: 2
Ivy Bridge is on the latest fab, there isn't much question about how well it wouldn't OC compared to Sandy and its very mature 32nm process.

I'm in the same boat though, i7 860 here and I have no pressing reason to upgrade. An SSD and new video card are all you really need to keep it current, and I'm covered on both with a 240GB Intel SSD and a GTX 680. Haswell may make me finally upgrade, we'll see. Until then I'm actually ok, I haven't been able to milk a single CPU/mobo like this for over a decade.

"Paying an extra $500 for a computer in this environment -- same piece of hardware -- paying $500 more to get a logo on it? I think that's a more challenging proposition for the average person than it used to be." -- Steve Ballmer

Most Popular ArticlesAre you ready for this ? HyperDrive Aircraft
September 24, 2016, 9:29 AM
Leaked – Samsung S8 is a Dream and a Dream 2
September 25, 2016, 8:00 AM
Inspiron Laptops & 2-in-1 PCs
September 25, 2016, 9:00 AM
Snapchat’s New Sunglasses are a Spectacle – No Pun Intended
September 24, 2016, 9:02 AM
Walmart may get "Robot Shopping Carts?"
September 17, 2016, 6:01 AM

Copyright 2016 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki