Intel's Ivy Bridge Architecture Exposed
by Anand Lal Shimpi on September 17, 2011 2:00 AM EST- Posted in
- CPUs
- Intel
- Ivy Bridge
- IDF 2011
- Trade Shows
Final Words
Ivy Bridge will bring about higher clock speeds thanks to its 22nm process, however the gains will likely be minimal at best. Intel hasn't been too keen on pursuing clock speed for quite some time now. Clock for clock performance will go up by a small amount over Sandy Bridge (4 - 6%), combine that with slightly higher clock speeds and we may see CPU performance gains of around 10% at the same price point with Ivy Bridge. The bigger news will be around power consumption and graphics performance.
Ivy Bridge will be Intel's flagship 22nm CPU for some time. The chip was originally due out at the end of this year but was likely subject to delays due to issues with the fab process and the chip itself. The move to 22nm is significant leap. Not only are these new transistors aggressively small but the introduction of Intel's tri-gate technology is a major departure from previous designs. Should the fab engineers at Intel do their job well, Ivy Bridge could deliver much better power characteristics than Sandy Bridge. As we've already seen, introducing a 35W quad-core part could enable Apple (and other OEMs) to ship a quad-core IVB in a 13-inch system.
Ivy Bridge's GPU performance is particularly intriguing. With a 33% increase in execution hardware and a near doubling of performance per EU, it's clear that Intel is finally taking GPU performance seriously. If Intel can hit its clock and performance targets, Ivy Bridge could deliver GPU performance on-par with AMD's Llano. By the time Ivy Bridge arrives however, AMD will have already taken another step forward with Trinity. The question is who will address their performance issues quicker? Will AMD improve x86 performance faster than Intel can improve GPU performance? Does it even matter if both companies end up at the same point down the road? Short of 3D gaming workloads, I believe that x86 CPU performance is what sells CPUs today. Intel's embracing of OpenCL however and AMD's efforts in that space imply things are finally changing in that regard.
Sandy Bridge brought about a significant increase in CPU performance, but Ivy seems almost entirely dedicated to addressing Intel's aspirations in graphics. With two architectures in a row focused on improving GPU performance, I do wonder if we might see this trend continue with Haswell. Intel implied that upward scalability was a key goal of the Ivy Bridge GPU design, perhaps we will see that happen in 2013.
Ivy Bridge can do very well in notebooks. A more efficient chip built using lower power transistors should positively impact battery life and thermal output. Desktop users who already upgraded to Sandy Bridge may not feel the pressure to upgrade, but having better graphics shipping on all new systems can only be good for the industry.
Facebook
Twitter
RSS
97 Comments
View All Comments
piroroadkill - Saturday, September 17, 2011 - link
I'd have liked a little more on this.. What's the source?I searched anyway, and found it is using thermal sampling. Presumably it's also seeded. Anyway, I thought it was of interest.
Jamahl - Saturday, September 17, 2011 - link
Don't you get tired of saying "intel is finally taking gpu performance seriously" every year? I do.JonnyDough - Monday, September 19, 2011 - link
I'd just like to say...=) Yes sir, I do.
imaheadcase - Saturday, September 17, 2011 - link
I heard when sandy bridge came out they was considering a GPUless version for enthusiasts who don't need it..is that something they will do eventually?I suspect its tied to the core, so not going to happen because of high costs. But wouldn't that save even more power/heat problems with that removed?
It just seems like its a mobile orientated cpu vs consumer. :D
DanNeely - Saturday, September 17, 2011 - link
With power gating if you're not using the IGP it doesn't consume any power; so the only thing they'd save on is die area by removing it.imaheadcase - Saturday, September 17, 2011 - link
Ah did not know that part. thanks.fic2 - Monday, September 19, 2011 - link
"I heard when sandy bridge came out they was considering a GPUless version for enthusiasts who don't need it..."Interesting since Intel did the exact opposite - put the only GPU with half decent performance into the enthusiast 'K' series.
JonnyDough - Monday, September 19, 2011 - link
The only people who would actually consider that are businesses and home users who don't play "real" games. :Pjjj - Saturday, September 17, 2011 - link
"I believe that x86 CPU performance is what sells CPUs today"That's not all that true anymore,there was a time when apps used by everybody required a fast CPU but that's not the case anymore nowdays..Just a few years ago playing HD content was a chalange on older systems but now ,if you look at usage paterns and what kind of perf is needed, the picture has changed. This is one of the reasons PC sales are not doing so great,there is no need to upgrade your system every 1-2 years.Even Windows is not driving system requirements up anymore.
In the consumer space GPU and battery life matter more now. Intel is trying to fight all this with lower power consumption, ultrabooks but that far from enough.If they want to survive the ARM "tsunami" (think about the financial part too here not just perf) , they got to push the software to be more demanding too and maybe the easiest is to do it on the GPU side -not in games.
MadMan007 - Saturday, September 17, 2011 - link
Intel's quarterly results say there is less to worry about than hyperbolic ARM domination headlines would lead one to think. One IDF slide showed large growth in emerging markets where the analysts aren't as able to get reliable data. Yes, PC upgrade cycles are longer, but that doesn't mean there is not net worldwide growth.There is room for growth in both areas, it's not a zero-sum game, and some things like mobile video consumption actually go hand-in-hand with faster beefy CPUs.