Latest Intel roadmap confirms PCI Express 3.0 for Ivy Bridge
Did you expect anything less?
Mar 28, 2011 in Chips, Desktop, Microprocessors, Mobile
Up until now Intel (INTC) hasn’t unveiled which PCI Express spec its upcoming LGA-1155 Ivy Bridge processors will support, but their latest roadmaps includes an update which reveals support for PCI Express 3.0, even though it’s a feature that’s not even highlighted on the slide. Keep in mind that this is a generic Ivy Bridge overview and not specific to a certain product SKU, but it looks like all CPUs top to bottom will get PCI Express 3.0 support.
Beyond PCI Express 3.0 the big news of Ivy Bridge is, of course, that it will be Intel’s first 22nm production CPU, other than the memory news, that is. We’ll see yet another new graphics core which will apparently bring support for DirectX 11 and OpenCL 1.1 as well as improved video encode, decode and transcode support. We can also expect support for faster memory, although as always, Intel is being very conservative and only moves on to DDR3 1600MHz which seems to be a small step forward considering the platform is still almost a year away.
The slide also verifies that the Panther Point chipset will be called the Intel 7 series, although exact chipset models may or may not follow the current numbering scheme, especially when you take the Z68 chipset into consideration, as this is a new addition to Intel’s otherwise somewhat predictable chipset numbering scheme. Chipsets with display connectivity will gain support for a third display, as the current 6 series chipsets are “limited” to two displays. SATA support appears to stay the same with a mere two SATA 6Gbps which is disappointing considering that AMD has had full SATA 6Gbps support for quite some time now.
The 7 series is of course Intel’s first chipset to gain native USB 3.0 support as well and as far as we know, nothing has changed here and we’ll see support for four USB 3.0 ports, although this might very well differ between the various chipset SKUs. Sadly it looks like Intel decided to stay with DMI 2.0 and as such there won’t be any additional bandwidth available for the PCH, something that might become an issue with an integrated USB 3.0 host controller, SATA 6Gbps and additional peripheral controllers.
Intel is expecting a 10-20 percent performance improvement over Sandy Bridge with Ivy Bridge, but it’s not clear if this is clock for clock as the slide doesn’t go into enough details to clarify what Intel based the performance gain figure on. We’ll just have to wait and see what Intel delivers, but that won’t happen until early 2012, at least not unless AMD can offer Intel some serious competition later this year that would cause Intel to change its current plans.S|A
32 Responses to “Latest Intel roadmap confirms PCI Express 3.0 for Ivy Bridge”
Trackbacks/Pingbacks
-
-
Apr 5, 2011
[...] 3.0 x16 interconnection as well as PCIe 2.0 x4 controller, according to a slide published by SemiAccurate [...]
-
-
Mar 30, 2011
[...] [it.com.cn] [semiaccurate] [...]
-
-
Mar 29, 2011
[...] [...]
-
-
Mar 29, 2011
[...] certain product SKU, but it looks like all CPUs top to bottom will get PCI Express 3.0 support. http://semiaccurate.com/2011/03/28/l…or-ivy-bridge/ I hope PCI-E 1.0 support dies, but I doubt it. [...]
-
-
Mar 29, 2011
[...] PCI Express 3.0 for Ivy Bridge http://semiaccurate.com/2011/03/28/l…or-ivy-bridge/ [...]
-
-
Mar 29, 2011
[...] semiaccurate.com [...]
-
-
Mar 29, 2011
[...] semiaccurate.com [...]
@3 JEEBEE And from which rotten part of this world do u come ;)
I have a dream that one day, Intel and nVidia will have a marriage just like AMD-ATI.
I have a dream that one day, there will be an equal amount of focus on making newer cards run cooler, more energy efficient and quieter… instead of just fps and eye candy when graphics settings are set to max.
I have a dream that one day, there will be no difference between CPU and GPU.
Charlie,
Any intelligence on if will be 32nm Sandy Bridge dual-socket Xeons or whether
Intel will go straight to 22nm/Ivy Bridge, staying with the Nehalem X56xx until then?
Cheers,
M
Most of motherboards that designed for Sandy Bridge can’t really support Ivy Bridge.
No hopes for same socket & full drop-in compatibility here please! :-)
Any linkies? What I’ve heard so far is that 6-series should support IvyBridge with a BIOS update…
the only advantage intel will have this time is the 22nm vs 32nm, power usage will be with intel from the smaller size etc, but the performance gain will be bigger from the AMD camp.
AMD has the advantage of speed/Architecture in bulldozer, same as with their gpu manufacturing style(once they are ready). Intels tick tock is finished after the 22nm, things will get better for AMD in the core race, where AMD is king.
AMD has a bright future ahead (Fusion FTW)
The 10% – 20% claim is the same claim they make every “tick”, and they make some absurd 30% – 100% or whatever every “tock”. This has less to do with any real world improvements, and more to do with trying to fool people into thinking that they’re actually improving with each generation.
Sandy Bridge failed to improve 10% – 20% over Nehalem in many real world benchmarks, why would a “tick” improve that much? The fact of the matter is, that x86 is quite mature, and there’s not much more they can do to improve general IPC past what the Core2/Phenom generations did. If they could’ve continued increasing IPC with additional transistors, they wouldn’t have bothered with multicore. They can’t increase clockspeed either, even though Intel marketing has done a fine job of convincing people that SB can easily OC to 5ghz. Even if some can OC past 4ghz, they absolutely will not last long at that frequency.
I’ve been running my i5-2500K at 4.3 GHz since day one. It’s rock solid for 24/7 folding. I’ll be sure to come back and update this if it ever shows signs of “not lasting long”.
pci-e3 is new revolution, paradigmn shift, entire new generation of game card….
vondrashek
Nah, just another way to shift digmns.
Article is just PR none sense. Why do people here refer that Z68 would have anything to do with ‘Intel 7 Series’ it’s a crappy chip with enabled GPU and it’s ‘Intel 6 Series’.
Also, this was damn hilarious:
“graphics core which will apparently bring support for DirectX 11 and OpenCL 1.1 as well as improved video encode, decode and transcode support.”
Now we barely have support for AMD OpenCL and NVIDIA CUDA who gives a shit about Intel. “Improved video encode/decode”, hah, get real the I’d like to see day Intel is willing to pay to get their features working. I bet it’s gonna be a killer party for OCL/OGL coders.
What are you rambling about? QuickSync is already the fastest thing on the planet.
Goddamn trolls with zero brain logic.
Dude, you’re a moron and a perfect mark for Intel. Quick Sync is fast but gives **** quality and is hence unusable for anyone who cares about actually watching the encoded material. Obviously, you have no critical thinking or research skills and only regurgitate th official Intel PR-material.
If 1 second of googling proves you 100% wrong maybe you should keep your opinions to yourself.
http://www.anandtech.com/show/4083/the-sandy-bridge-review-intel-core-i7-2600k-i5-2500k-core-i3-2100-tested/9
quicksync works only if graphics enabled…
SO, YOURE OUT OF LUCK ON P67 BOARDS
tell me, how much you got paid from intel for posts like yours ?
I think genetix was pointing out that the number of software titles that exploit OpenCL or CUDA is rather small.
Great, with Ivy Bridge, AMD will be ended in 2012. Everyone are going to be happy with Intel Inside Everything. Even many former AMD employee at Intel want AMD to be dead because that company is so poor and only churning out craps in the market by using crappy and outdated manufacturing.
All people will start to kill everything that is green. Children will start to call themselves Otellini, and kill all their classmates, if they ask for money or wears a green thirt. ATI fanboys will commit suicide on the streets. Everyone will celebrate. They have a new leader. The great, great Maddoctor. Successor to Gadaffi.
Well, considering AMD is, atm, moment offering 10 times more than Intel and at least they have some kind of innovation behind next generation chip where Intel is restricting the shit out of everything. I’d say this slide proves very well that Intel is no longer the way to go.
I personally didn’t believe in AMD before SB made to market and seems that intel is continuing the same crap they began with to the next year. Something I think nobody should take a part to.
Offering 10 times more of what? Innovation?
You’re a mess. And you’ll be a sad mess when Bulldozer prices come out.
Why would you be so happy about AMD’s hypothetical death ? What’s worth rejoicing in paying 300$ for sub-potential CPUs and the lack of innovation on a market dominated by a monopoly ?
AMD is late on the manufacturing process but it’s doing great on the GPU side. The raw CPU power doesn’t matter as much as it did a few years back and I’d say its future looks very bright with integrated graphics. What are the benefits of 22nm for Intel ? Maybe mainstream 6 cores instead of 4, a big deal for what, maybe 2% of the buyers. IPC should stand about the same. IGP will be incrementally faster, still crippled by lack of drivers support and AMD’s IGP will remain a lot better.
So tell us what’s so exciting about Intel and Ivry Bridge atm ?
maddoctor cant speak, he can only troll and trash if article has AMD in the name..
he is a sad, poor mess, not a real living person
How strong would a bridge made from ivy actually be? Ivy isn’t the strongest of plants, although it’s good at making good infrastructure crumble away.
In parts of India they train creeping figs to form bridges over hundreds of years. I think ‘creeping fig bridge’ is a far better name than Ivy Bridge.
No, they will most definitely stay with socket 1155 for the mainstream. Earlier on it was reported there will be a BIOS update for Intel 6-series chipsets to support 22nm processors. Remember the Core2.
I wonder why such claims arise of a 10-20% perf. increase. As we know from the Core 2 family, they also promised all kinds of performance increases. In the end, the 65nm and 45nm performed the same given the same amount of L2 cache and clockspeeds! The only difference was that the 45nm generation would consume slightly less power (same clocks) and generally could be OC’ed higher. Because of closer packing it also required a bit less voltage.
People, the performance will not increase for the same clocks/cores if the fundamental architecture is unchanged! These are bogus claims, don’t believe silly marketing talk! I get sick or reading such nonsense only to attract more attention to themselves!
Another new socket, then?