Intel is reportedly gearing up to showcase its 22 nm next-generation processors based on the Ivy Bridge architecture soon, possibly during Computex Taipei 2011 taking place from May 30 to June 4. Actual processors aren't expected to hit the market until the first half of 2012, keeping in tune with Intel's yearly release schedule.
Ivy Bridge's appearance seems rather premature, given that Intel just launched the 32 nm Sandy Bridge platform during CES 2011 back in January. However, the architecture will remain similar tos Sandy Bridge, but reduced down to a smaller die thanks to 22 nm manufacturing. Thus, the Ivy Bridge chips will be smaller but are expected to provide an increased performance per watt. The processors will also be compatible with the LGA1155 interface.
Additional reports indicate that Ivy Bridge will bring a DirectX 11-compatible graphics core combined with the second-generation QuickSync video technology. Ivy Bridge will also reportedly have 24 execution units compared to 12 today, and include HDMI 1.4 output. Integrated subsystems are also expected to receive improvements with the addition of native support for 1600 MHz DDR3 and PCIe 3.0. Overall performance is expected to increase 20 to 30-percent over the equivalent Core processors released in 2011.
As for efficiency, I doubt AMDs first generation HK/MG and early 32nm process will be more efficient than Gulftown let alone Sandy Bridge. And Intels 22nm looks to be amazingly efficient.
Add in the fact that Ivy Bridge will drop into current LGA1155 mobos and its hard to say where AMD will be with BD since BD will only work on AM3+.
and for those who don't know, the longest molecule strand known to man is 22nm long!! i forget where i read it though. but it was on the internet, so you know it was the truth
4nm by 2022.
You fail. By the time you reach those sizes, the interconnect paths are going to be limited by electron traveling. Those are the complications of Quantum Computing. By that time, there is going to be a new medium built instead of silicon wafer.
now if i recall correctly Brazo (fusion) was originally planned for 32nm but due to fab issues had to use 40 odd nm, AMD instead of growing the die size decided to optimize the chip instead and was able to achieve comparable performance without growing the die size at all, that tells me that fusion (in the brazo guise) has significant headroom for when it does go 32nm
yes current gen AMD chips are behind the curve on performance but there is a significant amount of real estate that becomes available when they go to 32nm, just filling that out with transistor count will get you some performance boost, but with an optimized new architecture there is potential for some real good performance gains to be had, not saying they will deliver just saying logically it's plausible
kcorp20034nm by 2022.Not going to happen. After you pass a certain component size quantum tunneling get so bad that further process shrinks are nearly impossible. Basically quantum tunneling means that electrons pass through the insulating dielectric materials and large leakage currents result, much higher than what we currently see. If I recall correctly 16nm is the smallest process that can be achieved with conventional silicon technology. Further miniaturization and clock speed increases will require the use of new materials like graphene.
Exactly. AMD better hurry up, Intel's getting farther ahead of them.