Intel claims that its 10nm processors will offer an effective performance improvement of around 25% and an energy efficiency of around 40%, compared to the current 14nm.
A few days ago we had confirmation from Intel that they already had the 10nm architecture ready for the processors that will clearly arrive this year, such as Cannon Lake and Coffee Lake. This new manufacturing architecture will give a significant leap in performance and consumption, assuming in the words of Intel, an improvement in performance per cycle of 40% compared to the current 14nm. It is a very important leap in power, but which must then be transferred from paper to the real world.
Intel has held the 'Tech Day' event to inform and publicize the technical data regarding the new 10nm node. The company has ensured that the density of the transistors increases by 2.7 times compared to the current 14nm. This has been achieved thanks to the fact that the 'minimum gate pitch' has gone from 70nm to 54nm and that the 'minimum metal pitch' has been reduced from 52nm to 36nm. This implementation assumes that the transistor density is 100.8 × 106, per square millimeter, which is roughly double that of 10nm chips from Samsung, TSMC or Qualcomm.
The jump also offers other improvements, such as the general performance that will be around 25%, while energy consumption is reduced by 45%. The 10nm revision, which has been dubbed 10nm ++, will improve overall performance by 15%, while power consumption will be improved by 30%.
All these improvements are very interesting, but this is on paper, then it is possible that they approach these figures or not, it will depend on many factors. The highlight of this node is that it will bring us in Cannon Lake the support of Quad Channel DDR4 and 2-Way SLI / CrossFireX x16 / x16, according to the data, which will mean a very important leap in the performance of gaming equipment Although the Quad Channel can be an ordeal right now, since the price of RAM is absurdly high.
Source: Kitguru
But Coffee Lake didn't come in the 14nm manufacturing process?
Yes ... as far as I know, CoffeLake will be the last Intel pseudo-architecture to use the 14nm ++ node, on the other hand I have seen that most confuse the aforementioned 40% extra performance with the IPC, but I would like to clarify that it is the throughput per wafer, which is easy to achieve with Intel's claimed 2.7x density increase!
Of course, it is that in different media they said that CL came in 14nm and others in 10nm. Even so, let's hope how and when Coffe Lake comes out, because Intel should have already taken the step due to ryzen.
That's right ... we will have to wait, although in advance .... To be honest I don't expect an increase of more than 9% for Coffe Lake comparing you to you with SandyBridge 5-6th Generation (KabyLake)
Me neither, really the most interesting thing is that supposedly Intel would introduce the 6 cores in the consumer ranges, in the i7 we go.