Intel Completes Development of 1.8nm and 2nm Production Nodes (tomshardware.com) 25
Intel has completed development of its Intel 18A (1.8nm-class) and Intel 20A (2nm-class) fabrication processes that will be used to make the company's own products, as well as chips for clients of its Intel Foundry Services (IFS) division, reports UDN. From a report: Wang Rui, president and chairman of Intel China, said at an event that the company had finalized the development of its Intel 18A (18 angstroms-class) and Intel 20A (20 angstroms-class) fabrication processes. This does not mean that the production nodes are ready to be used for commercial manufacturing, but rather that Intel has determined all specifications, materials, requirements, and performance targets for both technologies.
Intel's 20A fabrication technology will rely on gate-all-around RibbonFET transistors and will use backside power delivery. Shrinking metal pitches, introducing all-new transistor structure and adding backside power delivery at the same time is a risky move, but it is expected that 20A will allow Intel to leapfrog the company's competitors -- TSMC and Samsung Foundry. Intel plans to start using this node in the first half of 2024. Intel's 18A manufacturing process will further refine the company's RibbonFET and PowerVia technologies, as well as shrink transistor sizes. Development of this node is apparently going so well that Intel pulled in its introduction from 2025 to the second half of 2024. Intel originally planned to use ASML's Twinscan EXE scanners with a 0.55 numerical aperture (NA) optics for its 1.8 angstroms node, but because it decided to start using the technology sooner, it will have to rely on extensive use of existing Twinscan NXE scanners with 0.33 NA optics, as well as EUV double patterning.
Intel's 20A fabrication technology will rely on gate-all-around RibbonFET transistors and will use backside power delivery. Shrinking metal pitches, introducing all-new transistor structure and adding backside power delivery at the same time is a risky move, but it is expected that 20A will allow Intel to leapfrog the company's competitors -- TSMC and Samsung Foundry. Intel plans to start using this node in the first half of 2024. Intel's 18A manufacturing process will further refine the company's RibbonFET and PowerVia technologies, as well as shrink transistor sizes. Development of this node is apparently going so well that Intel pulled in its introduction from 2025 to the second half of 2024. Intel originally planned to use ASML's Twinscan EXE scanners with a 0.55 numerical aperture (NA) optics for its 1.8 angstroms node, but because it decided to start using the technology sooner, it will have to rely on extensive use of existing Twinscan NXE scanners with 0.33 NA optics, as well as EUV double patterning.
Are these fake? (Score:4, Interesting)
What are the performance charcateristics of the node? Specifically: 1. Transistors per square mm 2. power requirement per transistor 3. transistor switching speed (gate delay)
They all have been "fake" for a long time (Score:5, Informative)
All the "production nodes" have been fake for a long time. That's why they are called "production nodes" now and not gate sizes. The "size" given has no relation to actual gate length or pitch and hasn't had for almost a decade now. They are all just marketing terms. Actual gate sizes are much larger, and the whole thing is basically a big huge marketing snow job.
You'd think they'd just abandon the gate size as their metric and go with something else. But you see, it started with just minor fudging. Rounding down to the nearest nm here, calling a 11.5nm process that was as fast as a competitor's 10nm process 10nm, etc. Once they started down the garden path, the lies just got bigger and bigger and no one could stop. Pretty soon our 8nm gates will be labelled the "5picoM process" because they were able to eek out an extra couple hundred flops-per-watt.
Re: (Score:3)
Exactly right. Which is why tech journalists worth any salt should ask for the questions that matter which I asked in my comment. Plus maybe cost per transistor. Those four are the key metric.
Re: They all have been "fake" for a long time (Score:2)
Sounds like megahertz myth 2.0. Though you have to hand it to Intel that they finally understand the value of smaller numbers.
Re: (Score:2)
Boy, I can't wait to see how fast these chips are on a 10g phone network, which is five magnitudes and generations faster than a 5g!
Finished but not finished (Score:4, Informative)
Intel needs 20a production-ready by the end of 2024. They still haven't shipped anything on Intel 4 yet. For anyone not paying attention, Intel's road map calls for Intel 4 late this year as Meteor Lake, then Intel 3, and finally Intel 20a. Anyone wanna take bets on them iterating through three different nodes in the next year-and-a-half?
Re: (Score:3)
Despite the increased competition, now is not a good time to buy a laptop it seems. Intel is stuck on their crappy process for at least another couple of years, and the current parts run very hot.
AMD's chips are a lot better but their mobile offerings thus far seem to lack enough PCIe lanes and a proper USB4/Thunderbolt controller on-board.
Both Intel's 13000 series and AMD's 7000 series are disappointing for mobile users.
Re: (Score:2)
This matter extends far beyond laptops. If/when Intel fails to deliver on their node time line, they can either become a permanent vassal to TSMC or contract significantly.
Re: (Score:2)
This matter extends far beyond laptops. If/when Intel fails to deliver on their node time line, they can either become a permanent vassal to TSMC or contract significantly.
Hard, since Apple has gobbled-up all of TSMC's newest-node capacity.
https://appleinsider.com/artic... [appleinsider.com]
Re: (Score:2)
Not true. Intel is already paying TSMC for N3. TSMC built out extra capacity just for Intel.
Re: (Score:2)
AMD's chips are a lot better but their mobile offerings thus far seem to lack enough PCIe lanes and a proper USB4/Thunderbolt controller on-board.
Then they can hardly be called "a lot better" now, can they?
Full disclosure: I could care less about lame-ass x86/64 architecture. It is all too gluttonous, power-wise.
Re: (Score:2)
AMD parts are actually more power efficient than ARM chips for the same amount of work done, when comparing performance cores. What lets them down is Windows and Linux.
Re: (Score:2)
AMD parts are actually more power efficient than ARM chips for the same amount of work done, when comparing performance cores. What lets them down is Windows and Linux.
Citation, please.
First, define "work" in this context. Then define "done" in this context.
Re: (Score:2)
Okay Dr. Peterson, I think we all know what a CPU doing work means.
Re: (Score:2)
Okay Dr. Peterson, I think we all know what a CPU doing work means.
Do we?
Is the "work":
Integer Addition?
Floating Point Addition?
Integer Multiplication?
Floating Point Multiplication?
Integer Division?
Floating Point Division?
Integer Compare?
Floating Point Compare?
Push Register?
Pull Register?
Subroutine Call?
Subroutine Return?
Interrupt Service?
Interrupt Return?
etc?
Nevermind if we're talking single or multicore here! That adds a whole other world of variables
Each CPU architecture does each of those things with greater or lesser "efficiency", in terms of power and time.
So again:
Re: (Score:2)
Intel owning the M2/M3 "chips" that Apple (allegedly) invented.
What's with the scare quotes?
If you haven't noticed, Apple has been home-growing their own SoCs for well over a decade.
Even when Samsung was Apple's Fab, they were still Apple Designs.
Completes Development of 1.8nm and 2nm Nodes (Score:2)
There's a hilarious story about just this. I can't link directly to the story, but it was: "It was nothing, really!" by Theodore Sturgeon. An overview of it is here [stackexchange.com] or an alternate telling here [stackexchange.com].
odd post (Score:4, Interesting)
Re: (Score:2)
Good point.
I was expecting Intel and all the big guys not to do cutting edge research within China.
Maybe they are hoping that even if the R&D secrets leak, China will not be in a position to do anything cos they don't have access to the latest ASML (and Cannon and other company's in the field) gear?
Re: (Score:2)
Never go ASML
Re: (Score:2)
Re: (Score:1)
Uhhh (Score:2)
I thought they were supposed to build these in the US, and forbidden to export the tech to China?
But they built in China. I have a confused. Thanks Intel :(
So Intel had to go to China? (Score:2)
To finally make it? Sounds like it...