is 5nm the end for silicon?

3nm is

>>61903568
Yes

https://www.theinquirer.net/inquirer/news/3011257/ibm-breakthrough-paves-the-way-for-five-or-even-three-nanometer-chips

>>61903643
OY VEY DELET

File: intel did 911.jpg (312KB, 1434x834px) Image search: [Google]

312KB, 1434x834px

>>61903643
>ibm

DELETE THIS

>>61903568
I hope it will be also the end of gaymers threads here about CPU and GPU.

ITT: tards that don't realize node names have become entirely arbitrary/commercial and not based on transistor size.
The gate pitch of 14nm transistors is like 38nm.
Silicon will never arrive to "real" 5nm

>>61903568
10nm architecture is already starting to have some serious problems with electromigration; I literally wouldn't expect a 5nm chip to survive an entire month of normal operation. also the 14nm -> 10nm transition was where the cost per transistor stopped going down and started going back up due to ever worsening yields, so your 5nm cpu will be way more expensive in addition to being harshly lifespan-limited.

>>61903643
So POWER only?

Any feasible semiconductor alternatives? CNTs?

>>61903568
IBM has patents for 3.5nm supposedly. Not sure how that's gonna work but we'll see.

>>61906157
No, IBM is co-developing processes with Samsung; who is sharing fabs with GloFo, who makes chips for AMD.

So 5mm Zen 4 is a thing that could happen.

File: R14kkDj.jpg (35KB, 657x527px) Image search: [Google]

35KB, 657x527px

>you will never be a 32 nm silicon chipmin

File: 1400768188465.gif (973KB, 271x200px) Image search: [Google]

973KB, 271x200px

>>61906362
I am considerably more erect now.

>>61906362
>inb4 epyc cpus with SMT4.

>>61906306
N A N O T U B E S

Honestly though I think we need to stop obsessing with "muh shrink", and start innovating in other ways.

>>61906392
MOAR TRANSISTORS

>>61906392
We can't out-innovate big O

>>61903844
Ibm did 9/11

>>61903568
No 1nm is because then you run out of numbers.

Thank god, humanity was going crazy with that much technology improvement. I just want to be comfy with my Core2Duo

>>61906530
> 0.7nm
> 0.5nm
> 0.01488nm

>>61903568
>What that could mean for things like IoT sensors could be quite phenomenal.

-> Cameras and microphones everywhere.
We're fucked.

>>61906576
/pol/ posters confirmed idiots

>>61906392
Failing to shrink is why chips have barely improved over the last 4 years.

There just isn't any other way.

>>61906392
Miniaturization is literally how we went from industrial plant size computers to pocket calculators that could outperform them.

There is not much else that can be done to improve performance if miniaturization is no longer possible with silicon or other substitutes.

>>61906622
Oh sorry anon, did I hurt your feelings, how about a different measurement variation then.
> 7 ångström
> 5 ångström
> 1.488 ångström

Yes, I'm aware I'm getting into the diameter of an atom by that point.