Go Back   OC3D Forums > [OC3D] General Forums > OC3D News
Thread Tools Display Modes
Old 21-12-18, 02:38 PM
WYP's Avatar
WYP WYP is offline
News Guru
Join Date: Dec 2010
Location: Northern Ireland
Posts: 19,270
TSMC cleared to construct 3nm chip factory in Southern Taiwan

Production is set to start in 2022/2023.

Read more about TSMC's constructing a 3nm foundry in Taiwan.

Twitter - @WYP_PC
Reply With Quote
Old 21-12-18, 03:45 PM
Dawelio's Avatar
Dawelio Dawelio is offline
OC3D Elite
Join Date: Aug 2014
Location: Scandinavia
Posts: 5,123
The money involved, $19.5 billion dollars! Jesus christ!
Reply With Quote
Old 22-12-18, 03:01 PM
RobM's Avatar
RobM RobM is offline
OC3D Elite
Join Date: May 2016
Location: Durham
Posts: 1,558
Pretty impressive they are still able to push the envelope at that cost
ASUS ROG Crosshair VII wifi | Ryzen 5 3600 | Custom Loop | Patriot Viper Steel 4133 @ 3600 2 x 8Gb | Asus Strix RX 470 | Samsung 500gb nvme | Samsung 1Tb nvme | Samsung Evo 120gb ssd | Seagate Barracuda 1tb storage | Toshiba 1Tb HDD | EVGA 500w | Phanteks Entho Pro Tempered Glass | Windows 10 and Linux Manjaro
Reply With Quote
Old 22-12-18, 04:18 PM
ImprovizoR's Avatar
ImprovizoR ImprovizoR is offline
OC3D Crew
Join Date: Jan 2015
Posts: 640
There's only so much they can do with silicon though, right? We're almost at the end of the road.
Reply With Quote
Old 22-12-18, 11:26 PM
Avet's Avatar
Avet Avet is offline
OC3D Elite
Join Date: Dec 2016
Posts: 1,642
Originally Posted by ImprovizoR View Post
There's only so much they can do with silicon though, right? We're almost at the end of the road.
I think 2.5 nm is sort of the minimum. But we are long way from using something other than silicon. It is not the size that matters, it is how you use it. (That is definitely not what she said ) The most important of all is density, and then the architecture itself. Current multiple patterning lithography has reached it's limit. EUV lithography is what will be the main process for a long time. And for the far future, brains at MIT and University of Colorado are working on ALE (Atomic Level Etching) technique. That is some crazy science fiction stuff. Amazeballs! Also more, and more manufacturers are turning to software optimizations, rather than raw horsepower. There is so much performance to be gained there.

Silicon is here to stay. I don't see it leaving for decades.
Reply With Quote
Old 26-12-18, 03:42 AM
realneil's Avatar
realneil realneil is offline
OC3D Crew
Join Date: Sep 2010
Location: Virginia, USA
Posts: 607
3nm, Man how things progress over time!

Reply With Quote
Old 26-12-18, 02:13 PM
tgrech tgrech is offline
OC3D Elite
Join Date: Jun 2013
Location: UK
Posts: 2,170
Originally Posted by Avet View Post
Silicon is here to stay. I don't see it leaving for decades.
It's going to be a while, maybe another decade, but it's unlikely it will be much more than that, the closer we get to the limits, the more money gets thrown at finding alternatives, technological progress & improvements to efficiency on a grand scale are fundamental to the progress of humanity & are almost always worth far more money than is put into them(Even when that input money reaches the 10's to hundreds of billions), so whether we put more money into optimising designs in silicon or finding alternatives basically depends on which one has the best cost:benefit ratio, if rewriting decades worth of software from the ground up starts to become a requirement of utilising progressive technologies then the cost of the software development will become dominant and eventually that dynamic is going to flip.
There's a fair few materials that have had a few decades of research, which are mostly limited by the fact we don't have ways to cheaply/reliably mass produce them as circuits, but as MCM's are set to become ubiquitous in all high end processor industries by 2020 we're no longer going to have to use just the material that is the most versatile, we could see these currently exotic materials used in very small scales initially for specific areas they're suited to.

Optronics is another part of the puzzle, ditching electrical signalling altogether would solve bucket loads of problems if we had more efficient & cheaper semiconductor light emitter/readers, but this is again dependant on semiconductor materials progress.

But yeah, there's still plenty of optimisations to find in existing silicon transistors, as well as stacking techniques too, while MCMs would allow total die size to continue growing in line with Moores law till long after silicon scaling ends. I think the first uses we will see for carbon sheets(Graphene/carbon nanotubes) will be in stacked silicon dies primarily to improve heat dissipation, possibly with some functionality as I/O elements(We can already do basic circuits with nanotubes), while their use in wireless modems as their active-component-debut (Possibly 6G if not 5G networks) is almost guaranteed.
Reply With Quote

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

All times are GMT. The time now is 05:50 AM.
Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2021, vBulletin Solutions, Inc.