What's new

“Death Star” response from US would lock Russia out of 5G, advanced chips

Hack-Hook

ELITE MEMBER
Jan 11, 2012
14,364
3
14,027
Country
Iran, Islamic Republic Of
Location
Iran, Islamic Republic Of
lol 5G telecommunication does not depend on 7nm nodes or even 14nm nodes.

Huawei's advanced unique 5G chips which are still unsurpassed by others are not allowed to be fabricated by TSMC and no Chinese foundry can make those 5G chipsets. But they still hold all the patents on dedicated 5G chips and the end result is still clear.

China's 5G rollout is many times greater than Europe and USA's combined until only recently when western Europe and USA caught up. Speeds are still greater in China and utilization. 5G has been operating on farms, construction, mines, street cleaning vehicles and of course phone users since two years ago. They are barely reaching phone users in western Europe and USA now. The speeds are nothing comparable and already China has launched two separate 6G technology testers once in 2020 and one this year.

5G requiring 7nm nodes is absolute bullshit. The equipment is not a smartphone and limited by space. 5G equipment is like military hardware and even 200nm nodes will do as India uses for some of their military hardware.

China already can do 28nm nodes and trying to deliver commercial scale 14nm and 7nm. Lab level production has already been done for over two years. They cannot yet get it to commercial scale with controlled costs though. 28nm is already plenty for equipment. Even 280nm is acceptable. It's just 10cm^3 more bulk maybe.

What China did in meantime to overcome the volume issue is stacking and bonding techniques. Even Samsung is applying new bonding and stacking techniques following this path since it makes their products even more efficient on at least space.
I doubt they made those military equipment on 28nm even , more likely around 100nm .
the writer of the article has mistaken military equipment with, laptops and smart phones
 

KrimesKaster

FULL MEMBER
Dec 22, 2021
250
0
136
Country
Canada
Location
Canada
I doubt they made those military equipment on 28nm even , more likely around 100nm .
the writer of the article has mistaken military equipment with, laptops and smart phones
It depends. Autonomous drone platforms might need access to such semiconductor nodes just to get large enough memory chips to process scene data in real-time.
 

Hack-Hook

ELITE MEMBER
Jan 11, 2012
14,364
3
14,027
Country
Iran, Islamic Republic Of
Location
Iran, Islamic Republic Of
It depends. Autonomous drone platforms might need access to such semiconductor nodes just to get large enough memory chips to process scene data in real-time.
Actually they don't need such thing . They more likely needs more resilient chips and those industry and military grade chips are usually built on larger nodes
 

KrimesKaster

FULL MEMBER
Dec 22, 2021
250
0
136
Country
Canada
Location
Canada
Actually they don't need such thing . They more likely needs more resilient chips and those industry and military grade chips are usually built on larger nodes
Look at some of the new smart weapons from Israel like new ATR experimented in SPICE-250 bombs from Rafael. It uses deep-learning models for recognition and discrimination of targets. I highly doubt you will be able to evaluate a computer vision deep-learning model without several hundreds of megabytes to a giga-byte of fast and high-bandwidth RAM. To do it in realtime means being able to operate on large number of layer weights and inputs at a go and that needs bandwidth. Yes, I think you need new semiconductor processes for manufacturing those parts. And to be honest, 28-nm is not the newest.

 

serenity

FULL MEMBER
Jan 9, 2007
1,689
0
4,098
Country
China
Location
Australia
I doubt they made those military equipment on 28nm even , more likely around 100nm .
the writer of the article has mistaken military equipment with, laptops and smart phones

Yeah military equipment for example on a tank or in IFV for most part are like 200nm and above and that's already "advanced". For some military equipment though, more advanced chips are actually necessary for how they consume power and space and weight for example on a missile, EW unit, aircraft radar.

You can still make pretty good 2017 2018 level computers and phones with 28nm tech. Like it is better than not having any of the equipment for making sure your society still functions. The problem is they are not going to be sell-able if there are modern ones on the market. It is more a problem of making sure your country still runs even if total chip ban is implemented. It just means your own electronics devices are not competitive which means they cannot sell products and make money. The problem is Chinese electronics are so integrated with the global supply chain. Sony (Japan), Philips (Dutch), Siemens (German), General Motors (US) all have components that are Chinese supplied. Sometimes many components sometimes just one or two. Chinese OEMs and companies also buy a great deal of the latest chips for their products too. Will the supply chain stakeholders be able to carry such losses? They certainly haven't even tried to imagine not supplying China with chips. It is like asking Russia not to supply Europe with energy. It's got good side and bad side for both sides lol.

If USA thinks they have a net gain by doing something, they immediately do it. There is nothing they have on risk for implementing total chip ban. China is at 28nm already, 14nm process and 7nm along the same DUV method but just not able to make it commercial and economical. It will take time maybe months maybe years. 5nm and down to 2nm seems to require EUVL. This means China must master EUVL theory and then the technology itself AND also make it commercially viable which may take a long time if ever successful. Good thing is 2nm is limit and exponentially expensive that many industry experts predict 2nm will just be some halo products and super limited. Every major player (China, USA, Japan, Korea, Taiwan, UK) are currently chasing the holy grail alternative to silicon as a similar physics foundation and alternative physics for computing e.g. using photons and making quantum computing technology capable of performing real world useful computation tasks which they are unable to. That's a different side though and even if figured out and they have the dedicated "chips" or processes, they also have to make the size the same as today's Laptops and smartphones.

I mean the current Chinese, photon based 66 qubit system looks like this.

1643371437781.png



and IBM's 53 qubit system (silicon based I think?) looks like this.

1643371506631.png


So a long way from scaled down but these computers are the supercomputers of quantum computing world so normal devices will not be that large. These are the first steps only.
 
Last edited:

serenity

FULL MEMBER
Jan 9, 2007
1,689
0
4,098
Country
China
Location
Australia
Wouldn't they import semiconductors from China. China makes more chips anyways.

For some Russian space and military equipment that require advanced chips, they apparently used to get some from South Korea of chips they couldn't buy off the shelf commercially for whatever reason e.g. mods required for military task or security concerns.

Since then apparently South Korean supply has been disrupted for some reason probably US action. There were rumors that Russia have sought some support in chips supply from China after that and then no more whispers of those issues. China at 28nm as most advanced ability of lower production scale doesn't match South Korea's 7nm ability though! So there is some changes the Russians would need to do and work with China on if they previously used Korean 7nm or even 14nm chips.
 

Hack-Hook

ELITE MEMBER
Jan 11, 2012
14,364
3
14,027
Country
Iran, Islamic Republic Of
Location
Iran, Islamic Republic Of
Look at some of the new smart weapons from Israel like new ATR experimented in SPICE-250 bombs from Rafael. It uses deep-learning models for recognition and discrimination of targets. I highly doubt you will be able to evaluate a computer vision deep-learning model without several hundreds of megabytes to a giga-byte of fast and high-bandwidth RAM. To do it in realtime means being able to operate on large number of layer weights and inputs at a go and that needs bandwidth. Yes, I think you need new semiconductor processes for manufacturing those parts. And to be honest, 28-nm is not the newest.

If they want to use multipurpose commercial chips, but many time they use specialized chips that is suited for only one task, they cant compare it like that with a deep learning program that ran on your laptop
 

KrimesKaster

FULL MEMBER
Dec 22, 2021
250
0
136
Country
Canada
Location
Canada
If they want to use multipurpose commercial chips, but many time they use specialized chips that is suited for only one task, they cant compare it like that with a deep learning program that ran on your laptop
Fabrication process does NOT mean that chip will be multipurpose. Anyways, I was talking about memory. Each algorithm has memory requirements, does not matter where you run it. Deep Learning models have pretty high memory requirements.
 

Hack-Hook

ELITE MEMBER
Jan 11, 2012
14,364
3
14,027
Country
Iran, Islamic Republic Of
Location
Iran, Islamic Republic Of
Fabrication process does NOT mean that chip will be multipurpose. Anyways, I was talking about memory. Each algorithm has memory requirements, does not matter where you run it. Deep Learning models have pretty high memory requirements.
and you can build memory on 28nm nodes
 

925boy

SENIOR MEMBER
Sep 10, 2016
4,921
-30
5,481
Country
United States
Location
United States
What a nation of scheming pathological liars
But people like @F-22Raptor , the OP who started this thread, Believe all these lies.

@F-22Raptor why did you start this thread AND THEN START DUCKING? People are proving you wrong now ,cuz maybe US gov pays you to post propaganda on PDF, i dunno, just a guess, b ut why are you hiding? i havent seen you post since your original post on this thread, you afraid some of these PDF members will cook your bad arguments and propaganda? c'mon bro, do better in 2022!
 

Users Who Are Viewing This Thread (Total: 1, Members: 0, Guests: 1)


Top Bottom