Chinese semiconductor industry

Status
Not open for further replies.

quantumlight

Junior Member
Registered Member
Like I said, there is a roadmap for advancing lithography for the next decade. Who knows what will happen after that.
Anyone can come up with a roadmap, there have always been roadmaps... heck intel was supposed to be at 10Ghz ten years ago, according to one of their roadmaps... or 3nm by now, according to another roadmap... hasn't happened and won't happen.

Its not just the death of Moore's Law but also hitting the wall of Dennard scaling.. CPU stopped getting faster a long time ago, so they said well lets multicore and scale up the cores, problem is most applications were still optimized for single core, and not every application is going to benefit equally from multiple cores, and even core scaling quickly reaches point of diminishing returns

So they try this trick and that trick, 3D stacking this and DLSS that.... but all to find efficicnes still could not keep Moore's Law alive... You do not understand the definition of Moore's Law if you state the time factor doesn't matter.

As far as semiconductor goes, this is the end of the road, the writing has been on the wall for a long time, and over the last decade its already proven Moore's Law was dead and no longer tracking... whatever happens there may be a new technology that leapfrogs IC but it won't be with the current paradigm... we are not debating a decade from now, Moore's Law has already been dead for more than one decade. Moot point dude. but nice try
 

gelgoog

Lieutenant General
Registered Member
Yeah who is the joker.
...
Anyone can come up with a roadmap, there have always been roadmaps... heck intel was supposed to be at 10Ghz ten years ago, according to one of their roadmaps... or 3nm by now, according to another roadmap... hasn't happened and won't happen.

Its not just the death of Moore's Law but also hitting the wall of Dennard scaling.. CPU stopped getting faster a long time ago, so they said well lets multicore and scale up the cores, problem is most applications were still optimized for single core, and not every application is going to benefit equally from multiple cores, and even core scaling quickly reaches point of diminishing returns

So they try this trick and that trick, 3D stacking this and DLSS that.... but all to find efficicnes still could not keep Moore's Law alive... You do not understand the definition of Moore's Law if you state the time factor doesn't matter.

As far as semiconductor goes, this is the end of the road, the writing has been on the wall for a long time, and over the last decade its already proven Moore's Law was dead and no longer tracking... whatever happens there may be a new technology that leapfrogs IC but it won't be with the current paradigm... we are not debating a decade from now, Moore's Law has already been dead for more than one decade. Moot point dude. but nice try

Why not? The best AMD APU has an 8-core CPU with 4.4 GHz speed and 2150.4 GFLOPS GPU.
It is available and in stock at Amazon for $459. No need for a scalper. That is about the GPU performance of a GTX 1650.
Once GPU prices fall you get a card and dump it into the system.

There is some talk Intel might re-enter the GPU market. If that is true (I doubt it) then prices might come down.

Just like predicting the weather the farther way the prediction is the less accurate it will be.
We are talking about TSMC production 2 years from now.

With regards to scalar performance scaling, a lot of people have been working over the past decade on how to parallelize algorithms and software, the fact you get GPU software acceleration for coin mining is just one of the visible examples of that. GPU architectures and software models are precisely one of the ways to get around scalar performance limitations.

Moore's Law was never about processor clock speed. The human brain operates at around 100 Hz clockspeed and has more computer power than a supercomputer.
 
Last edited:

gelgoog

Lieutenant General
Registered Member
Also, if what you want is high clockspeeds, they have manufactured 50 GHz transistors in the lab with CVD Diamond instead of Silicon.
 

Oldschool

Junior Member
Registered Member
Related to xinjiang cotton, China is quick sanction on foreign companies, people and institutions.

But regarding to tech war, I have not seen any action from Chinese side.

It's own development will takes time. Come back and check in couple years.

Meanwhile it needs to show something in this tech war besides internal development.

sue TSMC and take over its Fab.

And forbid Chinese companies from using any taiwanese fabs. Might as well, since all the important players already got sanctioned by US. Don't let smaller and less significant players continue contribute to TSMC's revenue

Shut down some US semiconductor
companies in China.

Xinjiang cotton and tech war shows big discrepancy in actions. From outside , people only see China continue to beating on the tech war. Show something, anything.

Go to major chinese electronic new site, like eefocus, US company ads from TI, ADI filled up the site. What happen to deamericanize tech drive? Government ought to have task force to monitor and crack down on those US companies continuing to infiltrate Chinese society.
 
Last edited:

supersnoop

Major
Registered Member
Also forgot to mention in my previous post, actually a good example of trying to reinvent all components is Japan.
Not specific to semiconductors, but in many industries the Japanese companies want(ed) to control the whole supply chain for the product.

So a Panasonic CD player is using Panasonic laser diode, Panasonic Capacitors, Panasonic Batteries, etc., etc.
Toyota Car is using Aisin transmission, Denso air conditioner, etc.

The advantage is that the Japanese products have a good reputation for quality, but expensive to produce and when repairs are needed, the parts are hard to find and expensive since they were custom made for the product.

You can see the result as the entire consumer industry for them basically evaporated in the last 20 years since their cost became so uncompetitive and hardware became less important than software.

RISC-V is the future for China and the true free world with ARM falling into the Nvidia/US camp...

I did a post of this in the other technology thread. I hope RISC-V and OSS do gain more traction. I also see there is interest from the EU to reduce the dominance of American tech companies. We see a lot of software innovation in the last 20 years since the collapse of the wintel domination of the Personal computing market. I think OSS played a big role in it.
 

Appix

Senior Member
Registered Member
I know that in 2019 and 2020 three new EDA startups have been established. People with knowledge should keep an eye on those EDA start-ups in coming months and years. How much of a chance do they have in succeeding? Anyone with knowledge about this niche in the semiconductor supply chain.

1. X-EPIC in Nanjing (founding: March 2020)

Please, Log in or Register to view URLs content!

2. Hejian Industrial Software in Shanghai (founding: May 2020)

No website

3. Amedac (founding: September 2019)

Please, Log in or Register to view URLs content!

Source:
Please, Log in or Register to view URLs content!
 
Status
Not open for further replies.
Top