A reappraisal of China's semiconductor strategy
- Thread starter ZeEa5KPul
- Start date
- Status
Jura The idiot
General
I knew I'd receive some more, LOL all I can do is attaching 'Like'
in the meantime ate my dinner and read the first one I mean #314 weig2000, Today at 2:56 AM
Jura The idiot
General
kind of gallows humor inside:
“In the same way racial profiling of African Americans as criminals may create the crime of ‘driving while black,’ ” wrote Kim, who practices law at the Houston office of Greenberg Traurig, “profiling of Asian Americans as spies … may be creating a new crime: ‘researching while Asian.’ ”
Jura The idiot
General
after May 28, 2019
I thought it was all about financing
(funding bodies usually try not to commit their money if they know the research may get funds from elsewhere),
but I'm not sure anymore,
especially if this was true (source of course is ):
"One in five of the Chinese-named defendants was never found guilty of espionage or any other serious crime in the cases between 1997 and 2015—almost twice the rate of wrongful accusations among non-Chinese defendants."
in short, I don't tend to believe in coincidences, LOL
Quantum computers are still at very early stages, so China is well-poised simply because it's running this race from the same starting point as everyone else. It didn't jump in late like with conventional CPUs. There are actually quantum tunnelling effects that prohibit chips from having transistors smaller than 3nm (not really, but close enough) - just like heat sink limitations keep clock speeds at around 3GHz. That's one or two generations from the present state of the art (3 or 4 from where China is); Korean/Japanese/Taiwanese manufacturers will hit a wall in the coming years, which should give China a chance to catch up and a big role in deciding what comes next.
@ZeEa5KPul
That's a good observation about quantum computing.
The clock speed limit is however slightly above 5GHz at the moment and has been since mid 2000s. But there are still several innovation avenues to explore within conventional semicon and computing design, like:
1. moving away from silicon to higher performance materials like GaN. This could allow for higher clock speeds.
2. Improving power efficiency. We are still about 2 orders of magnitude above the theoretical limit
3. Design of reversible logic ICs for potentially collosal power efficiency improvements
That's a good observation about quantum computing.
The clock speed limit is however slightly above 5GHz at the moment and has been since mid 2000s. But there are still several innovation avenues to explore within conventional semicon and computing design, like:
1. moving away from silicon to higher performance materials like GaN. This could allow for higher clock speeds.
2. Improving power efficiency. We are still about 2 orders of magnitude above the theoretical limit
3. Design of reversible logic ICs for potentially collosal power efficiency improvements
Everyone laugh about analogue computers, forget that how difficult is to design them.
In the uni it took few hours of training to design (simple) digital circuit, but to design a simple analogue computer needs years of high level training ( complex ones usually designed by high level math guys)
The quantum computers way more difficult than the analogue computers. Digital needs monkeys to design and program, analogue require mathematicians and the quantum can use only geniuses.
The digital hardware technology died around 2012. There is still marketing lingering around, but that wall was hit more than half decade ago.
In the uni it took few hours of training to design (simple) digital circuit, but to design a simple analogue computer needs years of high level training ( complex ones usually designed by high level math guys)
The quantum computers way more difficult than the analogue computers. Digital needs monkeys to design and program, analogue require mathematicians and the quantum can use only geniuses.
The digital hardware technology died around 2012. There is still marketing lingering around, but that wall was hit more than half decade ago.
Optical Computing could be the next big thing. The problem would be creating an optical processor(photons instead of electrons) and therefore,more precisely, a photon enabled transistor. Something that'd capture photons and store it / release it - creating 1 and 0s . Such a processor would need a laser to "power the processor" and a optical to electronic converter.
The power sucked in would be by the laser( electricity to photons) and the optic to electrical converter and should stay within 65W to 100W. Mobile devices are out of question in the first stages BUT Servers, desktops and laptops ? Sure.
Such processors are still in research phase and the biggest hold-back is the lack of breakthrough in the creation of a viable Photonic transistor.
The performance gain would be massive, if breakthrough is achieved.
I imagine such processors would be 3D shaped like a rubix cube.
One can only dream.
The power sucked in would be by the laser( electricity to photons) and the optic to electrical converter and should stay within 65W to 100W. Mobile devices are out of question in the first stages BUT Servers, desktops and laptops ? Sure.
Such processors are still in research phase and the biggest hold-back is the lack of breakthrough in the creation of a viable Photonic transistor.
The performance gain would be massive, if breakthrough is achieved.
I imagine such processors would be 3D shaped like a rubix cube.
One can only dream.
- Status