Chinese semiconductor industry

[Xizor]

More people are probably more concerned with storage size than processing speed, unless they are using it for gaming.

People may not directly opt for processor capability but they indirectly vote towards it by having a positive bias/ choice for reduced load times of applications, faster running of processes ( image editing, video editing, gaming) and supporting more feature heavy applications.

So its an indirect vote. Processor capability aren't that noticeable for smartphones released an year or two apart BUT it is noticeable for models with bigger gaps like four or five years.

 

[FairAndUnbiased]

People may not directly opt for processor capability but they indirectly vote towards it by having a positive bias/ choice for reduced load times of applications, faster running of processes ( image editing, video editing, gaming) and supporting more feature heavy applications.

So its an indirect vote. Processor capability aren't that noticeable for smartphones released an year or two apart BUT it is noticeable for models with bigger gaps like four or five years.

how many people do image or video editing on phones besides some simple filters? For nonprofessional stuff, compression algorithms for the images can really reduce load time. For professional stuff people use desktop.

 

[antiterror13]

how many people do image or video editing on phones besides some simple filters? For nonprofessional stuff, compression algorithms for the images can really reduce load time. For professional stuff people use desktop.

definitely not me .. I would do it on my super powerful and big screen PC

 

[Xizor]

how many people do image or video editing on phones besides some simple filters? For nonprofessional stuff, compression algorithms for the images can really reduce load time. For professional stuff people use desktop.

Can VR and AI powered features be done with speed on old processors? Can 4K video capture and handling be done with old processors?
And gaming, battery life...

I prefer using the smartphone for handling documents and signing them.

 

[Overbom]

Can VR and AI powered features be done with speed on old processors? Can 4K video capture and handling be done with old processors?
And gaming...

I prefer using the smartphone for handling documents and signing them.

All these features need fast (and specialised) chips.

Also the current "hot" tech which is actively being developed is Augmented Reality (AR). And AR eats processing power like its nothing

As such, customers will keep (as they should) demanding more and more performance from chips in mobile phones.

I predict that by 2023 AR will seriously start taking off. Apple is already well positioned for such a future

 

[AndrewS]

PLA may not even need to destroy the fabs themselves, if the rumors about TSMC destroying its factories upon capture are true.

I think it very likely that TSMC have such contingency plans to destroy their fabs if there is an invasion.

They probably think it would serve as a deterrent to China. And perhaps it was actually a real deterrent in times past.

But now, China can simply threaten severe punishment to anyone who carries out such plans.

After all, they will not be able to escape Taiwan as there are no planes or ships able to leave

 

[FairAndUnbiased]

Can VR and AI powered features be done with speed on old processors? Can 4K video capture and handling be done with old processors?
And gaming, battery life...

I prefer using the smartphone for handling documents and signing them.

You can get around that by architecture i.e. instead of trying to cram everything onto a SoC, you separate CPU and rudimentary video functions to one side, and drive a dedicated GPU when doing something graphics intensive. You can stack this in one package with through silicon vias so there's no additional board space required.

BTW, 4k video

Please, Log in or Register to view URLs content!

which is made on

Please, Log in or Register to view URLs content!

There was even a

Please, Log in or Register to view URLs content!

And remember you are not talking about ordinary market conditions. You are talking about a total cutoff. So it's not "do I choose between 14-22 nm domestic chips with lower battery life, or foreign chips with better battery life" it is "do I choose between 14-22 nm domestic chips with lower battery life, or do I go back to the stone age".

 

[caudaceus]

Since you work in the industry @FairAndUnbiased . Do you think it's feasible for China to have domestic EUV by 2025-2027?

 

[FairAndUnbiased]

I think it very likely that TSMC have such contingency plans to destroy their fabs if there is an invasion.

They probably think it would serve as a deterrent to China. And perhaps it was actually a real deterrent in times past.

But now, China can simply threaten severe punishment to anyone who carries out such plans.

After all, they will not be able to escape Taiwan as there are no planes or ships able to leave

Don't think it matters. The fabs are junk once combat starts anyways, because the power grid need to be targetted and if a fab loses power unexpectedly for even a few hours everything in progress is junk and alot of the equipment goes down.

If they don't destroy it, the only worth of the fabs will be salvaged equipment (maybe some DUV litho tools and wet chem tools). Alot of the gas phase chemistry equipment like ion implanters, etchers and deposition are going to be junk due to losing vacuum and leaking corrosive, toxic chemicals. The engineers are likely to flee or go into hiding.

 

[FairAndUnbiased]

Since you work in the industry @FairAndUnbiased . Do you think it's feasible for China to have domestic EUV by 2025-2027?

I think 14-28 nm domestic immersion DUV is going to come out soon, within 1-2 years, since ArF laser is already made so the big bottlenecks are the mechanical stage and water management system. Once you have ArF laser, it is not such a large leap to go to immersion.

Building on immersion DUV experience, the transferrable tech to EUV is only the mechanical stage. However, everything else will be different, since EUV does not propagate in air and thus you need vacuum for transmission, since EUV photoionizes air. That means everything must now be retooled from atmospheric pressure to 100% gas tight vacuum including literally every single component. That is why ASML takes so long: EUV is far more rigid in its requirements than DUV and every single part is much more expensive.

The problem is that with vacuum, outgassing becomes a huge deal, including the specific chemistry of the outgassed species since EUV can also photopolymerize contaminants on top of the wafer or on mirror optics and destroy it. You cannot have any plastics in the EUV light stream either, since they will degrade, outgas, and potentially photopolymerize elsewhere. That will require new development of all mechanical stages and optics for vacuum and EUV compatibility.

Due to the massive number of parts that must be vacuum certified or developed to be vacuum compatible instead of merely cleanroom compatible, I put EUV at 5-8 years out. A 2025-2027 timeframe is optimistic. I put it at "pre 2030".