Chinese semiconductor industry

Orthan said:

Perhabs you know more than me, but i refuse to believe that one researcher produces the same as five researchers. It depends on how do you organize the work. And they already have a lot of competition.

Click to expand...

1) We’re not talking about researchers but engineers. Researchers explore the physics and figure out the mechanisms. Engineers build the actual products. Companies don’t do research. They build products.

2) We’re not talking about the number of engineers, but the number of teams formed to do the work. In other words, the question of the number of companies in this space is a question of how you organize the work, and per my original comment, if you have too many engineers working on the same problem that usually impedes rather than accelerates progress.

Anyways, how innovation dynamics work in a market environment is usually a pretty complicated process not conducive to reductive truisms that media commentators like to employ. I was just pointing out the contradictory attitudes people seem to have about these things. It seems which truism they believe has more to do with the identity of the subject than the logic of the truism.

China NAND Flash Maker YMTC Aims to Lead Rivals, Ex-Chairman Says

latenlazy said:

We’re not talking about researchers but engineers. Researchers explore the physics and figure out the mechanisms. Engineers build the actual products. Companies don’t do research. They build products.

Click to expand...

How will they learn how to build better products if they dont do research? In order to build anything, its obvious that you must know how to do it.

And companies dont do research? thats a joke.

Orthan said:

How will they learn how to build better products if they dont do research? In order to build anything, its obvious that you must know how to do it.

And companies dont do research? thats a joke.

Click to expand...

The research for the products these companies are trying to build has already been done. Otherwise they wouldn’t have any product to chase commercial opportunities with. But more to the point the number of firms entering to build the equipment has no bearing on the effectiveness or quality of research, which itself has no bearing on whether firms are being spread thin when trying to engineer the product.

I probably should have specified “The goal of these companies is not to do research but to build products”. The point is that insofar as companies do do research the ultimate goal is still to assemble a product, so the point about whether there’s a limit to the number of engineers you can productively have working on any one project still stands. If you needed me to specify all that to understand my basic point I have trouble believing that you’re being earnest in this discussion.

from cnTechPost

Former Xiaomi chip chief allegedly joins NIO to lead the company's chip ambitions
2020-11-03 17:12:37 GMT+8 | cnTechPost
3

Former OPPO hardware director and head of Xiaomi's chip business Bai Jian has now joined NIO as hardware VP, reporting directly to NIO founder and chairman William Li Bin, according to industry media .
Bai graduated from Beijing University of Posts and Telecommunications with a doctorate degree in cryptography and communications engineering, and his appointment shows NIO's determination to develop its own chips, the report said.


The report is consistent with recent rumors. As previously mentioned by cnTechPost, NIO is planning to develop its own self-driving computing chip, and the project is being driven primarily by Li himself.

In line with the plans, NIO has also set up a separate hardware team, internally called "Smart HW (Hardware)". According to 36kr.com, Li has been looking for a technical leader with a silicon-based background since a few months ago.
Sources close to Li said that although the chip project has not yet been discussed by the board of directors, Li's intentions have been clear, and he is thinking about the final structure and has been making some advance communication to the company's executives and shareholders.
The report quoted a chip practitioner as saying that the capital investment required to develop their own chips could range from 1 billion yuan to 1 billion US dollars, depending on the extent of research and development.

Just this August, Ren Shaoqing, former director of R&D at Momenta, has joined NIO as assistant vice president, reporting directly to William Li Bin, who, according to public records, was formerly director of R&D at Momenta.
According to public records, the latter, a former director of R&D at Momenta, is a graduate of the joint PhD program between the University of Science and Technology of China and Microsoft Research Asia.
In response to the chip-making report, NIO was quoted by iAutoDaily as saying, "NIO has been continuing to invest in the field of automated driving technology. We will communicate details about AD's specific technology route, research and development hardware and software when the time is right, and we can't respond to market rumors one by one."

To the experts a question, if Huawei can able to produce and design this flash memory, then they can able to replicate a NVIDIA like GPU? or the capability to produce one?

from cnTechPost
Huawei reportedly uses new super-fast flash memory in Mate40 Pro that it developed itself
2020-11-03 19:36:35 GMT+8 | cnTechPost
1

Recent reviews by several digital bloggers have claimed that the Huawei Mate40 Pro's flash memory read and write speeds are superfast, and they speculate that this could be because Huawei has optimized the UFS 3.1 technology to the max. However, the newly released teardown report shows that this is not the case.
A Chinese tech reviewer first spotted the flash memory with the HiSilicon logo on it after the Mate40 RS Porsche design was torn down, which seems to have indicated that Huawei already has its own flash memory.


Other users have claimed that Huawei's Mate40 Pro, 40 Pro+, and Mate40 RS Porsche designs use a new type of flash memory (possibly sfs 1.0) that Huawei has developed in-house.

However, Huawei has never publicly mentioned this technology, nor did it mention it at the launch, so technical details are still unclear.
However, measurements have already shown that the Huawei Mate40 Pro's flash memory performance is extremely strong, with the Huawei Mate40 Pro achieving sustained read and write speeds of 1,966MB/s and 1,280MB/s respectively.


For comparison, the Mi 10 Supreme Edition reads and writes at 1772MB/s and 789MB/s, respectively; the Samsung Note 20 Ultra at 1750MB/s and 736MB/s.
Evidently, the Huawei Mate40 Pro flash memory write improvement is very obvious, an increase of more than 70%.
The Mate40 Pro's random reads are 383MB/s, and random writes are an exaggerated 548MB/s, almost double that of a typical UFS 3.1 model.

Worst case so what if Huawei cellphone and base station business died.

Still has ZTE to pick up the slack. now it can do both like huawei. Unisoc also make 5g cellphone chipsets.

Huawei has other roles to play. Spread its harmony OS and HMS like Google. And bring up domestic equipments.

latenlazy said:

The research for the products these companies are trying to build has already been done.

Click to expand...

If so, why doesnt SMEE build EUV if they know how to build it?

latenlazy said:

But more to the point the number of firms entering to build the equipment has no bearing on the effectiveness or quality of research

Click to expand...

When you "divide" the companies, you also divide the researchers, with each company trying to do the same research work, instead of pulling minds together and dividing work.

latenlazy said:

I probably should have specified “The goal of these companies is not to do research but to build products”.

Click to expand...

I agree, but in order to build products, you have to know how to do it, right?

Orthan said:

If so, why doesnt SMEE build EUV if they know how to build it?

Click to expand...

...because it takes time to build products and calibrate and test them for performance and reliability. For example, most of ASML’s development time for EUV was not spent doing research. It was spent building and testing the machines and reiterating on the engineering and design in order to fit customer requirements. Building a product isn’t inventing new knowledge. It’s about figuring out the gritty details of how to turn mature knowledge into practical solutions.

When you "divide" the companies, you also divide the researchers, with each company trying to do the same research work, instead of pulling minds together and dividing work.

Click to expand...

Most of the primary research is done in academia or shared across firms. Insofar as research is done within a firm, it’s usually secondary to a specific applications method, aka auxiliary to the some engineering approach specific to the product they’re building. That research can be published openly, and companies also can set up collaborative ventures and information sharing schemes. Just because there are more players that doesn’t mean the research is also being split and isolated. And even in research you don’t need 100 people looking at the same exact idea or hypothesis. You actually cover more ground with more people working in smaller teams breaking down a subject area and exploring different routes. The knowledge is what needs to be shared, not the projects or the company identity.

Nothing stipulates that the knowledge needs to stay in the company it originated from. Anyone who’s studied how Silicon Valley even got set up would know that within an industry knowledge spillover across firms is rather typical and a primary driver for developing strong ecosystems. If China has over 500,000 EE grads, whether they’re engineers or researchers, it doesn’t need to concentrate 100,000 each into five companies. This would be far more wasteful than having 1,000 each in 500 companies.

Furthermore, and more to the point, most new entrants are not going to focus on research. New entrants especially are primarily taking knowledge and talent that already exists and focusing on execution. No one researches their way into catching a market. You catch markets by building and selling product.

I agree, but in order to build products, you have to know how to do it, right?

Click to expand...

Tech companies aren’t siloed islands, and firms aren’t where most of the knowledge building happens.

visitor123 said:

building a house is not hard man. Just watch a youtube video or something. You don't need to do any research. You don't even need to be literate. Why don't you do it instead of paying for a house?

are you blind and deft and illiterate at the same time? Or you are mentally incapable (most likely)?

Edit: oh and if you try to do it, why isn't it finished TODAY??? how about TOMORROW??? NEXT WEEK??? LMAO

Click to expand...

Deaf.

Not illiterate.

pg