Opinionated - China Chipping Away to Semiconductor Dominance

[j20blackdragon]

Chinese x86 Zhaoxin KX-7000 update.

First signs of tapeout and functional samples of the processor began to appear in September 2019.

Zhaoxin KX-7000(OctaCore) appeared on GeekBench in November 2019.

Wang Weilin, chief engineer of Zhaoxin, said that the performance goal of KX-7000 is to reach the level of AMD in the same period, and the target product is at least Zen2 architecture level.

At present, Zhaoxin has achieved 3GHz under the 16nm process.

Goal of KX-7000 is to achieve 8-core 4.0GHz at 7nm.

Best case scenario, can reach 80-90% performance of AMD Ryzen 7 3700X.

 

[Paul2]

That article from simmtester is a copy from article in EEtimes by Junko Yoshida two days earlier. That $70 billion claimed in 2017 never materialized.
https://www.eetimes.com/changxin-emerging-as-chinas-first-only-dram-maker/

If those $70B never materialises, then where did they go then?...

Almost 450,000,000,000 CNY is not a pocket change, you know. A second south-north water diversion project, or a second manned space program, a second 3 gorges dam, a half of existing HSR network, a big chump of an OBOR programme, or an entire megacity built from scratch...

 

[Han Patriot]

If those $70B never materialises, then where did they go then?...

Almost 450,000,000,000 CNY is not a pocket change, you know. A second south-north water diversion project, or a second manned space program, a second 3 gorges dam, a half of existing HSR network, a big chump of an OBOR programme, or an entire megacity built from scratch...

I am pretty sure you are not Chinese for one. Two China is rich but not that rich that we just buen 70 bil$, we are not US. Not materialise means they bought the land but the project did not progress beyond that and now they are earning from the land appreciation. But that deviates from their initial cause. Anyhow, China has twch foe 2 three types of mem now, NOR NAND and DRAM.

 

[Paul2]

Two China is rich but not that rich that we just buen 70 bil$, we are not US. Not materialise means they bought the land but the project did not progress beyond that and now they are earning from the land appreciation.

Yes, doing land speculations with state money on an astronomical scale is certainly not what their mandate is. Would they be a normal state company, their directors would've been double regulated years ago.

For such amount of cash they could've bought any memory maker other than Samsung few times over. And if they were hellbent on doing it with indigenously developed tech, they could've done it like 10 time over if they weren't hiring such incompetents.

China already had quite successful niche DRAM players who were doing just fine, so the argument that "Chinese tech can't do that" is no more than a joke. Xingmem has a thriving speciality RAM business, a fully privately owned CanSemi also does embedded DRAM. Both of them have no problem doing without astronomical sums in state subsidies.

Previously Unigroup bought and merged RDA and Spreadtrum in a "brilliant strategic move," and now the resulting company seem to show no life signs. Same was for their "Domestic FPGA" effort. Such cases are many.

I don't know what they are trying to do, and I think their directors don't know either.

We saw same patterns over and over. A state company opens a tons of nondescript holding companies for its investments, which in turn do the same, each then engaging in trading and loaning their shares to god knows whom for no apparent reason, with all men involved somehow making millions out of nowhere: https://zh.wikipedia.org/wiki/陈良宇事件

 

[Mohrenn]

Yes, doing land speculations with state money on an astronomical scale is certainly not what their mandate is. Would they be a normal state company, their directors would've been double regulated years ago.

For such amount of cash they could've bought any memory maker other than Samsung few times over. And if they were hellbent on doing it with indigenously developed tech, they could've done it like 10 time over if they weren't hiring such incompetents.

China already had quite successful niche DRAM players who were doing just fine, so the argument that "Chinese tech can't do that" is no more than a joke. Xingmem has a thriving speciality RAM business, a fully privately owned CanSemi also does embedded DRAM. Both of them have no problem doing without astronomical sums in state subsidies.

Previously Unigroup bought and merged RDA and Spreadtrum in a "brilliant strategic move," and now the resulting company seem to show no life signs. Same was for their "Domestic FPGA" effort. Such cases are many.

I don't know what they are trying to do, and I think their directors don't know either.

We saw same patterns over and over. A state company opens a tons of nondescript holding companies for its investments, which in turn do the same, each then engaging in trading and loaning their shares to god knows whom for no apparent reason, with all men involved somehow making millions out of nowhere: https://zh.wikipedia.org/wiki/陈良宇事件

Dude, you have a serious case of being a moron

 

[Paul2]

Dude, you have a serious case of being a moron

Good boy! Go get yourself a candy now

 

[Han Patriot]

Yes, doing land speculations with state money on an astronomical scale is certainly not what their mandate is. Would they be a normal state company, their directors would've been double regulated years ago.

For such amount of cash they could've bought any memory maker other than Samsung few times over. And if they were hellbent on doing it with indigenously developed tech, they could've done it like 10 time over if they weren't hiring such incompetents.

China already had quite successful niche DRAM players who were doing just fine, so the argument that "Chinese tech can't do that" is no more than a joke. Xingmem has a thriving speciality RAM business, a fully privately owned CanSemi also does embedded DRAM. Both of them have no problem doing without astronomical sums in state subsidies.

Previously Unigroup bought and merged RDA and Spreadtrum in a "brilliant strategic move," and now the resulting company seem to show no life signs. Same was for their "Domestic FPGA" effort. Such cases are many.

I don't know what they are trying to do, and I think their directors don't know either.

We saw same patterns over and over. A state company opens a tons of nondescript holding companies for its investments, which in turn do the same, each then engaging in trading and loaning their shares to god knows whom for no apparent reason, with all men involved somehow making millions out of nowhere: https://zh.wikipedia.org/wiki/陈良宇事件

Paul stop flooding like an Indian. Firstly there is no 70 bil invested, secondly thwir original intention was to set up a mem giant, guess who fcked it up, the land thwy acquired appreciated, they made a profit, not the intention none the less. BUY SAMSUNG? WHAT ARE YOU SMOKING? They tried buying micron, guess what happened? I am not arguing private companies are better the point is private companies will nEVeR invest in this expensive shit. The returns are too far away, China does it due to strategic reasons. GOING BY YOUR logic that our SOEs are like your PSUs, we should be India now and not a threat to US right? Common sense.

 

[Paul2]

Firstly there is no 70 bil invested

Where those 70 bil are then? And where were they supposed to get them in the first place?

secondly thwir original intention was to set up a mem giant, guess who fcked it up

Who?

They tried buying micron, guess what happened?

They haven't tried hard enough, low end people. Have they ever tried suing back? Have they considered an option buying Micron's overseas subsidiaries outside of CFIUS jurisdiction.

I am not arguing private companies are better the point is private companies will nEVeR invest in this expensive shit.

Do they need to invest in expensive shit in the first place? DRAM is an unprofitable cyclical industry, and out fabs have just started working when prices hit rock bottom. Speciality DRAM would've been a perfect bang for buck, with potential for further expansion into mainstream.

 

[Han Patriot]

Where those 70 bil are then? And where were they supposed to get them in the first place?

Let me teach you simple math and logics, if I intend to buy a house for 100 dollars but end upp buying the land for 20 dollars, where is the remaining 80$? Come on use some common sense my dear yindoo fren.

Who?

You typed a flood of contents and then you dunno who? Read again your post and mine.

They haven't tried hard enough, low end people. Have they ever tried suing back? Have they considered an option buying Micron's overseas subsidiaries outside of CFIUS jurisdiction.

Hahahh sure geniua, we should appoint you as the advisor right? Buying micron subsidiaries? Gosh, people need to sell for you to buy, and how does buying a subsidiary give you access to technology genius?

Do they need to invest in expensive shit in the first place? DRAM is an unprofitable cyclical industry, and out fabs have just started working when prices hit rock bottom. Speciality DRAM would've been a perfect bang for buck, with potential for further expansion into mainstream.

Do you know what is the memory market size in China? All semiconductor are expensive shit bhai. Tell me what are 'cheaper' speciality DRAM? What's the market size? That's the reason India is no where, talk too much. Now we are a threar to the Korean US monopoly. That's good for the market.

 

[JSCh]

Oppo developing its own mobile chips, says report | DIGITIMES
Lena Li, Taipei; Steve Shen, DIGITIMES
Friday 13 December 2019
Chinese handset vendor Oppo is currently developing a mobile processor in house, and the chip, dubbed M1, is an auxiliary processor, according to a China-based tech.qq.com report, citing company sources.

The report said company vice president Liu Chang had made the revelation at a recent technology forum. He said that Oppo currently boasts the ability to make its own chips, highlighted by the demonstration of its in-house developed VOOC fast-charging chips at the forum.

The report did not provide details of the chip.

At the forum, which was held to mark the completion of Oppo's new office building, company CEO Tony Chen also disclosed that Oppo will invest CNY50 billion (US$7.172 billion) to deepen its software and hardware prowess in the next three years, said the report.

In addition to enhancing its product portfolios, the company's future R&D push will be also eyeing the development of emerging technologies including 5G/6G, AI, AR and big data, Chen was quoted as saying.

Liu, who also serves as dean of Oppo's research institute, revealed that the company is scheduled to launch an array of new products, including smartwatches, true wireless stereo (TWS) headsets, and 5G CPE (customer premises equipment) products, in the first quarter of 2020.

Oppo is also developing new products utilizing flexible panels, Liu added.

 

[Paul2]

Hahahh sure geniua, we should appoint you as the advisor right? Buying micron subsidiaries? Gosh, people need to sell for you to buy, and how does buying a subsidiary give you access to technology genius?

Their fabs abroad work under subsidiaries. I myself had a coworker who once worked in Micron's China fab.

Do you know what is the memory market size in China?

Will it make any difference if you can't make money from it man?

Hynix will steamroll whatever newcomers in China now when they don't have to pay the tariff.

Buddy, I worked in electronics for longer than you

 

[Han Patriot]

Their fabs abroad work under subsidiaries. I myself had a coworker who once worked in Micron's China fab.
Will it make any difference if you can't make money from it man?

Subsidiaries and patents are 2 different things genius. At most you are buying facilities, equipment and sales channel, not patents or technology. Understand? What has making money got to do with buying technology or developing it? Strategically it is imperative for China to maintain a semiconductor industry even if it means losing some money. Why do you think US feels threatened, because we want technology regardless of the cost. They used the market economy argument to deter a potential rival. How many billions you think US sunk in research before passing it to the civillian economy?

Hynix will steamroll whatever newcomers in China now when they don't have to pay the tariff.

Buddy, I worked in electronics for longer than you

It doesn't matter, the market is so big in China, with a stroke of pen, all will be required to buy local if Trump does a national security shit again. We can use the same argument right? You need to look at this as an ecosystem not just one company. If Hynix is so confident, then good, nobody should be worried, let the chinks do the stupud thing, but history tells us whenever China does something, in the end we always get what we want, its a matter of time.

 

[Olli Ranta]

Subsidiaries and patents are 2 different things genius. At most you are buying facilities, equipment and sales channel, not patents or technology.

Just FYI, that article from Junko Yoshida has a last discussion item from Junko stating an agreement on patents:
http://www.wilan.com/news/news-rele...cense-and-Acquisition-Agreements/default.aspx As we speculated in our story, ChangXin Memory Technologies, Inc. has entered into a patent license agreement with WiLAN to acquire DRAM patents developed by former Qimonda.

 

[Han Patriot]

Just FYI, that article from Junko Yoshida has a last discussion item from Junko stating an agreement on patents:
http://www.wilan.com/news/news-rele...cense-and-Acquisition-Agreements/default.aspx As we speculated in our story, ChangXin Memory Technologies, Inc. has entered into a patent license agreement with WiLAN to acquire DRAM patents developed by former Qimonda.

Yup i am aware of it, even the ex tech chief from Qimonda is in CXMT and apparently he developed some new tech for CXMT.

 

[j20blackdragon]

AMD Epyc 7742 is a 64-core CPU with a 2.25GHz base clock, 3.4GHz max clock, and 225W TDP at 7nm.

Compare that to the Kunpeng 920 from Huawei.

Btw, just in case you haven't noticed. Huawei has already rolled out 64 core on 7nm. Why do you think the US government is so terrified?
View attachment 560451

64-core CPU is for high-performance computing, enterprise workloads, artificial intelligence and cloud-native apps.

But Huawei also has 4/8 core CPU for desktop.