Chinese Economics Thread

[AndrewS]

@Ultra

Let's settle this score shall we? Let's see what gets to Exa-scale computing FIRST. I will bet it will not be China. The reason why is outline in my previous post, it would take quantum leap for them to get to Exa-scale supercomputing as it will simply become too costly (100-150 million cores, 200MW power requirement) to build or scale up using the current tech which China obviously mastered. Let's see if China can indeed out innovate America in this regard shall we?

I'd be more than happy to bet money on China getting to exa-scale before the USA for the following reasons.

The US is aiming for 2023, whilst China is aiming for 2020 deployment, which is a huge 3 year gap which is an entire generation in chip design terms.

China currently has 3 different pre-exascale projects underway, to be completed by 2018.
If we look at chips, both the Shenwei chips and the Knight's Landing chips have only recently become available in the past year, and are comparable in performance/energy terms.

Also note that the European Union also has an pre-exascale prototype (based on ARM) due for completion in 2018.

===

But in the bigger picture, what we see is that China already has more raw compute power than exists in the USA. And that existing Chinese supercomputer plans will result in China in pulling ahead in this metric, as those supercomputers will be needed for R&D purposes.

Last year, Nature magazine actually forecast that China will be spending more on R&D than the USA in 2020, which is only 4 years away.

 

[SamuraiBlue]

@Ultra



I'd be more than happy to bet money on China getting to exa-scale before the USA for the following reasons.

The US is aiming for 2023, whilst China is aiming for 2020 deployment, which is a huge 3 year gap which is an entire generation in chip design terms.

China currently has 3 different pre-exascale projects underway, to be completed by 2018.
If we look at chips, both the Shenwei chips and the Knight's Landing chips have only recently become available in the past year, and are comparable in performance/energy terms.

Also note that the European Union also has an pre-exascale prototype (based on ARM) due for completion in 2018.

===

But in the bigger picture, what we see is that China already has more raw compute power than exists in the USA. And that existing Chinese supercomputer plans will result in China in pulling ahead in this metric, as those supercomputers will be needed for R&D purposes.

Last year, Nature magazine actually forecast that China will be spending more on R&D than the USA in 2020, which is only 4 years away.

There are more then one way to measure super computers and

Please, Log in or Register to view URLs content!

is one of them.
This measures how fast a computer can process big data or non linear data in which you can see K super computer is the fastest.

Please, Log in or Register to view URLs content!

Data intensive supercomputer applications are increasingly important for HPC workloads, but are ill-suited for platforms designed for 3D physics simulations. Current benchmarks and performance metrics do not provide useful information on the suitability of supercomputing systems for data intensive applications. A new set of benchmarks is needed in order to guide the design of hardware architectures and software systems intended to support such applications and to help procurements. Graph algorithms are a core part of many analytics workloads.

Backed by a

Please, Log in or Register to view URLs content!

of over 50 international HPC experts from academia, industry, and national laboratories, Graph 500 will establish a set of large-scale benchmarks for these applications. The Graph 500 steering committee is in the process of developing comprehensive benchmarks to address three application kernels: concurrent search, optimization (single source shortest path), and edge-oriented (maximal independent set). Further, we are in the process of addressing five graph-related business areas: Cybersecurity, Medical Informatics, Data Enrichment, Social Networks, and Symbolic Networks.

This is the first serious approach to complement the Top 500 with data intensive applications. Additionally, we are working with the SPEC committee to include our benchmark in their CPU benchmark suite. We also maintain a list of the most energy efficient systems for data intensive computing. The

Please, Log in or Register to view URLs content!

will be released together with the Graph500 list. We anticipate the list will rotate between ISC and SC in future years.

The Graph 500 was announced at ISC2010 and the first list appeared at SC2010.

 

[AndrewS]

Now we're getting into how do we measure compute power.

But remember that the trend is pretty clear on how China will need a lot more computing power than the USA, and that China also has the resources required to build this.

 

[manqiangrexue]

I never say "nuclear weapon or nuclear fission were innovations while improving a supercomputer is not". I say what China did with their supercomputer is not innovation because they just add more cores! And I predict they will soon hit that a wall they can't break through by just adding even more cores.

Let's settle this score shall we? Let's see what gets to Exa-scale computing FIRST. I will bet it will not be China. The reason why is outline in my previous post, it would take quantum leap for them to get to Exa-scale supercomputing as it will simply become too costly (100-150 million cores, 200MW power requirement) to build or scale up using the current tech which China obviously mastered. Let's see if China can indeed out innovate America in this regard shall we?

If they do, then I stand corrected. If not, then you obviously don't know what you talking about!

Were you not paying attention at all? They didn't "just add more cores!" They used a design that happened to have more cores, but the Chinese machine is SMALLER than the American Titan AND it is much more energy-efficient! TaihuLight is half the size of Titan! Read that and understand that you can't just take what's there, add more cores, and then have it become smaller!! What you said would only be plausible if TaihuLight is 5x faster than Titan, but also 5x larger and uses 5x the energy! But this is clearly not the case and Taihulight is a better designed machine! As a matter of fact, they say that TaihuLight's design is only beginning to reach part of its potential and can most easily be scaled up, though I doubt China wants to make a computer that's just 8x larger.

See who gets to exascale first? Sure, let's wait. US estimates they can do it by 2023. Japan by 2020. China's already got it going and estimates 2017-2020.

 

[tphuang]

There are more then one way to measure super computers and

Please, Log in or Register to view URLs content!

is one of them.
This measures how fast a computer can process big data or non linear data in which you can see K super computer is the fastest.

SamuraiBlue, as I said before, please stay away from non-military stuff, since your posts do tend to cause heated arguments.

 

[Engineer]

I never say "nuclear weapon or nuclear fission were innovations while improving a supercomputer is not". I say what China did with their supercomputer is not innovation because they just add more cores! And I predict they will soon hit that a wall they can't break through by just adding even more cores.

Let's settle this score shall we? Let's see what gets to Exa-scale computing FIRST. I will bet it will not be China. The reason why is outline in my previous post, it would take quantum leap for them to get to Exa-scale supercomputing as it will simply become too costly (100-150 million cores, 200MW power requirement) to build or scale up using the current tech which China obviously mastered.

If they do, then I stand corrected. If not, then you obviously don't know what you talking about!

It is obvious that you are so desperate in pandering a false narrative when your own statements are self contradictory.

Right in your last post you claimed taking unrealistic risks is what matters in innovation, and now you claim doing things "too costly" is not innovation. In the post before that, you claimed only being the first matters in innovation, and now you try to justify how if China achieving exa-scale computing first would not be an example of innovation.

Your posts are great at demonstrating why the US are slowing down at being innovative vis-a-versus China. Instead of being productive to society, too many people like yourself are wasting time on mental gymnastics, in desperate attempts to portray China's successes as failures.

Let's me also correct you, China didn't "just add more cores", China innovates. That is why China is currently in the lead in supercomputing over the US.

Let's see if China can indeed out innovate America in this regard shall we?

China already out-innovated America in supercomputing by getting to 125 Teraflops first. US knew of this fact long before TaihuiLight came out, which was why US Government tried to impede China's progress through banning sales of CPUs to China.

 

[Engineer]

There are more then one way to measure super computers and

Please, Log in or Register to view URLs content!

is one of them.
This measures how fast a computer can process big data or non linear data in which you can see K super computer is the fastest.

Ah yes, the same approach to Beijing Olympics in medals count: If the West is not winning, then just redefine the meaning of winning.

You know the prospect of China overtaking the West is very real when the pundits are actually going through the five stages of grief.

• Denial: China just has a stroke of luck. Its bubble will burst any days now.
• Anger: China copies everything!
• Bargaining: What China is dong is not really innovation.

I am going to enjoy seeing the next two stages when they arrive, especially the first one:

• Depression
• Acceptance

 

[AndrewS]

The way the US counts medals is off topic but it really is just how they think and do things there.

Everyone in Western Europe (and the rest of the world) thinks that the US is crazy to count the total number of medals instead of taking golds first

 

[ahojunk]

Please, Log in or Register to view URLs content!

2016-07-27 09:04 | Global Times | Editor: Li Yan

The State Council Information Office issued a document on Tuesday to guide the restructuring and merger of the country's centrally administered State-owned enterprises (SOEs), said a statement on its website.

More government capital should be channeled toward crucial industries to support their development, according to the statement.

SOEs should become more innovative by optimizing their research and development programs and strengthening basic research for eventual application.

Companies in sectors including equipment manufacturing, construction and iron and coal should merge so as to pool their resources and cut production overcapacity.

Early in February, the State Council pledged to cut crude steel production by 100-150 million tons in the next five years .

Also, it plans to shut down 500 million tons of coal capacity and consolidate another 500 million tons over the next three to five years.

Through restructuring and mergers, centrally administered SOEs should achieve clear strategic positioning and rational overall structures, and enhance their innovation capacity and international competitiveness, said the statement.

Many centrally administered SOEs have announced or completed plans to merge.

For example, China North Locomotive and Rolling Stock Industry Co and China South Locomotive and Rolling Stock Co merged into CRRC Co, said an announcement on the company's website.

 

[ahojunk]

Please, Log in or Register to view URLs content!

2016-07-29 11:30 | Ecns.cn | Editor: Mo Hong'e

(ECNS) -- The end of China's one-child policy has greatly stimulated the development of the baby care market, with everything from vitamins to photography seeing strong demand and rising prices, Guangzhou Daily reported.

Sales of pregnancy and prenatal vitamins and fertility drugs for women increased by 66.9 percent in the second quarter from last quarter on Jianke.com, a leading online drug store.

Female visitors to the website also rose by 45.3 percent, while fertility drugs for men and male visitors grew by 39 percent and 51 percent respectively.

Xie Fangmin, CEO of Jianke.com, said the two-child policy has directly fueled the growth of baby care products and means a bright future for the online pharmacy business.

Guangzhou Baiyunshan Zhongyi Pharmaceutical said a drug it made to assist fertility and protect the fetus saw sales growth of 169 percent in the first six months.

Many maternal and child healthcare hospitals in Guangzhou City, capital of South China's Guangdong Province, are already crowded with pregnant women, who might need to register to see a doctor one week earlier.

Other products and services targeting mothers and children have also seen golden opportunities following the birth policy relaxation. Guangzhou now boasts over 500 baby and mother specialty stores. In the city's major commercial street Shangxiajiu, many clothing stores have transformed themselves into baby bath services.

A shopping mall owned by Guangzhou Friendship Group has increased its area for baby specialty products by 30 percent, and also offered new playgrounds to attract consumers.

Wu Ruixi, manager of a local child photography studio, said the business will grow at least 30 percent and that the store has started training new employees and expanding its office space.

Insiders said the market for toys, clothes and other baby products will benefit hugely as more families have two children.