News on China's scientific and technological development.

BoraTas

Captain
Registered Member
Their loss
I'm somewhat confused as to how China was able to continue being competitive in supercomputing, given as that FT article itself writes:
"The local developers of the chips used in the two giant new systems — Tianjin Phytium Information Technology and Shanghai High-Performance Integrated Circuit Design Center — were both on last year’s US sanctions list.
“I think it’s quite impressive that they were able to put in place a system based on their own technology over a very short period of time,” said Dongarra. He added that it was unclear whether the chips were manufactured in mainland China — which is still years behind in matching the world’s most advanced chip fabs — or in Taiwan."

Is there any likelihood at all that the semiconductors themselves were fabbed in the mainland, or is it virtually a given that they were fabbed in Taiwan?

If they were fabbed outside of the mainland, then I would that not present a rather obvious bottleneck that the US would seek to exploit in a rather obvious manner?
Processor performance is an extremely multifactorial thing. Multi-node computer system performance is even more multifactorial since communication throughput and latency between processors and nodes become important as well.

Starting around 2005, computer performance in terms of instructions per clock per core, and clock speed has mostly stalled. Since then the main method to increase computer performance is increasing parallelism. This is achieved by more cores, more CPUs and even more nodes. Decreasing the node size used to enable more advanced architectures and higher clock speeds. Now it mostly doesn't and that means unless the difference is extreme you can make up for it with just bigger and more chips. SMIC 12 nm and N+1 (roughly 8 nm equivalent) are more than enough for competing with computers using TSMC's 5 nm node if the Chinese government is willing to pay the price.
1652999026577.png
 

4Runner

Junior Member
Registered Member
I'm somewhat confused as to how China was able to continue being competitive in supercomputing, given as that FT article itself writes:
"The local developers of the chips used in the two giant new systems — Tianjin Phytium Information Technology and Shanghai High-Performance Integrated Circuit Design Center — were both on last year’s US sanctions list.
“I think it’s quite impressive that they were able to put in place a system based on their own technology over a very short period of time,” said Dongarra. He added that it was unclear whether the chips were manufactured in mainland China — which is still years behind in matching the world’s most advanced chip fabs — or in Taiwan."

Is there any likelihood at all that the semiconductors themselves were fabbed in the mainland, or is it virtually a given that they were fabbed in Taiwan?

If they were fabbed outside of the mainland, then I would that not present a rather obvious bottleneck that the US would seek to exploit in a rather obvious manner?
The secret sauce of HPC in general and supercomputing in particular is not dependent only on fast and dense CPUs. It also include but not limit to critical factors of high-performance inter-connect and scheduling algorithms and intelligent storage. So even some CPUs or GPUs got hit by the sanctions, China is able to use its own microprocessors instead of Intel/Nvidia/AMD CPUs or GPUs. In some sense, high-performance inter-connect and software is more important than raw compute power.

Disclaimer: I am not an expert in supercomputers. But I have some experiences in HPC and high-performance NFs.
 

FairAndUnbiased

Brigadier
Registered Member
I'm somewhat confused as to how China was able to continue being competitive in supercomputing, given as that FT article itself writes:
"The local developers of the chips used in the two giant new systems — Tianjin Phytium Information Technology and Shanghai High-Performance Integrated Circuit Design Center — were both on last year’s US sanctions list.
“I think it’s quite impressive that they were able to put in place a system based on their own technology over a very short period of time,” said Dongarra. He added that it was unclear whether the chips were manufactured in mainland China — which is still years behind in matching the world’s most advanced chip fabs — or in Taiwan."

Is there any likelihood at all that the semiconductors themselves were fabbed in the mainland, or is it virtually a given that they were fabbed in Taiwan?

If they were fabbed outside of the mainland, then I would that not present a rather obvious bottleneck that the US would seek to exploit in a rather obvious manner?
Supercomputer system performance doesn't necessarily correlate to the single processor performance because the bottleneck isn't necessarily only in how fast 1 processor works but how to get 10000 processors to work together, how to talk to memory chips, etc.

And much of the improvement in <14 nm is from power savings, not maximum performance.
 

tokenanalyst

Brigadier
Registered Member
I'm somewhat confused as to how China was able to continue being competitive in supercomputing, given as that FT article itself writes:
"The local developers of the chips used in the two giant new systems — Tianjin Phytium Information Technology and Shanghai High-Performance Integrated Circuit Design Center — were both on last year’s US sanctions list.
“I think it’s quite impressive that they were able to put in place a system based on their own technology over a very short period of time,” said Dongarra. He added that it was unclear whether the chips were manufactured in mainland China — which is still years behind in matching the world’s most advanced chip fabs — or in Taiwan."

Is there any likelihood at all that the semiconductors themselves were fabbed in the mainland, or is it virtually a given that they were fabbed in Taiwan?

If they were fabbed outside of the mainland, then I would that not present a rather obvious bottleneck that the US would seek to exploit in a rather obvious manner?

HOW CHINA MADE AN EXASCALE SUPERCOMPUTER OUT OF OLD 14 NANOMETER TECH​

Please, Log in or Register to view URLs content!

Because when power consumption is not an issue, is not enterally necessary to use latest shiniest node to get computer power, can get more computing power out of relative "older nodes" by using advanced packaging techniques.

Please, Log in or Register to view URLs content!

Or using a completely different chip architecture that better uses vector units and matrix coprocessors to speed things up instead of using generic manycore CPUs. Better software, better interconnections, etc.
And yes, the chips for these supercomputers in China could have been made at SMIC or HLMC.
 

Blitzo

Lieutenant General
Staff member
Super Moderator
Registered Member
Sunway Oceanlight uses 14nm chip made by SMIC. Here is the detailed calculation of how that is possible.

Please, Log in or Register to view URLs content!

It may not be as green as the US machines, but it is much cheaper than US to build, not because it is China building it but 14nm tech is cheaper than denser chips.

I think 14nm is mature tech in SMIC's hand according to the semiconductor thread in this forum.

Thanks, so the article does show that it is indeed 14nm processes -- but are we sure they are fabbed at SMIC?
Reading the article itself and some of the past articles, I am unable to find a smoking gun that the SW26010-Pro are actually fabbed by smic, but I'd be happy to be corrected.

I do appreciate that leading edge processes are somewhat less important for supercomputing needs and competitiveness. But in the longer term, if the gap in leading edge processes is unable to be closed, then if everything else is held constant, then a degree of bottleneck will still remain.


Edit:
I presume the SW26010-Pro are assumed to be from SMIC given the original SW26010 are *confirmed* to be fabbed by SMIC?
 
Last edited:

SanWenYu

Captain
Registered Member
A few days ago, we have seen this work by a Chinese team. They designed and built devices that generate electricity from the ambient relative humidity changes.

Now we have another Chinese team published a paper about their work that "opens new possibilities for continuously efficient energy harvesting utilizing the sun and the universe".

I remember having seen news about this paper a couple of weeks ago. But I cannot recall whether I have posted it here. The paper was published on April 22. The Chinese MSM picked up it just now.

Please, Log in or Register to view URLs content!

Self-adaptive integration of photothermal and radiative cooling for continuous energy harvesting from the sun and outer space​

The sun (∼6,000 K) and outer space (∼3 K) are two natural energy resources for humans. However, most of the approaches of energy harvesting from the sun and rejecting energy to outer space are achieved independently using absorbers and emitters with static spectral properties. Herein, a spectrally self-adaptive structure with strong solar absorption and switchable emissivity within the atmospheric window (i.e., 8 to 13 μm) is experimentally demonstrated to achieve diurnal solar thermal and nocturnal radiative cooling efficiently. The experiment shows that the proposed structure not only can be heated to 185°C in diurnal mode but also be cooled to −12°C in nocturnal mode. This work opens new possibilities for continuously efficient energy harvesting utilizing the sun and the universe.
The sun (∼6,000 K) and outer space (∼3 K) are two significant renewable thermodynamic resources for human beings on Earth. The solar thermal conversion by photothermal (PT) and harvesting the coldness of outer space by radiative cooling (RC) have already attracted tremendous interest. However, most of the PT and RC approaches are static and monofunctional, which can only provide heating or cooling respectively under sunlight or darkness. Herein, a spectrally self-adaptive absorber/emitter (SSA/E) with strong solar absorption and switchable emissivity within the atmospheric window (i.e., 8 to 13 μm) was developed for the dynamic combination of PT and RC, corresponding to continuously efficient energy harvesting from the sun and rejecting energy to the universe. The as-fabricated SSA/E not only can be heated to ∼170 °C above ambient temperature under sunshine but also be cooled to 20 °C below ambient temperature, and thermal modeling captures the high energy harvesting efficiency of the SSA/E, enabling new technological capabilities.

News report in Chinese by Xinhua:
Please, Log in or Register to view URLs content!


新华社合肥5月19日电(记者徐海涛、周畅)白天可比环境温度高170摄氏度,夜晚可比环境温度低20摄氏度,无需外部能源消耗……近期,中国科学技术大学教授裴刚、研究员邹崇文等人研制出一种分别以太阳、太空为热源、冷源的“冷热双吸”材料,可24小时捕获利用能量,有望在改善地球温室效应、供应太空基地能源等方面发挥作用。
 
Last edited:

ZeEa5KPul

Colonel
Registered Member
I'm somewhat confused as to how China was able to continue being competitive in supercomputing, given as that FT article itself writes:
"The local developers of the chips used in the two giant new systems — Tianjin Phytium Information Technology and Shanghai High-Performance Integrated Circuit Design Center — were both on last year’s US sanctions list.
“I think it’s quite impressive that they were able to put in place a system based on their own technology over a very short period of time,” said Dongarra. He added that it was unclear whether the chips were manufactured in mainland China — which is still years behind in matching the world’s most advanced chip fabs — or in Taiwan."

Is there any likelihood at all that the semiconductors themselves were fabbed in the mainland, or is it virtually a given that they were fabbed in Taiwan?

If they were fabbed outside of the mainland, then I would that not present a rather obvious bottleneck that the US would seek to exploit in a rather obvious manner?
Please, Log in or Register to view URLs content!
 

Strangelove

Colonel
Registered Member
Please, Log in or Register to view URLs content!

China delivers world's first 100,000-tonne 'mobile fish farm'

Please, Log in or Register to view URLs content!


ac400c11badb48c9b54222650f198976.jpeg

Colored smoke is released at the delivery ceremony for the Guoxin-1, the world's first 100,000-tonne aquaculture vessel, Qingdao, east China's Shandong Province, May 20, 2022. /Qingdao Conson Development Group

In a world first, China delivered a 100,000-tonne smart aquaculture vessel on Friday in Qingdao City, east China's Shandong Province.
The ship, named Guoxin-1, is the realization of an innovative mobile fish farming model regarded as a technological breakthrough for China's deep-sea aquaculture industry.

The fish farm vessel is 249.9 meters long, with a deadweight of 100,000 tonnes and a displacement of 130,000 tonnes, large enough to survive typhoons.

The cabin has 15 rooms, each larger than two standard swimming pools.

According to CMG, the ship can handle 3,700 tonnes of fish annually, including yellow croakers, groupers and Atlantic salmons.

The key technology is the non-stop water exchange between the cabin and the sea that makes the water environment on the ship enclosed and controllable, said Dong Shaoguang, deputy general manager of the ship's state-owned funder, Qingdao Conson Development Group.

Dong told the media that the Guoxin-1's breeding density is three to five times higher than traditional cages, and the aquaculture cycle can be up to a quarter shorter.

The ship will farm in the Yellow Sea, East China Sea and South China Sea, with the first batch of yellow croakers raised expected to be on the market by the fall of 2022.

The Guoxin-2 and Guoxin-3 ships of the same model are due for delivery by March 2024. The construction of an improved version, the Guoxin-4, is planned to begin in late 2023.
 
Top