News on China's scientific and technological development.

FairAndUnbiased

Brigadier
Registered Member
I like the "NOT to submit to Top500". It is a move to discredit or downgrade the list. Instead, people in the world are forced to listen to two voices and make the choice of who to rely on. Build my own rival stage of show. In a way, up to today, people is like "this is that because somebody says so", in the future more question would follow "are you sure? because I heard something else."
This is an extremely important move. We always have to evaluate, when someone says something is "good", that there are many questions unanswered by these assertions:

Good? By what metric?
Why was that metric chosen?
Who participated in the comparison?
Who judged?
What is the end result of the comparison?
 

Jiang ZeminFanboy

Senior Member
Registered Member
Big HPC news. China has already broken the (true/LINPACK) exascale barrier in 2021—on more than one machine.
It seems they decided NOT to submit to Top500 due to political issues
Please, Log in or Register to view URLs content!
From the article:
And here’s another subtle detail. Our source confirms these LINPACK results for both of China’s exascale systems—the first in the world—were achieved in March 2021. When did the entity list appear citing Phytium and Sunway and the centers that host their showboat systems? In April 2021.
Wonder how it may affect the next supercomputers which will follow? 2022/2023 could be hard. Do we know at what nm process these processors are made? 28nm line should be ready, now only ramp up with capacity. 2022/23 we may see a 14nm and 7nm Chinese domestic line and then ramp up with capacity.
 

taxiya

Brigadier
Registered Member
From the article:

Wonder how it may affect the next supercomputers which will follow? 2022/2023 could be hard. Do we know at what nm process these processors are made? 28nm line should be ready, now only ramp up with capacity. 2022/23 we may see a 14nm and 7nm Chinese domestic line and then ramp up with capacity.
Please, Log in or Register to view URLs content!

Surprisingly, in case of Sunway, this article predicted 14nm for the coming exaflops, 8nm for the next 10 exaflops. SMIC has been mass producing 14nm.
So, if you were building the Sunway TaihuLight supercomputer, how would you scale it from 125 petaflops peak double precision performance up to at least 1 exaflops? There are a bunch of ways to do it, but according to this paper, NRCPC took some obvious approaches enabled in part by the chip process manufacturing shrink that was no doubt used from Semiconductor Manufacturing International Corp, the big indigenous foundry in China that is trying to get on par with Taiwan Semiconductor Manufacturing Corp, Samsung Electronics, and Intel with its chip foundries. SMIC’s 14 nanometer FinFET processes were in volume production in 2019, so the SW26010 is using a much older process, and
Please, Log in or Register to view URLs content!
is said to be equivalent to Samsung’s 8 nanometer process and better than TSMC’s 10 nanometer. This N+1 process is a little young to be used in a chip that might go into production this year. So assuming that the SW26010 was made with 28 nanometer processes, we think that the future Sunway exascale processor – let’s call it the SW52020 for reasons that will be obvious in a minute – will have the ability to shrink the processor die size quite a bit, keep clocks the same or lower, and significantly boost the compute elements on the chip. And by sticking with 14 nanometer, NRCPC has the 8 nanometer shrink — or smaller — to try to push to 10 exaflops with a future machine.
The 3rd exaflops (not in the reports and about two months behind the two reported), Shuguang (Sugon) intended to use 7nm for their production exaflops, the prototype (2018) was not 7nm.
 
Last edited:

AssassinsMace

Lieutenant General
It shows that demands for China to be transparent is all so the US knows which industries they should target to undermine China. Yeah, that Top 500 list is now irrelevant because now it's just going to be a propaganda tool touting Western computing achievements in the guise of being international. What happened to Japan? I remember reading article a while ago saying Japan was back in the competition for world's fastest computer and now there's nothing.
 

gelgoog

Brigadier
Registered Member
It shows that demands for China to be transparent is all so the US knows which industries they should target to undermine China. Yeah, that Top 500 list is now irrelevant because now it's just going to be a propaganda tool touting Western computing achievements in the guise of being international. What happened to Japan? I remember reading article a while ago saying Japan was back in the competition for world's fastest computer and now there's nothing.

Japan's Fugaku supercomputer is the #1 in the current Top500 list.
Please, Log in or Register to view URLs content!

It uses 48x ARM cores per processor at 7nm with HBM2 memory. The A64FX processor was designed by Fujitsu and has its own special 512-bit vector instructions to increase peak FLOPS. It is fabbed at TSMC.
 
Last edited:

AssassinsMace

Lieutenant General
Please, Log in or Register to view URLs content!

Yes China setting up smart cities is a threat to the world when it's really about stopping Chinese advancements in technology. The US has a hard time getting 4G to cover where people live. The US ain't having smart cities anytime in the foreseeable future. They know that hence why they're very sensitive to anyone having something they don't have.
 

Strangelove

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

Chinese tech giant Baidu Inc will cultivate 5 million artificial intelligence talents in the next 5 years, aiming to create a well-rounded tech labor force and power AI growth, said Robin Li, co-founder and CEO.

Li made the remarks at the Emerging Engineering International Forum 2021, hosted by his alma mater Peking University, on Saturday. "In daily work, it is very important to shift our mindset from a scientific one to an engineering one – going from a one-step process to a step-by-step process," Li said.

While a scientific approach can produce fruitful and thorough results, an engineering approach is incremental, strategic and ultimately more useful, he added.

According to Li, both Baidu and Tesla are adopting a step-by-step engineering approach to autonomous driving. While Tesla is gradually advancing the autonomous driving levels, Baidu prioritizes expanding the driving range of self-driving vehicles.

By practicing the operation of self-driving cars in smaller areas where the conditions are more conducive to testing, Baidu said they are able to gradually learn and improve to expand the geographical scope of autonomous driving, while gradually upgrading the abilities of the autonomous cars.

Since 2015, the Beijing-based company has supported the Ministry of Education's industry-university cooperation and collaborative education project, training more than 3,000 teachers across more than 700 universities.

Li said in today's world, it is expected that job applicants have an understanding of artificial intelligence to find a job in the technology industry. As such, he has committed the company to develop 5 million AI talents over five years to help job seekers grow their skills toolbox, creating more well-rounded talent.
 
Top