News on China's scientific and technological development.

xlitter

Junior Member
Registered Member
Development of LiDAR in China.
Please, Log in or Register to view URLs content!

At present, China has gathered a number of lidar enterprises with a certain market scale. Data disclosed by Yole show that since 2018, there are about 55 ADAS preloading points in the world, among which the lidar companies from China account for half. Hesai Technology overtook Valeo with 27% to rank the first in the world, and Astral Juhuang ranked the second in China with 16% of the number and the third in the world.

The agency estimates that more than 200,000 LiDAR will be delivered to vehicles in 2022, with the top five companies accounting for 84 percent of the global total. Three of them will come from China, followed by hesai Technology, Sagitar Juchuang and Huawei.
 

tphuang

Lieutenant General
Staff member
Super Moderator
VIP Professional
Registered Member
Development of LiDAR in China.
Please, Log in or Register to view URLs content!

At present, China has gathered a number of lidar enterprises with a certain market scale. Data disclosed by Yole show that since 2018, there are about 55 ADAS preloading points in the world, among which the lidar companies from China account for half. Hesai Technology overtook Valeo with 27% to rank the first in the world, and Astral Juhuang ranked the second in China with 16% of the number and the third in the world.

The agency estimates that more than 200,000 LiDAR will be delivered to vehicles in 2022, with the top five companies accounting for 84 percent of the global total. Three of them will come from China, followed by hesai Technology, Sagitar Juchuang and Huawei.
hmm, Huawei is not big here at all. Most of the new brands are picking Hesai. BYD has picked Robosense, which means Robosense will probably be the largest one in a few years.
https://www.businesswire.com/news/home/20211227005067/en/RoboSense-Receives-BYD’s-Strategic-Investment-and-Becomes-an-Engaged-Participant-of-the-Auto-Industry
 

sunnymaxi

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

(Yicai Global) Sept. 9 -- China has entered the top tier of artificial intelligence capabilities for the first time and the gap with the US is narrowing, according to a recent research report.

China has been inducted into the first class of the ‘Artificial Intelligence Index Report 2021’ due to continuous improvements in the comprehensive strength of the country’s AI technology, said the head of the research team at the Institute of Scientific and Technical Information of China. In the previous two years that the report was compiled, only the US made this class.

Although the US still comes top with a score of 59.42, China ranks second for the second year running with a score of 50.14, and the gap between the two is closing, Xu Feng, who is also executive deputy director of ISTIC’s Artificial Intelligence Development Research Center, told Yicai Global.

South Korea placed the highest in the second tier, with a score of 37.29, and Israel was at the bottom of this group with 26.37. Altogether, of the 46 countries evaluated, there were nine countries in the second class, 13 in the third group, including Sweden and Luxembourg, and 22 in the fourth class, including India and Russia.

The research shows the quantity and quality of China’s AI enterprises are growing rapidly, said Zhao Zhiyun, director of ISTIC. The number of Chinese AI open-source projects climbed to 158 last year from 139 the year before.

China now has the most supercomputer centers in the world, Zhao said. As of June 2021, it had 188 supercomputing centers in the global Top 500, accounting for 37.6 percent.

“The number of supercomputing centers in China has surpassed that of the US, but there is still a gap compared with the US in terms of computing power, mainly because the number of high-quality data centers needs to increase,” Xu said.

The 10 countries with the most AI firms account for 82.9 percent of the total number of firms among the countries polled, at 9,020, the report said. The US accounts for half of this with 4,664 and China has 880. In terms of listed AI firms, the US has the most at 63, followed by China at 60 and Canada with 12.

China’s AI companies had received total funding of USD46.2 billion as of September last year, ranking second in the survey. But the average financing amount per enterprise was USD53 million, ranking first.

Talent Pool

North America and Europe remain the main clusters of AI expertise. And the top five countries in the AI talent subindex were the US, Singapore, China, Latvia, and Israel.

China has made great efforts to nurture AI expertise in recent years, Xu said. And although the talent pool is increasing, because it has not been a long time, there is still a gap in the overall quality of skilled workers with developed countries such as the US, especially in terms of the number of top AI scholars.

Chinese universities’ offerings in basic AI disciplines are still inferior to those of the US, the report said. Although the number of professionals in the industry is second only to the US, the proportion per 10,000 people in the labor force is not high.
 

sunnymaxi

Major
Registered Member
Who made this score? Is it like the COVID preparedness score the US made ~2 years ago?
who made this score ? probably America or some Ai related competitions held in western world. i also don't believe on this rankings

China is the global leader in 4th industrial revolution. number of Ai related firms are in thousands. but in this article they said 880 only. LOOL .. some mistakes but fine.
 

tphuang

Lieutenant General
Staff member
Super Moderator
VIP Professional
Registered Member
Actually, I think China probably is leading the world in AI by this point. It really depends on who you ask.

Anyhow, something that got missed in the entire Biren/Inspur test result was what Alibaba did with their submission.
Please, Log in or Register to view URLs content!
Alibaba submitted five systems that are composed of variations of two to six nodes, running a mix of Intel Xeon and Nvidia GPUs. An Alibaba software program, called Sinian vODLA, automatically partitions the tasks of a neural network across different processors among the different computers.

The most striking feature is that the Sinian software can decide on the fly to apportion tasks of a neural net to different kinds of processors, including a variety of Nvidia GPUs, not just one, so that the difference in abilty of each processor is not an obstacle but a potential advantage.
The benchmark results from the Alibaba Cloud Server showed some eye-popping numbers. On the BERT language task, a four-node system with 32 Nvidia GPUs in total was able to run over 90,000 queries per second. That is 27% faster than the top winning submission in the Closed category of data center machines, a single machine from Inspur using 24 GPUs.
You can always stack together more GPUs to get more computation, but this multi-node approach really seems to bring out the most optimal approach. It should that you don't need the latest GPUs to carry out high number of AI computation. Now, if they can only onboard more domestic chips to their platform.

You will also see how involved Inspur is everywhere. They are huge.
Please, Log in or Register to view URLs content!
They seem to be between 10 and 11.5% in market share globally for the past few quarters.
Global server shipments were 3.305 million, an increase of 20.7% year-over-year (YOY). Sales were $27.21 billion, an increase of 29.3% YOY. Worldwide, all regions recorded YOY growth in revenue during the first quarter. The Americas saw YOY growth of 32.2% and Asia-Pacific (excluding Japan) saw 34% growth. Investment in hyperscale data centers in these regions remained high. Revenue in Japan and EMEA (Europe, Middle East, and Africa) increased 13.4% and 21.4% YOY, respectively. This was a result of a backlog of orders being filled and strong demand for cloud products.
So, it's absolutely humongous for Inspur to be work with Biren Technology and Kunlun. They will be able to export AI servers with Chinese GPUs.

Note that they have already helped building Cambrian AI servers for Nanjing AI data center, so this not anything new
在智算中心领域,浪潮与寒武纪一起建设了南京智算中心,实现了国产化的算力系统,服务于各类企事业单位

Inspur also has it's own AI ecosystem called 元脑生态
Please, Log in or Register to view URLs content!
Back when Biren signed up with Inspur, they talked a lot about the pluses of this ecosystem for Biren GPUs.
Please, Log in or Register to view URLs content!
this is when Kunlun signed up with Inspur and this ecosystem
Please, Log in or Register to view URLs content!

GPUs themselves is just one part of the equation. You can always put in more GPUs and use more space/power to get more computation through. The bigger difference makers in AI are the ecosystem, the data and technology that are used to perform the AI work. Without big players like Inspur, Alibaba and Baidu, it would be hard to see Chinese chip design house succeed. With their help, I think the task gets a lot easier. With their help + the ban of Nvidia products, Biren and others should be able to take large market share.
 
Top