Shulou(Shulou.com)11/24 Report--

Since the beginning of this year, the large model industry has shown a blowout development. At present, more than 100 large models have been released publicly in China, which accelerates the scarcity of intelligent computing power, and the situation of "one card is difficult to get" and "queue reservation" appears in GPU.

How does the industry respond to the outbreak of demand for intelligent computing and the resulting surge in energy consumption? At the Green Computing Forum in the Cloud original era of the 2023 inclusion Bund Conference recently, guests from the fields of politics, industry, learning, research and application shared the development direction under the new trend.

At present, the domestic policy attaches great importance to the green development of the computing industry. He Feng, deputy director of the Shanghai Communications Administration, proposed that Shanghai carried out green computing related work around the "Computing Pujiang Action Plan", and proposed to speed up the research and development of key technologies of green computing to improve the resource utilization of data centers; speed up the standardized measurement and definition of data centers; deepen industrial coordination and build a new ecology of green computing.

(photo: he Feng, Deputy Director of Shanghai Communications Administration, delivered a speech)

Xie Ji, secretary-general of the China small and medium-sized Enterprises Association and former inspector of the Climate Department of the National Development and Reform Commission, said that the China small and medium-sized Enterprises Association attaches great importance to the development of small and medium-sized enterprises in a green and low-carbon economy. The Association, together with relevant units in Shanghai, jointly issued the group standard of "ESG Information Disclosure Guide for small and medium-sized Enterprises and ESG Star rating for small and medium-sized Enterprises" to promote the green and low-carbon transformation of small and medium-sized enterprises, practice "green computing" and achieve "carbon neutralization" in the computing field.

Ant Group proposed to pay attention to the "end-to-end green computing system", especially the green computing in the computing application stage, which led to break through the bottleneck of computing and achieve green collaborative development, which was highly recognized by the participants.

End-to-end green computing, that is, from power production, computing production (including intelligent computing center builders, hardware manufacturers, cloud manufacturers) to computing application of green computing in the whole industry. In the Green Computing White Paper for Computing applications issued by the China Academy of Communications in 2023, it was mentioned that green computing in computing applications has the largest space to improve intelligent computing efficiency and achieves the fastest results. It is also a part that has not been given enough attention and still needs to be strengthened.

At the level of computing hardware infrastructure, the industry has achieved remarkable results through measures such as architecture upgrading and reducing data center energy consumption. Duan Xiaodong, deputy dean of China Mobile Research Institute, said that China Mobile has built a new intelligent computing center from the five core aspects of "new interconnection, new computing effect, new storage, new platform and new energy saving". It has built a unified standard, advanced technology, compatible and open base for intelligent computing. "in the face of the explosion of intelligent computing, it is necessary to build a computing network through openness, collaboration, inclusive benefits and win-win results. We will promote the construction of a better green and energy-saving computing foundation in the future intelligent computing era.

Using AI technology to achieve energy-saving optimization of data centers and related industries is also an important practice direction. Liu Yunxin, professor Guoqiang of Tsinghua University and chief researcher of the Intelligent Industry Research Institute, shared the latest results with Zhang Yimin, chief scientist of Intel Research Institute in China and director of the academic Cooperation Center, including core enabling technologies in two directions: energy-efficient intelligent computing systems and the use of AI to improve quality and efficiency, and then build a "double carbon" intelligent computing platform, taking advantage of the greening of the data-driven intelligent decision-making enabling industry.

Statistics estimate that after the data is produced, only 0.2% of the data can be effectively retained, used and then brought into play. A large amount of data can not effectively generate business value, and its bottleneck lies in the current data computing and processing capacity. In this regard, Professor of Huazhong University of Science and Technology, Vice President of Chinese computer Society and IEEE Fellow Jin Hai put forward three modes of near data processing from the point of view of new data technology: storage up-moving, computing power sinking, and distributed expansion, which put forward an effective solution to solve the performance bottleneck of "storage wall". Looking to the future, Jin Hai believes that "the data-centric computing architecture will become the trend direction of innovating big data's processing model, which will also greatly optimize the computing efficiency in data processing."

(photo: professor of Huazhong University of Science and Technology, Vice Chairman of Chinese computer Society, IEEE Fellow Jin Hai delivered a keynote speech)

In the era of big models, technology companies are at the forefront of a shortage of computing power. He Zhengyu, Chief Technology Officer of Ant Group and president of platform Technology Group, said that since the era of CPU computing, Ant Group has begun to build a green and efficient intelligent computing system, forming four core competencies: low-carbon computing power selection, global resource scheduling optimization, workload optimization, and green monitoring and evaluation, which will provide "weights and measures" for green computing in the intelligent computing era. He Zhengyu said that in 2022, this system supported the intelligent scheduling of computing power recommended by Alipay's home page of 3 billion PV per day, saving 41.4 percent of machine resources throughout the year. "in the era of large AI models, technology sinks, new optimization space is opened, and targeted optimization of the underlying software stack should be paid attention to."

(photo: he Zhengyu, Chief Technology Officer of Ant Group and President of platform Technology Group, delivered a keynote speech.)

End-to-end green computing requires industry upstream and downstream cooperation to build a green computing ecology. Ren Xiang, director of the Integrated Circuit Evaluation Center of the China Institute of Electronic Technology Standardization and head of the standard test group of the Green Computing Industry Alliance, said that pooling resources upstream and downstream of the industry chain through the green computing industry alliance can effectively promote standard innovation, personnel training, international cooperation and the landing of technological product research and development and applications.

Paying attention to the problem of carbon perception and constructing the way of green measurement and evaluation is also the first step in the systematization of industrial development. Wang Chaoyang, General Manager of Ariyun IDC R & D Department, Zhao Shuai, General Manager of Tide Information Server Product Line, and Zhu Youhong, Director of PICC Technology data Center Operation, shared different views from the perspectives of green data center, hardware and data processing, and constructively proposed the construction of correlation indicators between computing power and carbon emissions, providing industry reference for effective measurement of green computing practice.

In the Bund Conference, the value of computing power as the base of digitization and deep intelligence has been recognized. Under the general trend that computing resources are in short supply, it has become an industry consensus to drive intelligent computing power to break through the bottleneck through the cooperation of software and hardware.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.