On June 27, according to Alibaba Cloud official micro, the company and Fudan University jointly built the largest cloud scientific research intelligence platform CFFF (Computing for the Future at Fudan) officially launched.。It is understood that the platform will provide ultra-kilocalorie parallel intelligent computing in the form of a public cloud model, while supporting large model training with up to 100 billion parameters.。

"Che Wen" No.1 + "Jin Si" No.1 = CFFF platform left and right brain

It is understood that the framework for building the CFFF platform is divided into two parts: intellectual support and computational support.。Among them, the "cut question" responsible for intellectual support 1 mainly provides the intelligent computing foundation for multidisciplinary integration and innovation, while the "near thought" 1 responsible for computing power support is equipped with a dedicated high-performance computing cluster for high-precision research.。

According to the relevant introduction, "Chequan" 1 is hosted in the Aliyun Wulanchabu data center 1500 kilometers away from Fudan University, while "Jinsi" 1 is deployed in Fudan University.。The two are equivalent to the left and right brains of the CFFF platform, complement each other, and with the help of Alibaba Cloud's large-scale heterogeneous computing power fusion scheduling technology, hierarchical storage technology, AI and big data integration technology, the combination of two into one, so that the CFFF platform in the true sense of the "super cloud scientific research intelligence platform."。

According to Professor Haoqing of Fudan University and President of the Institute of Artificial Intelligence Innovation and Industry, Qi Yuan, relying on the public cloud model, projects running on the CFFF platform can enjoy super-kilocalories of parallel intelligent computing power, kilocalories of parallel effective computing power to reach 92% of the industry's leading, scalability to reach 10,000 cards, 10,000 cards of parallel effective computing power can also reach 90%。

At the same time, the public cloud is also more low-carbon and green。Alibaba Cloud Green Data Center technology combines the natural climate advantages of Wulanchabu. The CFFF platform can achieve an average annual PUE of less than 1.2. It can save more than 2,000 kilowatts of total electricity each year, save 5 million yuan in electricity bills, and save 15 tons of carbon each year。

"Born to discover and solve complex scientific problems"

After the launch of the CFFF platform, Fudan University revealed on its official website the vision of the platform - "born to discover and solve complex scientific problems."。It is reported that the CFFF platform has now landed the first large scientific model based on the platform training results, can be completed in one day 4.5 billion participants of the large model, the prediction speed from the original hour level to 3 seconds.。

Fudan also revealed on the official micro that it has signed strategic cooperation agreements with Alibaba and China Telecom respectively, and will establish a new pattern of long-term, close and stable school-enterprise cooperation.。In the future, the CFFF platform will continue to expand the scale of computing power and gradually open up to scientific research institutions, hospitals and high-tech enterprises outside Fudan University.。

For the launch of the platform, Alibaba Cloud and Fudan University have praised and placed high hopes on it.。Wang Jian, academician of the Chinese Academy of Engineering and founder of Alibaba Cloud, believes that this is the first time in China and the world that universities have the same research and computing platform as technology giants, allowing research to truly enter the era of computing-driven。

Jin Li, president of Fudan University and academician of the Chinese Academy of Sciences, said that the intellectual computing platform represented by the CFFF platform, as an emerging scientific research supercomputing architecture, will become an important supporting force for scientific research, greatly improving scientific research efficiency, reducing scientific research costs, accelerating the discovery of scientific principles and technological breakthroughs, and effectively promoting the landing of large scientific models.。