On March 15th, it was announced that the "Qingyang Telecom" has completed the deployment of 1000 GPU servers including NVIDIA H800 and Huawei H910B by the end of February at the Qingyang Computing Center of China Telecom's East-West Calculation National Hub. The computing power of the center has reached 3500P.
With the stable operation of these 1000 GPU servers, the computing power of the Gansu hub node cluster at the national integrated computing network has exceeded 5000P, reaching 5300P.
According to media reports, 'P' represents the computing power of the supercomputing center. In the industry, it is stated that 1P is equivalent to computing ten quadrillion times per second, and 100P of computing power is equivalent to 500,000 computers.
Taking a scientific research scenario as an example: when astronomers need to locate certain features in a star map of 200,000 celestial bodies, with insufficient computing power, it might take over 100 days. However, with 100P of computing power, it would only take 100 seconds.
Currently, the Qingyang data center cluster has established network links from Xi'an to key cities such as Beijing, Shanghai, Guangzhou, and Chengdu, with 15,000 cabinets deployed and running. It has formed a computing power scale of 5300P, with intelligent computing power absorption reaching 100%.
It is worth mentioning that the official statement indicates that the Qingyang Computing Center of China Telecom plans to deploy an additional 400-500 GPU servers by the end of June, adding 2500P of computing power.
