Computing power serves as a new quality productive force in the digital economy era. Hebei has vigorously promoted the construction of the Hebei Computing Power Monitoring and Scheduling Platform, actively integrating into the national integrated computing power network.
As of now, the platform has supplied over 3,680P of computing resources in total, providing computing support for more than 6,000 enterprises and youth entrepreneurship teams in the Beijing-Tianjin-Hebei region, covering multiple scenarios such as urban operations and governance, industrial intelligence, and AI model training and inference, effectively reducing enterprise research and innovation costs and computing usage expenses.
"Look, this is the Hebei Computing Power Monitoring and Scheduling Platform we built," said Xing Yanbo, General Manager of Langfang Shucheng Technology Group Co., Ltd., pointing at the large screen in the command center of the Hebei Computing Power Monitoring and Scheduling Platform located at the Beijing-Tianjin-Hebei Big Data Innovation and Application Center in Langfang on September 29.
Through the screen, real-time dynamic data of computing resources across the province scrolls continuously, with monitoring information from data centers across the region clearly presented across five dimensions: computing investment, computing supply, computing applications, computing-electricity coordination, and computing-network security.
The Hebei Computing Power Monitoring and Scheduling Platform is an important vehicle for Hebei to build the Beijing-Tianjin-Hebei integrated computing power network. A computing power network connects servers and data centers distributed in different locations to form a shared computing power pool. Entities with computing needs can access and pay for computing power on demand, just like using water and electricity.
Divided by function, the Hebei Computing Power Monitoring and Scheduling Platform is subdivided into three sub-platforms: computing monitoring, computing operations, and computing scheduling, forming a complete intelligent system.
Clicking on the screen to enter the computing monitoring sub-platform, data from data centers across the province is displayed on a single screen, covering 112 indicators including project investment, computing capacity, and storage capacity. The computing monitoring sub-platform establishes a full-dimensional computing monitoring system covering the entire province by connecting interfaces with data centers, achieving dynamic perception of computing resources within the province. The platform also continuously pushes data to the national integrated computing power network monitoring and scheduling trial verification platform, actively participating in the verification of national computing monitoring interfaces and capability assessment standards, striving to achieve more results in the field of automatic computing data collection.
On the interface of the computing operations and scheduling sub-platform, information such as total schedulable computing power, completed orders, computing unit prices, and supply-demand matching progress is updated in real time.
"In the digital economy era, computing power injects strong momentum into all industries," said Xing Yanbo. In the past, enterprises had to contact data centers one by one to find computing power. Opaque pricing and information asymmetry led to weak alignment between computing supply and demand. Now, the computing operations and scheduling sub-platform coordinates and integrates scattered computing resources across the province, using intelligent scheduling algorithms to flexibly allocate computing power, providing cost-effective and elastically scalable computing services for high-frequency demand scenarios such as AI training and scientific research computing.
"With this platform, we can purchase computing power on demand, which has reduced our large model inference costs," said Qiao Jialin, Chief Technology Officer of Tianmou Technology (Beijing) Co., Ltd., recently. The company successfully completed a computing power transaction through the Hebei Computing Power Monitoring and Scheduling Platform.
Tianmou Technology is a company developing time-series database software and time-series large models. Recently, the company needed substantial computing power to develop its TimechoAI time-series large model cloud platform and submitted a computing power purchase request to the Hebei Computing Power Monitoring and Scheduling Platform. Platform staff quickly matched high-quality computing power to ensure the smooth progress of large model inference tasks. Currently, the company's large model inference computing cost has decreased by an average of 50%.
Building the Hebei Computing Power Monitoring and Scheduling Platform is like assigning a "steward" to the province's computing power, allowing computing resources dormant in server rooms to "flow," transforming from hardware assets into tradeable and serviceable digital goods, and promoting efficient connection between computing resources and market demand.
In March last year, Langfang Shucheng Technology Group Co., Ltd. planned the construction plan, organized construction forces, and compiled a demand list to launch the platform construction. After iterative upgrades, in April this year, all core functions of the platform were deployed and entered the trial operation phase.
Why was the platform built in Langfang? The reason is that Langfang is adjacent to Beijing and Tianjin, possessing both locational advantages and computing power advantages. At the same time, the latency from the Langfang computing cluster to the Beijing computing platform is controlled within 2 milliseconds, enabling it to undertake Beijing's massive computing demand nearby.