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

The strong demand for computing power generated by the rapid spread of AI applications is making the intelligent computing center a new type of public computing infrastructure in the age of mathematical intelligence. Data show that in the past two or three years, at least 30 cities in China have built or are building intelligent computing centers. However, the construction of the intelligent computing center is also faced with many problems, such as high cost, high energy consumption, and idle computing caused by blindly following the trend.

Dai Wei, vice president of ▲ Lenovo Group and general manager of China Program Service Business Group, said in an interview with reporters at the 2023 China Computing Conference that just concluded, Lenovo Vice President and General Manager of China Program Service Business Group said that the regional intelligence center will become the core direction of future computing power construction, the "background" is green, and subscription will also become the mainstream of future construction.

At this conference, Lenovo Group launched for the first time the core products of Smart Computing Center solutions and services as a whole, including three major types of solutions-Lenovo brain embedded solutions, Lenovo hybrid cloud solutions, Lenovo green low-carbon computing center infrastructure and two major types of services-end-to-end full-cycle Lenovo Computing Center services. and on-demand subscription, flexible payment Lenovo computing service 2.0. " Lenovo will provide end-to-end solutions and full-cycle intelligent services for the construction of the intelligent computing center, breaking many unsolved problems. " He said.

Many problems to be solved in the construction of intelligent computing center with high cost and high energy consumption

Statistics show that China ranks second in the world in terms of computing power. From the perspective of infrastructure construction, China's computing industry mainly includes traditional data center, super computing center, intelligent computing center and so on. At present, under the catalysis of artificial intelligence technology and application acceleration, the intelligent computing center is ushering in a new round of outbreak.

According to the White Paper on the Development of Intelligent Computing Industry just released by the 2023 China Computing Congress, intelligent computing accounted for more than half of China's new computing facilities in 2022, and it is expected to account for more than 85% in 2025. However, in the construction of the intelligent computing center, the whole industry still faces many problems, including the high investment cost in the construction, the huge energy consumption in the use process, the idle calculation and so on.

According to media reports, taking 100p as the unit of measurement, the construction cost of an intelligent computing center with 16-bit precision computing power per 100p in a city in 2020 is 75 million yuan. Under the same conditions, the construction cost in 2021 will reach 460 million yuan. For intelligent computing centers with the same positioning and similar functions, the difference in construction investment is up to 6.2 times.

The PUE of the Intelligence Center is also a problem that can not be ignored. In June this year, six departments, including the Ministry of Industry and Information Technology, issued an action plan to improve industrial energy efficiency, which clearly stipulates that by 2025, the PUE of the new large and super-large data center will be better than 1.3. In fact, after the completion of the construction of some intelligent computing centers, the initial PUE value has been high, even as high as 1.9. Under the spring tide of "new infrastructure" of the Intelligent Computing Center, enterprises not only need to bear the high investment costs, but also have to face huge electricity consumption.

In addition, there are some problems in the intelligent computing center, such as idle calculation, confusion of concepts and so on. On the whole, the state and industry have begun to strengthen overall planning and guidance from the source. For example, the National Information Center issued the Guide to the Planning and Construction of the Intelligent Computing Center in 2020 to further clarify the standards of the Intelligent Computing Center, guide the Intelligent Computing Center to define the application scenarios and priority tasks, and ensure that the construction of computing resources is reasonable and green.

Let computing power become public resources like "hydropower". The construction of regional intelligence centers will become the mainstream.

In response to the above problems to be solved, Dai Wei believes that in the construction of future intelligence centers, there will be two types, one is regional intelligence centers, and the other is enterprise intelligence centers, in which the former will become the mainstream of future development, and the background color will be "green." the mode of service will be subscription-based, pay-on-demand mode.

According to reports, the Regional Intelligence Center takes large-scale AI training as the main application scenario, achieving regional radiation, and even supporting national computing scheduling and application needs, with one cloud multi-core, multi-cloud management, cost control and high-standard operation and maintenance capabilities. The Enterprise Intelligence Center uses small and medium-sized computing resources to support the core applications of enterprises.

He believes that the regional green computing center will have a large-scale hybrid computing architecture and resource pool, while providing high-standard operation and maintenance capabilities through green and efficient technologies such as liquid cooling technology and multi-cloud management and fee control. to meet the needs of many companies, users or applications in the region.

Under the background of the strategy of calculating the east and the west and the construction of a national network of computing power, computing power will become a public resource like hydropower. In the future, most enterprises will no longer need to build their own intelligence centers, and they can get corresponding support in the regional intelligence centers. He said that Lenovo Intelligent Computing Center solutions and services will provide end-to-end full-cycle support for heavily invested intelligent computing centers in the future.

Among them, the solution embedded in Lenovo brain solves the problem of "what is" in the intelligent computing center, and promotes the landing application of intelligent computing power by providing core algorithms and intelligent scene models in various industries. Lenovo Lenovo xCloud hybrid cloud solution meets a variety of complex needs, such as multi-core cloud, public cloud, private cloud, as well as hybrid computing, such as computing, supercomputing and computing. The infrastructure of Lenovo Green low-carbon Intelligent Computing Center creates the "green" background of the Intelligent Computing Center.

The two major services respectively provide end-to-end, full-cycle companion services from consulting design to system operation, and innovative on-demand subscription and flexible payment business models to help reduce the cost of investment in the center.

Case of intelligent lighthouse in multiple benchmarking areas strong plus intelligent computing power new infrastructure

At present, Lenovo Group has helped to build a number of large-scale regional intelligence centers in Ningxia, Gansu and other places, comprehensively accelerating the large-scale landing of advanced technologies such as AI.

▲ Gansu Zijinyun Public Computing Center Zijinyun Public Computing platform is currently the largest scientific computing and intelligent computing center in Gansu, which not only undertakes the collaborative hub function between the western and eastern regions, but also faces a large number of computing and storage needs in the northwest region, so it is necessary to promote the complementary and integrated and balanced development of eastern and western computing power.

Starting from scratch, Lenovo helped to build the first phase of the computing platform of the center with the full-process services of design, construction, and subsequent operation and maintenance, and the peak computing speed of the overall system reached 2.73PFLOPS computing capacity per second. In the future, on the basis of the first phase of the project, the computing power of 3PFLOPS will be increased. In the aspect of green and low carbon, the overall energy saving and carbon reduction management is realized through the overall heat dissipation system such as "natural cooling + air cooling + evaporative cooling" and waste heat recovery scheme, from hardware equipment to overall operation.

In Zhongwei, Ningxia, Lenovo helped parallel Technology to create the first intelligent computing center relying on improved computing services, creating a precedent for the subscription model of computing services. With Lenovo computing services, parallel technology can cost-effectively expand its cloud services infrastructure. The solution uses an on-demand billing model, which means that parallel technology only pays for the computing resources it consumes without having to buy the hardware itself. At the same time, parallel technology obtains the convenience of transparent and quarterly payment, avoiding a huge amount of upfront capital investment.

"Lenovo Computing Service has helped us transition to a more agile operating model," said Liu Haichao, Vice President and Director of Resource Systems of parallel Technologies. "We can easily expand capacity and resources according to the actual needs of users, and minimize risk and IT expenses. In addition, the quarterly payment model can not only save initial costs, but also free up valuable funds for other more important business activities in the future. "

In fact, there are many such cases. Dai Wei said that under the wave of the fourth industrial revolution, intelligent computing is becoming a solid base and powerful engine for the high-quality development of the digital economy. Lenovo will use the original solutions and services of AI to continuously create green and low-cost intelligent computing power for Qianhang Baiye, accelerate the integration of AI and vertical scenarios, and help Chinese enterprises achieve the acceleration of intelligent transformation.

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.