Flowing "computing power": Digital China's "new Yellow River" continues in the intelligent era

Among all the civilizations in the world that rely on the river system, the Yellow River civilization is unique. For thousands of years, the Chinese nation, represented by the Central Plains civilization, has had a profound connection with the Yellow River. Our ancestors made full use of the Yellow River resources through water conservancy, farming, shipping, and fisheries, turning the water of the Yellow River from "usable water" into a "useful spring" that nourishes civilization.

In the intelligent era, computing power is the new water power. How to make the computing power resources flowing through the land of digital China become the "new Yellow River" in the intelligent era has become a profound proposition of this era.

In the fertile land of the Central Plains, beside the Yellow River, an action to transform new computing power into a new Yellow River is underway. From September 27 to 29, the 2024 China Computing Conference was held in Zhengzhou. Intelligent computing construction and application became the most concerned topic of this year's computing conference.

During the Computing Power Conference, Lenovo hosted the 2024 Heterogeneous Intelligent Computing Industry Ecosystem Alliance Technology Forum with the theme of "Heterogeneous Intelligent Computing for the Future".

From this we can get a glimpse of the latest trends in computing power construction for intelligent development under the general context of Digital China construction, as well as what preparations Lenovo has made in response to the trends and demands of the times.

"Easy to use computing power" is the new coordinate in the eyes of Lenovo and the entire intelligent computing industry. Following its direction, we can discover and build a "new Yellow River" for digital China.

As a key force in developing new types of productive forces and the foundation of digital China, computing power construction has become an urgent need of the times.

The 2024 Government Work Report proposes to "moderately advance the construction of digital infrastructure, accelerate the formation of a national integrated computing power system, and cultivate a computing power industry ecosystem." Since then, the Ministry of Industry and Information Technology and six other departments have issued favorable policies such as the "Action Plan for High-Quality Development of Computing Power Infrastructure" to guide the vigorous development of China's computing power industry.

In this era, China's total computing power continues to rise. By the end of 2023, China's computing power will reach 230EFLOPS, of which intelligent computing accounts for nearly 30%, storage capacity will reach about 1200EB, and advanced storage capacity will account for more than 25%.

After a considerable period of accumulation of computing power, the key to the development of China's computing power industry is how to achieve a leap from quantitative change to qualitative change. While continuing to expand the scale of computing power and achieve a balance between supply and demand, it is also necessary to solve various difficulties in the application of computing power in various industries, such as the mismatch of storage, computing, and network resources; the management and operation and maintenance problems of heterogeneous intelligent computing; and the deep integration of computing power with real application scenarios. In summary, the core of the current development of the computing power industry is to achieve the transition from expanding the scale of computing power to releasing the value of computing power with high quality.

The China Computing Power Conference is the industry's top stage, known as the "computing power industry weathervane, computing power innovation source, and computing power ecosystem incubator". At this conference, how to promote the leap from quantitative change to qualitative change in China's computing power industry in the AI ​​era has become the most concerned, discussed and exchanged topic in all walks of life. Among them, Lenovo gave its own answer - to make the leap from "computing power available" to "computing power is easy to use".

Chen Zhenkuan, vice president of Lenovo Group and general manager of China Infrastructure Business Group, believes that "in the process of computing power empowering economic and social development, it is necessary to complete the key transition from computing power technology accumulation to computing power value release. The core of this process is to achieve an important upgrade from 'computing power available' to 'computing power is easy to use'."

The value of "easy-to-use computing power" can be seen from the expansion of new computing power application scenarios brought by AI big models. Big models bring huge computing requirements, as well as complex computing architectures, new computing models, and rich software and hardware combinations. This means that if you want to make good use of big models and fully exert their intelligent value, you can't just expand the scale of intelligent computing, but also put forward diversified and systematic high requirements for computing power users.

For intelligent computing scenarios, the availability of computing power means that computing power users can build large models and run multiple types of computing tasks through a stable combination of software and hardware. Then, the availability of computing power requires intelligent computing power to become a direct support for AI applications. This requires providing end-to-end overall computing power optimization from chips to applications to achieve a significant upgrade in computing power efficiency and performance, thereby meeting the implementation needs of diverse and complex AI applications.

It can be said that "computing power is easy to use" will be the core goal of the construction of digital China's computing power in the next cycle. Following this goal, the entire industry is jointly looking for and exploring a "computing power yellow river" that allows computing power to flow continuously and allows the value of computing power to benefit thousands of industries.

Lenovo has made full preparations for this.

The key to moving from "computing power available" to "computing power is easy to use" is to create an integrated and highly complete computing power infrastructure solution so that the complex computing power and intelligent computing needs of various industries can be properly addressed. This is like digging a river channel, and all the joints must be opened up to allow the water to flow into thousands of households.

In order to achieve this goal, it is necessary to build full-stack AI advantages and computing power advantages. Based on this, Lenovo began to proactively deploy AI technology seven years ago. Facing the wave of the explosion of AI big models today, Lenovo timely proposed a "full-stack AI" strategic layout based on the new IT architecture of "end-edge-cloud-network-intelligence".

This can be seen especially at the Computing Power Conference. Unlike other exhibitors who mainly focus on the B-end market, Lenovo is the only manufacturer that can provide AI terminals, AI infrastructure, and AI computing power solutions and services, truly realizing the full stack AI "from pocket to cloud".

Among them, in response to the demand for AI solutions and services, Lenovo has built the "One Engine and Three Arrows" based on the Dynaudio architecture; in response to the AI ​​iteration of terminal products, Lenovo released the world's first AI PC in May this year, making the PC a true personal assistant.

Back to the intelligent computing infrastructure, Lenovo continues to upgrade its computing power strategic framework and build the Lenovo Wanquan Heterogeneous Intelligent Computing Platform. It can unify and manage heterogeneous computing power to maximize intelligent computing efficiency. At the same time, Lenovo has created a complete product portfolio of "one horizontal and five vertical" to provide comprehensive support for diverse AI infrastructure needs.

The full-stack AI strategy forms the strategic foundation for Lenovo to explore "computing power for practical use". In response to the upgrade needs of the intelligent computing system, Lenovo further consolidates its AI infrastructure product portfolio to provide computing power, storage power and carrying capacity for AI scenarios.

In terms of computing power, Lenovo is committed to providing a more powerful and comprehensive intelligent computing power product portfolio. After a long period of planning, Lenovo has built an AI full-scenario server product portfolio covering large model training, reasoning, training-in-one, and edge reasoning.

In terms of storage capacity, Lenovo has built a full-scenario, high-performance intelligent storage base, providing large-capacity, high-density, green and energy-saving storage products for data preparation; for model training and reasoning, it provides storage products of different model sizes, supporting flexible expansion and high-performance storage.

In terms of carrying capacity, Lenovo has created a multi-architecture, low-latency intelligent network architecture, enabling data network products to support multiple AI network architectures and create high-speed AI network solutions for different AI scenarios.

With full-stack AI as the main channel and the continuous construction of computing power, storage power, and carrying capacity as tributaries, Lenovo has opened up a new computing power Yellow River under the goal of "computing power is easy to use" in the AI ​​era.

"Easy to use computing power" is not an overnight achievement. After the "river channel" is connected, we must work tirelessly to open up the channel between AI infrastructure and AI scenario needs. Continuous technology upgrades, product improvements, and ecological promotion are the support for the new Yellow River of computing power to continue to flow towards the goal of "easy to use computing power".

At the 2024 China Computing Conference, Lenovo brought its latest moves in the field of AI infrastructure, which we can summarize into three new channels.

1. New product channels.

In response to the continuous upgrading of intelligent computing needs, Lenovo has released a new generation of AI servers for AI large model training, Lenovo Wentian WA7880a G3. This server is the first in China to implement the OAM 2.0 module, which is compatible with the OAM GPUs of mainstream domestic GPU manufacturers, meeting customers' needs for diversified computing power. For AI needs such as model training, reasoning, high-performance computing, and data analysis, WA7880a G3 can give full play to its advantages in strong performance, flexible configuration, and green energy saving to help users achieve their intelligent goals.

In addition, Lenovo has also released Lenovo AIPod, a solution for enterprise AI application deployment to meet more corporate needs for AI scenarios.

2. Green new channel.

In the era of dual carbon, both enterprise users and intelligent computing centers pay more attention to the energy efficiency of deploying AI capabilities. Under this trend, liquid cooling + AI has become a general trend. Last year, Lenovo launched a liquid cooling strategy with "green empowerment" as the core to the market. This year, Lenovo has fully upgraded Lenovo AMT liquid cooling technology, based on Poseidon liquid cooling technology, integrating Lenovo's long-term rich accumulation in the field of liquid cooling technology, while insisting on localization with the Chinese market, comprehensively improving the comprehensive competitiveness of Lenovo AMT liquid cooling technology.

3. New ecological canal.

In the current environment, active ecological development is the best way to develop China's computing industry and the core driving force for the development of intelligent computing. Against this backdrop, under the guidance of the China Intelligent Computing Industry Alliance, Lenovo and many partners jointly launched the Heterogeneous Intelligent Computing Industry Ecological Alliance, relying on important national strategic think tanks such as the Chinese Academy of Sciences and the China Academy of Information and Communications Technology.

During this computing power conference, Lenovo continued to promote the development of the heterogeneous intelligent computing industry ecological alliance and released the "Heterogeneous Intelligent Computing Industry Trends and Technology Development White Paper" with alliance partners. In the frontier fields of AI computing power applications such as intelligent risk control, biomedicine, natural language processing, image recognition and processing, Lenovo and alliance members demonstrated their latest progress and promoted the overall development and standardization of the industry.

With the continuous advancement of the full-stack AI strategy and the full support of new products, new technologies, and new ecosystems, Lenovo is turning the era goal of "easy-to-use computing power" into a yellow river of computing power that flows in all directions. What flows in it is not only the water of computing power, but also the source of innovation for new quality productivity.

From traditional productivity to emerging productivity since the 16th century, and then to today's new quality productivity, the core of the driving mechanism is to realize integrated innovation behind science and technology as the main driving force for the development of productivity.

"Easy-to-use computing power" is such an integrated innovation. It connects the needs of the times and guides the progress of science and technology. By continuing to explore along this goal, the water of intelligent computing will nourish thousands of industries and bloom new productivity.