Starting from the latest technological development trends of large models, this paper first analyzes the architectural characteristics and computing power demand features of multimodal, long sequence, and mixture of experts models. Further, it focuses on the requirements of the latest large models for massive computing power scale and complex communication patterns. It quantitatively analyzes the current development problems and technical challenges faced by large model computing infrastructure from two aspects: computating efficiency and cluster interconnection technology. Finally, it proposes a high-quality computing infrastructure development trajectory oriented by applications, centered on systems, and targeted at efficiency.
Against the backdrop of the explosive development of generative Artificial Intelligence (AI) large language models, the importance of intelligent computing is becoming increasingly prominent. Currently, large language models with hundreds of billions or even trillions of parameters have raised demanding requirements for intelligent computing in terms of chip computing power, memory capacity, and interconnection speed during the training process of massive data. Starting from the demand for key technologies of intelligent computing, this study focuses on analyzing the development trend and the current industry situation, in dimensions such as computing chip technology, software technology, and interconnection technology. Combining with the opportunities and challenges faced by the current industry development, it proposes future development strategies for intelligent computing.
The new generation of information technology represented by artificial intelligence is accelerating innovation, posing new requirements for the development of database technology in terms of data structure, deployment methods, and application scenarios. Focusing on the new demands and challenges of database innovation in the era of artificial intelligence, this paper first summarizes the main directions of current database innovation. Then, it focuses on the integration and practical application of artificial intelligence and database, and analyzes the current development trends and competitive landscape of the database industry both domestically and internationally. Finally, it puts forward suggestions for the development of technological innovation in China.
As the demand for the level of immersion in extended reality (which includes virtual reality and augmented reality) continues to grow, video see-through (VST) has become a more promising mixed-reality technique than optical see-through (OST) due to its advantages in environment brightness adaptability, field of view and occlusion implementation. However, a few products equipped with VST have showcased multiple problems as the technique is still in an early phase. Therefore, the analysis has been conducted in the aspects of temporal, spacial and image quality, focusing on problems of image latency, point of view accommodation, oculus resolution, perspective distortion and dynamic range. Hopefully, this paper can provide an insight into VST status quo and facilitate its future development.
Performing data asset assessment is a crucial step in realizing data assetization, holding significant importance for enterprises in optimizing data asset governance and guiding the valuation and pricing of data assets. This research establishes an evaluation index system of enterprise data assets by focusing on four aspects of data assets: quality, cost, risk, and application. It also constructs a quantitative data asset evaluation model that incorporates techniques such as the stepwise normalization method and the Analytic Hierarchy Process.
Industry is a key area of energy consumption and greenhouse gas emissions in China. Due to the increasingly strict climate change regulations and the continuous rise in carbon emission costs, the demand for carbon emission management systems in industrial enterprises and parks continues to increase. With the comprehensive empowerment of advanced computing technologies such as big data, artificial intelligence, and blockchain, China’s digital carbon management technology system has basically formed. It is crucial for China to accelerate the application and promotion of digital carbon management systems and deepen their development in order to build a basic capacity for dual control over the amount and intensity of industrial carbon emissions. Starting from research on the current development situation of digital carbon management at home and abroad, this article analyzes the architecture of industrial digital carbon management system and the application of advanced computing technology. It also analyzes the hot trends of advanced computing technology empowering the development of industrial digital carbon management. Combining with the key issues faced by current application promotion, it puts forward suggestions for the development of industrial digital carbon management in China.
This paper proposes a design of monitoring service platform for computing power industry based on large language model technology, aiming to provide comprehensive support for intelligent decision-making in monitoring the development of the computing power industry by efficient technical means. The platform adopts a layered architecture design, with a bottom layer supported by a domestic large model base with 10 billion parameters, while the upper layer serves as a business application system, providing computing power development insights, intelligent analysis assistance, and other services.
At present, the wave of information revolution is surging, with rapid development in digitization, networking, and intelligence. New technologies, new formats, and new models are emerging. Digital infrastructure, as an important physical carrier of digital technologies such as communication networks and digital computing power, has become a key construction object in the digital transformation process of the Shanghai Cooperation Organization (SCO) states. Based on the core essence of digital infrastructure, this paper constructs an evaluation index system for the development level of digital infrastructure in the SCO states in six dimensions: development environment, market demand, network facilities, computing power facilities, service capabilities, and growth trends. The combination weighting-TOPSIS model is used to measure the development index of digital infrastructure in the SCO states. The advantages, experience, and characteristics of developing digital infrastructure in each SCO state are summarized to know their overall development levels, and regional disparities are analyzed. This paper also proposes a high-quality development path for digital infrastructure in the SCO states, in order to provide an important basis for infrastructure construction, technical cooperation, and joint research in the digital field among the SCO states.
Due to the evolution of 5G technology, new solutions and approaches have been provided for the comprehensive application of all element information required for intelligent connected vehicles, especially high-level autonomous driving. Therefore, this article elaborates on the performance testing and application pilot of the latest 5G/5G-A technology in the field of vehicle networking from four aspects: technological evolution, solutions, performance testing, and application incubation, and provides a comprehensive solution for all element vehicle networking.
With the rapid development of mobile internet technologies and the rapid popularization of electronic devices, various applications present explosive growth, and the problems of application security and personal information protection are increasingly prominent. Focusing on the problems and challenges faced by application security and personal information protection, this paper describes the objectives of HarmonyOS application security and personal information protection, focusing on typical issues related to application security and personal information protection. For each stage of application development, release and operation, it proposes the design and reflections of security protection measures throughout the entire lifecycle of HarmonyOS. It aims to build a more secure application ecosystem.
Currently, the technology and applications of large language models (LLMs) have been developed rapidly. The capabilities of LLMs with the capabilities of semantic understanding and content generation have broad application prospects in the field of digital government. This paper summarizes the development of the LLMs for digital government from the perspectives of technology, industrial chain and application scenarios. Then, this paper puts forward the architecture, key technologies, and implementation paths of the LLMs for digital government, expounds the problems that limit the further development of LLMs’ applications, and finally gives suggestions for further development.
The large amount of real traffic data generated by the infrastructure of the Internet of Vehicles can be used to build a scenario library for simulation testing of intelligent connected vehicles. This article proposes a method system for generating a test scenario library based on the collection of data from the Internet of Vehicles infrastructure, including data collection, scene extraction, scene generation, scene distribution, data desensitization, and looks forward to future development trends.
This paper summarizes the origin and development of the European Union’s International Data Spaces (IDS) and China’s Trusted Data Matrix (TDM). In view of the current system and technology bottlenecks of TDM, a new underlying technical framework of “Spatio-temporal Coding + Internet of Data” is introduced. On this basis, the concept connotation, system architecture, functional view and characteristic fusion base of the new TDM are systematically proposed. The main innovation of this paper is reflected in the trustworthy system and implementation path, technical framework and system architecture, platform positioning and ecological cultivation. It is suggested that the new TDM should be the key technical route for the construction of China’s data circulation infrastructure, and the cooperation of technical research and development should be carried out to accelerate the pilot verification demonstration.