Shulou(Shulou.com)12/24 Report--

There is no doubt that people are ushering in an intelligent world that is accelerating. From cloud computing to edge computing and accelerated computing, with the rise of artificial intelligence era represented by generative AI, digital scenes and intelligent applications flourish in thousands of industries. However, behind this intelligent era, IT infrastructure is also facing modern upgrades. As the cornerstone of data center computing power, changes in the server market largely reflect the degree of digitalization of IT infrastructure, directly affecting the effectiveness of business applications. In IDC's 2023H1 China Server Market Tracking Report, the overall performance of the domestic server market in the second quarter was better than that in the first quarter, with revenue up 6.9% year-on-year. Within this pattern, accelerated compute servers and ARM servers are growing particularly strongly, and this trend is expected to continue.

Under this background, Chaoyun successfully ranked among the top ten server suppliers in China by virtue of the eighth place in overall sales volume of domestic brands in the first half of the year. It is understood that Chaoyun was jointly funded by China Electronics and Cloud Base. It is a national high-tech enterprise and a state-level specialized and new "Little Giant" enterprise. It focuses on data center products and services. After more than ten years of development, Super Cloud products have covered many fields such as intelligent computing, information innovation, private cloud and cloud services, serving many industries such as Internet, finance, telecommunications, manufacturing, education, energy and public services.

As a growth enterprise, it is quite rare for Chaoyun to maintain rapid growth in a highly competitive market. In recent years, traditional general purpose server shipments have been declining, and vendors are looking for new market opportunities. One direction is that cloud service providers deploy ARM architecture servers on a large scale through self-research, which brings opportunities to many ODM vendors. The other direction is to build "dedicated" servers around edge computing and accelerated computing, especially in AI, which has greatly boosted the growth of computing, networking, storage and other related markets.

Layout intelligent calculation main AI and liquid cooling

In the face of market changes, Chaoyun clarified the development direction of intelligent computing, and increased investment in AI and liquid cooling technology research and development efforts. This is in line with industry trends and contributes significantly to its market share and performance growth. Tian Feng, general manager of Chaoyun Intelligent Computing Division, said that AI computing industry is in a stage of rapid development from policy support, technology trends to customer demand. "Since the second half of the year, the development of super cloud in AI field has grown significantly." The accelerated construction related to intelligent computing will inevitably lead to brand-new business application scenarios, while general computing will still have some matching construction requirements.

Supercloud has launched a number of series of intelligent computing products for AI training, reasoning, graphics workstations, edge AI computing and other computing needs under different scenarios, among which R8868G13 has attracted much attention. The R8868 G13 is a high-performance AI server based on Intel ® 4th Generation Xeon ® Scalable Series processors with a heterogeneous acceleration architecture that delivers extreme scalability, high performance, high reliability, and easy maintenance. This server focuses on large model training, large model reasoning, HPC, deep learning computing, rendering and other applications, which can greatly reduce the training cycle and is the first choice for AI computing and high performance computing business.

Super Cloud R8868 G13 High Performance AI Server

In the second half of this year, Super Cloud, jointly with Intel and Smart Spectrum AI, released AI large model training and pushing all-in-one machine, based on Super Cloud R8868 G13 server. This "out-of-the-box" trainer is powered by Intel's 4th Generation Xeon Scalable Series processors, compatible with modules such as Intel ® Habana® Gaudi®2C, and features superior heterogeneous acceleration architecture. It can be used for large model experience evaluation, pre-research and demonstration of large model application scenarios, large model Prompt development, large model application development and large model teaching training, etc. It can easily solve technical problems such as customers without professional computer room, soft and hard separation, expansion difficulties, and long delivery cycle.

Today, intelligent computing is not only a test of data center performance, but also a test of green energy efficiency, which is also the product concept that Super Cloud has always adhered to. From initially providing liquid cooling services for supercomputers to liquid cooling technology now applied to AI products, Super Cloud hopes to solve the problems of high heat generation and unstable performance caused by rapid increase of TDP of chips through liquid cooling solutions, and reduce the overall energy consumption of data centers. Based on mature cold plate technology, Super Cloud has launched a series of products such as cold plate liquid cooling server, complete cabinet liquid cooling server and liquid cooling micro module. In terms of immersion liquid cooling, Chaoyun has established a joint laboratory for thermal control of electronic equipment jointly with Beijing University of Chemical Technology and Zhejiang Institute of Chemical Technology to develop new liquid cooling materials so as to realize localization substitution as soon as possible. In addition, Chaoyun also participated in the compilation and release of two standards: Technical Reference Architecture for Liquid-cooled High-performance Computers and Technical Requirements for Immersion Liquid Cooling Systems for High-performance Computers, which filled the gaps in the design, development and application of liquid cooling technology in the field of high-performance computers and provided important reference basis for cold plate and immersion liquid cooling architecture and technical design.

Supercloud liquid cooling products have also been verified in practice. In an energy cold plate liquid cooling room project, Chaoyun designed a liquid cooling system for both CPU and GPU. The system's one-to-nine liquid module is decoupled from the whole machine, so that DCU (PCIe card) can be maintained independently and operated conveniently. The liquid distribution module is an integrated design with a height of 1U. The liquid inlet and outlet are of "upper inlet and lower outlet" structure. The main liquid inlet is located in the middle, and 9 groups of branch ports are located at both ends. Among them, 8 groups are GPU liquid cooling liquid supply and 1 group is CPU liquid supply, realizing uniform flow distribution and balanced heat dissipation performance. The whole machine liquid cooling 3560W, of which, CPU 560W, GPU 350W. Cold plate overall use of metal cold plate + hose combination form, high reliability, flexible deployment, excellent performance has been affirmed by customers.

Layout Scheme of Super Cloud Cold Plate Liquid Cooling Machine Room

Add code to create a safe computing base

AI and Xinchuang are the two pillars supporting the rapid growth of super cloud business. In the field of Xinchuang, Chaoyun cooperates with Feiteng, Longxin, Kunpeng and other partners to create a multi-computing development strategy. In cooperation with C86 platform, Super Cloud independently developed a variety of products and obtained excellent shipment performance.

Qianxin Technology Group Co., Ltd. chose Super Cloud to help it realize the implementation of cloud security management platform. This project needs to meet a number of requirements, including platform standardization, modularization, orchestration, traffic scheduling, security service chain, etc. At the same time, it needs domestic servers equipped with relevant software to provide services. Through software-defined security architecture, Qianxin adopts Feiteng server R2211 developed by Super Cloud in IT infrastructure layer. This 2U single-socket rack server is flexible in configuration and widely suitable for cloud computing virtualization, database, big data and other workloads. In addition, the server supports IPMI2.0 technology, which can provide various management functions such as status monitoring and KVM, realize remote fault diagnosis and alarm of the server, reduce the difficulty of operation and maintenance, improve the operation and maintenance efficiency of the data center, effectively reduce the operation cost and improve the return on investment.

Based on the security computing base built by Super Cloud, Qianxin has created leading cloud security solutions, formed rich security capabilities, built and deepened the cloud security defense-in-depth system, and met the security requirements of cloud business applications and tenants. At the same time, the performance guarantee of Super Cloud Virtual Machine helps users on cloud to build collaborative defense system, improves the efficiency of Incident Response Service, meets the steady landing of customer innovation projects, and further consolidates Qianxin's leading position in network security field.

Super Cloud R2211 Server

Information security is the key field to maintain national security, and trusted computing is the basis to ensure information security. As a professional data center product and service provider in China, Super Cloud not only has a mature basic product line, but also increasingly improves the underlying trusted computing technology system. This year, Chaoyun launched the C86 platform security trusted all-in-one solution, which recently won the benchmark award of the 3rd Photosynthetic Organization Solution Competition, and has been successfully delivered to government agencies and central state-owned enterprises and other units with extremely high security requirements.

Super Cloud Security Trusted All-in-One Solution

Based on the trusted design of C86 CPU, this scheme adopts the built-in secure processor chip (PSP) in C86 CPU as the computing unit of TPCM, establishes resource isolation and interaction mechanism, and constructs active immunity dual-architecture. Through TPCM operating system suitable for PSP chip and its resources, PSP platform has active measurement function, becomes the trust root of the whole hardware platform, realizes core functions such as active measurement of memory and establishment of trust chain, and realizes complete trust chain through trusted transformation of BlOS and operating system, forming a high-performance and high-security trust-body scheme.

It can be seen that Chaoyun's years of deep cultivation in AI and Xinchuang fields have achieved initial results. Under the uncertain market structure, the firmness and self-confidence shown by Super Cloud are not only reflected in business growth, but also in customer trust. With the continuous development of China's digital economy industry, Chaoyun's high-quality growth path will surely move forward steadily. Looking forward to the future, Chaoyun hopes to achieve sustained high growth of AI business in 2024, stabilize CSP customers, strengthen and expand industry customers such as operators; continue to exert efforts in the field of information innovation, laying a solid foundation for digitalization of China's industry.

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.