2024 MWC Shanghai | H3C fully demonstrates new achievements in intelligent computing, accelerating the construction of a new intelligent era

From June 26 to 28, 2024, the Mobile World Congress Shanghai (MWC Shanghai) grandly opened at the New International Expo Center. Themed "Future First," this year's conference focused on cutting-edge topics such as "5G and Beyond," "AI Economy," and "Manufacturing DX." Under the theme "Dedication, For A Smarter Future", H3C comprehensively showcased its innovative products and solutions in the fields of "Cloud-Network-Security-Computing-Storage-Terminal," as well as the practical outcomes of AI technology applications across various industries.

During the exhibition, H3C's booth attracted close attention from domestic carriers and industry customers, as well as overseas customers and partners from Spain, Turkiye, Saudi Arabia, Mexico, Vietnam, Thailand, Indonesia, and the Philippines. They delved deep into H3C's full-stack technical strength and its successful cases and practical experiences in the global market.

Computing Power × Connectivity: Unleashing the Infinite Potential of AIGC

Amidst the explosive growth in computing power demand, H3C, through technological innovation, has achieved a deep integration of diversified computing power and standardized connectivity, breaking the "invisible barriers" within AIGC infrastructure. This allows for more flexible scheduling and release of computing power, providing more efficient and stable support for model training and inference.

LinSeer Intelligent Computing Solution: H3C fully demonstrated the industry-leading LinSeer Intelligent Computing solution, fully leveraging the multiplying effect of "computing power x connectivity." Anchoring the "1+N" large-scale model strategy (LinSeer private large-scale model + N selected ecological large-scale models), it accelerates the formation of new quality productive forces across various industries.

Multi-Heterogeneous Computing Power: Facing the continuous growth in computing power demand from generative AI, H3C introduced the new multi-heterogeneous computing power platforms H3C UniServer R5330 G7 and H3C UniServer R5500 G7. With multiple advantages such as diverse heterogeneity, powerful computing capacity, flexible architecture, and strong scalability, these platforms provide robust support for different AI workloads, contributing to the high-quality development of the digital economy.

The AMPHA Computing Power Platform 3.0: Centered around "computing power as service," the platform achieves efficient utilization of computing resources through unified intelligent scheduling of heterogeneous computing power, AI full-cycle development services, and large-scale intelligent computing center operations management. Supporting the scheduling of unified computing power across large-scale clusters and compatible with various architecture acceleration cards, it significantly enhances resource utilization. Additionally, it provides AI full-cycle development services from data processing to application deployment, accelerating the practical application of AI technologies.

Intelligent Computing Center Network: Aimed at AI large-scale model training scenarios, H3C's end-to-end collaborative intelligent computing network solution features large-scale networking capabilities, highly efficient computing communication, and superior operational capabilities. With an advanced DDC (Distributed Data Center) architecture, it offers innovative zero-congestion lossless solutions, enhancing computing communication efficiency. H3C also teamed up with China Mobile to launch the industry-leading Global Scheduling Ethernet (GSE) solution, jointly creating a new intelligent computing network. H3C showcased the world's first single-chip 51.2T 800G CPO silicon photonics switch, capable of supporting an AIGC cluster size exceeding 32,000 nodes and improving GPU computational efficiency by 25%, ensuring maximum AIGC cluster computational performance.

Wide Area Computing Network: Addressing the computational power needs of customers across various scales and industries, H3C has built a high-speed, low-cost, computational power-adapted "computational power highway" by upgrading 400G ports and incorporating computational power factors into routing algorithms. Additionally, it launched service-based computational power private lines, providing elastic, differentiated computational power transmission services for sudden large-scale computational power demands, enabling on-demand construction and significantly enhancing the cost-effectiveness and service flexibility of government and enterprise customers, promoting the inclusive supply of computational power resources.

Converged Edge Cloud Network: H3C provides critical technical support for the business transformation of To C/To H/To B through computational power sinking to build ecological platforms, attracting multiple businesses to settle at the edge. On the cloud side, edge application servers adopt a modular design, enhancing GPU inference capabilities, adapting to various environments, and ensuring green and efficient delivery. On the network side, the edge cloud gateway's control-separation design and large-capacity high-density characteristics support multi-business offloading, achieving cloud-edge collaborative optimization.