H3C UniServer R5300 G7 Server

H3C UniServer R5300 G7 is a cutting-edge 6U dual-socket AI server, expertly designed by H3C. It combines advanced architecture, outstanding performance, and flexible scalability. This server supports up to 8 high-performance GPUs within its compact 6U form factor and is powered by Intel's latest Birch Stream platform processors, offering storage for 12 drives. It delivers robust and versatile computing power for AI applications like deep learning and AIGC, enabling enterprises to establish an efficient and reliable computing platform for the intelligent era.

Product features

Diverse Computing Power: Building a Future-Ready Intelligent Computing Hub

H3C UniServer R5300 G7 PCIe GPU server serves as a future-ready intelligent computing hub, fully compatible with various brands of AI accelerator cards through its diverse computing engines. Its multiple heterogeneous architectures cater to the distinct needs of training, inference, and HPC. Supporting CX8 SuperNIC and PCIe Switch architectures, it integrates an 800Gb/s high-speed network, increases total cluster Scale-out bandwidth by 300%, and achieves up to 400Gb/s bandwidth per GPU. With five topologies available—AI serial, AI parallel, HPC, CX8 SuperNIC, and passthrough—users can easily configure based on specific AI requirements, striking the perfect balance between performance and TCO for comprehensive intelligent computing support.

Exceptional Hardware Configuration: Redefining AI Computing Limits

H3C UniServer R5300 G7 features two Intel® Xeon® 6th Gen (Granite Rapids) Scalable Processors, each supporting up to 86 cores. Coupled with 32 channels of 6400MT/s DDR5 memory, it can expand to a massive 8TB capacity, dramatically enhancing throughput for high-concurrency tasks.

For storage, this server boasts 16 hot-swappable 2.5-inch drive bays, enabling mixed deployment of SATA/SAS/NVMe. It can be configured with up to 200TB of all-flash memory, fully addressing storage needs for high-throughput tasks like AI training and big data analytics.

Additionally, this model includes one dedicated OCP 3.0 network adapter slot and multiple PCIe expansion slots, supporting high-speed network adapters (200/400Gb) and GPU accelerator cards. This ensures optimal collaboration between computing, storage, and network performance, providing a flexible and efficient intelligent computing solution for AI training, inference, and high-performance computing (HPC) applications.