H3C R5300 G3 G5

H3C R5300 G5 can support various types of artificial intelligence accelerator cards to meet the needs of heterogeneous computing power in different artificial intelligence scenarios.H3C R5300 G5 can not only support HGX A100 4-GPU module, 4 A100s are interconnected through 600GB/s NVLINK; it can also support various types of PCIe AI accelerator cards, up to 8 double-width or 20 single-width accelerator cards. The CPU and GPU mount ratio can support various topology configurations such as 1:4 and 1:8, flexibly adapting to the needs of different artificial intelligence scenarios.
H3C R5300 G5 supports the new Intel® Xeon® Scalable family of processors, which can achieve up to 52% increase in processor floating point performance and 42% increase in core count. 60% bandwidth improvement. The H3C R5300 G5 supports up to 24 2.5/3.5-inch hard drives, 8 of which support NVMe, which can meet the storage space requirements of artificial intelligence massive data. The R5300 G5 can support up to 20 PCIe 4.0 slots and 1 OCP 3.0 dedicated network card slot, and supports 200GE network cards to meet the high-speed demand of artificial intelligence for network bandwidth.

H3C R5300 G3 is a brand-new independent research and development for artificial intelligence and high-performance computing

Generation GPU server, which can accelerate the construction of deep learning models, analyze data and export valuable information at extremely fast speed.

interest. Leverage super computing power to predict future activities, behaviors and trends, and use big data analytics to discover and solve hidden problems in the data.

The H3C R5300 G3 server is a 4U rack server built for accelerated computing. support 2 Intel® Xeon® Scalable processors. Up to 8 double-width GPUs or 20 single-width GPUs can be configured. configure Redundant hot-swappable power supply fans, through technology-leading heat dissipation design, ensure that the server can operate 7×24 hours long.Time stable operation.