ZTE has unveiled its first "flagship GPU server" for China, featuring NVIDIA's H800 AI GPUs, coupled with Intel's Xeon Scalable CPUs.
ZTE Plans On Stepping In The AI Bandwagon As Well, With Its Flagship GPU Server Featuring 8 NVIDIA H800 GPUs
AI developments in China have been tremendously growing in recent days, with every other tech firm stepping into the game with their own AI "solutions". ZTE, a firm that is known for its server, Smartphones, Networking, & enterprise solutions, has decided to make an entry in the AI industry, debuting its very own R6900 G5 flagship GPU server.
ZTE’s flagship GPU server R6900 G5 was unveiled at #2023China Mobile Global Partner Conference# ] Recently, ZTE’s flagship GPU server R6900 G5, specially designed for large-scale model training, was unveiled at the China Mobile Global Partner Conference, with excellent computing performance, high-speed network communication capabilities and innovative energy efficiency performance have brought new breakthroughs in fields such as artificial intelligence and deep learning.
Diving into specifications, the ZTE R6900 G5 server features NVIDIA's H800 AI GPUs. The NVIDIA H800 GPUs are designed specifically for China, in an attempt to work around the sanctions placed by the US government. Compared to the mainstream H100s, NVIDIA's H800 comes with degraded specifications, with slacks made within the chip-to-chip interconnect speed.
While ZTE themselves haven't disclosed the specific performance data for their flagship GPU server, it is disclosed that the R6900 G5 will pack eight H800 NVLINK GPU modules, totaling up to 297 TFLOPs of FP64 compute power. ZTE hasn't disclosed what model of Intel's Xeon Scalable processor is being featured, but it will indeed be in the higher-end range.
Speaking of networking capabilities, the server features RoCE networking between individual nodes, giving 4.8 TB/s of non-blocking bandwidth. On paper, the server looks like a great addition to ZTE's arsenal for tech development, and will definitely aid in catering to the high computing demand from the industry.
With such high-end components packed together, having an adequate cooling solution is a huge task to fulfill, and by the looks of it, ZTE has been successful. It is said that the R6900 G5 GPU server features fan adjustment capabilities as well as a "dual-cooling" liquid technology, which will provide optimal heat dissipation for both the CPU and GPU inside. Moreover, the server has the capability to reduce the power consumption by a whopping 1000W as well, demonstrating a power-effective design.
Details such as the cost of the server machine and the overall performance figures are missing for now, however, we will try to dig them out, since they will indeed be something to look at.
News Source: ITHome
About the author: Muhammad Zuhair is a hardware and technology reporter for Wccftech, specializing in the semiconductor industry and the complex interplay between technology, manufacturing, and geopolitics. His coverage focuses on the corporate strategies and technological roadmaps of industry giants like TSMC, NVIDIA, Samsung, and Intel. Zuhair's expertise lies in deconstructing complex topics such as fabrication nodes (e.g., 2nm process), the economic impact of policies like the CHIPS Act, and the strategic development of AI infrastructure from NVIDIA, AMD and Intel.
Follow Wccftech on Google to get more of our news coverage in your feeds.
