Some companies will never learn their lesson. Wasted money and time.
ZTE’s new flagship GPU server to tackle AI is equipped with NVIDIA H800
ZTE, which is committed to the development of servers, mobile phones, networks and enterprise solutions, has recently taken advantage of the AI craze and released a flagship GPU server called "R6900 G5". This server, combined with NVIDIAH800 AI GPU and Intel Xeon Scalable CPU, is expected to bring new breakthroughs in the development of AI and deep learning.
According to Wccftech reports, ZTE’s flagship GPU server R6900 G5 released at the China Mobile Global Partner Conference is specially designed for large-scale model training and has excellent computing performance, high-speed network communication capabilities, and innovative energy efficiency performance.
The R6900 G5 is equipped with the H800 GPU designed for the domestic market to circumvent the US ban. Compared with the mainstream H100, the H800 has lower specifications and slower interconnect speeds between chips.
R6900 G5 will be equipped with 8 H800 NVLink GPU modules, which can provide 297TFLOP of FP64 computing power. In addition, the R6900 G5 supports RoCE network technology between nodes and a non-blocking bandwidth of 4.8TB per second.
ZTE’s new flagship GPU server to tackle AI is equipped with NVIDIA H800
ZTE, which is committed to the development of servers, mobile phones, networks and enterprise solutions, has recently taken advantage of the AI craze and released a flagship GPU server called "R6900 G5". This server, combined with NVIDIAH800 AI GPU and Intel Xeon Scalable CPU, is expected to bring new breakthroughs in the development of AI and deep learning.
According to Wccftech reports, ZTE’s flagship GPU server R6900 G5 released at the China Mobile Global Partner Conference is specially designed for large-scale model training and has excellent computing performance, high-speed network communication capabilities, and innovative energy efficiency performance.
The R6900 G5 is equipped with the H800 GPU designed for the domestic market to circumvent the US ban. Compared with the mainstream H100, the H800 has lower specifications and slower interconnect speeds between chips.
R6900 G5 will be equipped with 8 H800 NVLink GPU modules, which can provide 297TFLOP of FP64 computing power. In addition, the R6900 G5 supports RoCE network technology between nodes and a non-blocking bandwidth of 4.8TB per second.