Ziguang, a Chinese chipmaker has revealed that it is working on the development of 7nm-class network chips with a core count of more than 500 and up to 40 billion transistors. In comparison, NVIDIA’s A100 Tensor core accelerator features 54.2 billion transistors on TSMC’s N7 (7nm) node. Named the H3C ENGIANT 800 series, they are slated to begin trial production by the end of this year, with a launch slated in 2022.
For the time being, the chipmaker is working on the mass production of its ENGIANT 660 chip fabbed on a 16nm-class node with up to 256 cores and 18 billion transistors. The trial production of these parts has been officially completed with a launch expected in the last quarter of the year.
The 800 series is expected to offer a performance boost of over 122% over the 660 series and will form the “ZhiQing” series of network processors for use in high-end routers.