Sources say NVIDIA is developing China-specific versions of AI chips HGX H20, L20 PCle and L2 PCle

The latest news shows that according to reports from Science and Technology Innovation Board Daily and Blue Whale Finance, industry chain sources revealed that Nvidia has developed the latest version of AI chips suitable for the Chinese market, including HGX H20, L20 PCle and L2 PCle. As of now, NVIDIA has not commented.

People familiar with the matter said that these three chips are based on NVIDIA H100 improvements, and NVIDIA expects to release them as soon as possible. It will be announced after November 16, and domestic manufacturers will get samples as soon as these days.

After checking public information, we learned that NVIDIA H100 Tensor Core GPU adopts a new Hopper architecture, based on TSMC N4 process, integrating 80 billion transistors . Compared with the previous generation product, it can provide up to 9 times the training speed for multi-expert (MoE) models

It is equipped with the fourth generation Tensor Core and Transformer engine (FP8 accuracy), and also has highly scalable NVLink Interconnect technology (up to 256 H100 GPUs can be connected, compared to the previous generation using HDR Quantum InfiniBand network, the bandwidth is 9 times higher, and the bandwidth speed is 900GB/s) and other functions.

Advertising statement: The external jump links contained in the article (including but not limited to hyperlinks, QR codes, passwords, etc.) are used for Convey more information and save selection time. The results are for reference only. All articles on this site contain this statement.

The above is the detailed content of Sources say NVIDIA is developing China-specific versions of AI chips HGX H20, L20 PCle and L2 PCle. For more information, please follow other related articles on the PHP Chinese website!