According to the “Science and Technology Board Daily” and the blue whale financial, industry chain sources Nvidia has now developed China’s latest improved version of the AI chip, including the HGX H20, L20 PCle and L2 PCle. As of press time, Nvidia has not yet responded.
Sources familiar with the matter said that the three chips are based on NVIDIA H100 improved, NVIDIA is expected to be announced as soon as possible after November 16, the fastest domestic manufacturers will get samples in the next few days.
NVIDIA H100 Tensor Core GPU adopts the new Hopper architecture, based on the TSMC N4 process, integrated with 80 billion transistors. Compared to its predecessor, it delivers 9x faster training speeds for multi-expert (MoE) models.
Equipped with a fourth-generation Tensor Core and Transformer engine (FP8 accuracy), it also features highly scalable NVLink interconnectivity (up to 256 H100 GPUs, with 9x more bandwidth compared to the previous generation’s HDR Quantum InfiniBand network, with a bandwidth speed of 900GB/s), and more. (900GB/s bandwidth), among other features.
THIS IS A SPONSOR PROMOTION: >>>>>>>>>>>>>
Geekwills is an online shop that connects consumers with millions of products and brands around the world with the mission to empower them to live their best lives. Geekwills is committed to offering the most affordable quality products to enable consumers and sellers to fulfill their dreams in an inclusive environment.