The new chip, called the H200, will surpass Nvidia's current flagship chip, the H100. The new chip's main upgrade is higher bandwidth memory, which is one of the most expensive parts of the chip and determines how fast it can process data.
Nvidia's H200 chip will handle larger artificial intelligence systems
Nvidia currently has a monopoly on the AI chip market, supplying chips to OpenAI’s ChatGPT service and a number of AI services that generate human-like responses. Adding more high-bandwidth memory and faster connections to the chip’s processing elements means services like ChatGPT can respond more quickly, according to Reuters.
The H200 has 141 gigabytes (GB) of high-bandwidth memory, up from 80GB in the previous H100 chip. Nvidia did not disclose the memory supplier for the new chip, but Micron Technology said in September that it was working to become a supplier to Nvidia.
Nvidia further revealed that Amazon Web Services, Google Cloud, Microsoft Azure, and Oracle Cloud Infrastructure will be the first cloud service providers to exploit the H200 chip, in addition to professional AI cloud service providers such as CoreWeave, Lambda, and Vultr.
Microsoft launches AI chip
Source link
Comment (0)