The upcoming event, where NVIDIA executives are ready to talk about innovations in the graphics field, is the virtual part of CES 2022, which will begin work this week. In March, the company will traditionally hold a conference for developers, and then the debut of the computing architecture Hopper, which will replace Ampere in the server segment, may take place.
The growing influence of server solutions on the business of NVIDIA allows this company to justify the pronounced “evolutionary distance” of their respective architectures. If, within the Ampere family, products for server and gaming use had the same conventional designation and differed, in addition to functional nuances, also in the manufacturing process, then in the case of the successors, as is commonly believed, both families will be even more separate from the start.
For the server segment, NVIDIA will release 5nm Hopper accelerators, which will try on a spatial “2.5-dimensional” arrangement with the combination of several crystals on a single substrate. According to rumors, gaming representatives of the 5nm generation of NVIDIA GPUs will appear only in the second half of the year under the symbol Ada Lovelace, they will retain a monolithic design, which is more profitable in mass production.
The company has already appointed schedule of speeches of their representatives at the spring session of GTC 2022. On March 21st, the head and founder of NVIDIA Jensen Huang will speak to the audience, and although the content of his speech has not yet been disclosed, it can be assumed that he will tell at least in general terms about the architecture of Hopper … The server segment is very inert in terms of bringing the presented new products to the market, so we will see carriers of the Hopper architecture in existing server systems towards the end of this year. March in this sense is a suitable period for a formal announcement.