NVIDIA has a major product refresh pending which, if all goes according to plan, should happen tomorrow. There is still nothing official, but from what we have seen in the latest leaks, it seems that we will see new GeForce RTX 30 productsfor both desktop and laptop.
Apart from those products for general consumption, which as many of our readers will know also have a place in the hands of professionals thanks to the support of NVIDIA Studio drivers, we know that the green giant plans to celebrate the GTC 2022 as of March 21, and that this event is planned as face-to-face.
How could it be otherwise, the CEO of NVIDIA, Jensen Huang, will be in charge of opening the GTC of 2022, but what do we hope to see at that event? Without a doubt, the most important thing could be the presentation of Hopper, a new GPU architecture that will be used in the new high-performance graphics accelerators that will compete with the AMD Instinct MI200.
Hopper will therefore be an architecture directed, and limited, to the professional sector, and will coexist with Ada Lovelace, which will be a next-generation architecture from NVIDIA designed for the general consumer market. This approach is not new, we already saw something similar when the American company limited the Volta architecture, the first to have tensor cores, to the professional sector.
Ada Lovelace will maintain a monolithic core design, and in general everything seems to indicate that it will have the same base that we have seen in Ampere. This means that the specialized hardware division between raster engines, shaders, tensor cores and RT cores will be maintained. This architecture will be used in the GeForce RTX 40 series, and in the new RTX AXXX, formerly known as Quadro.
By cons, the Hopper architecture will be NVIDIA’s first to feature a multi-chip design, and it will be used in A-series graphics accelerators, formerly known as Tesla. This means that they will combine two or more chips to create a super GPU, as we told you at the time in this article. We do not yet have specific details on the configurations that the chip giant could use, but we hope to get out of doubts in the coming months.