If we recapitulate everything we have seen in the first 6 months of this year, we will see that the price forecasts for the new GPUs are not at all optimistic. The world situation is not going to help either, so after mistreating the average consumer and gamer, companies like NVIDIA and AMD are going to have to use their ingenuity to sell. In the case of Huang’s, they have a strategy that we are going to call RTX+DLSSbut what’s new with this in NVIDIA?
When the RTX 20 came out, the world marveled at what NVIDIA had just achieved: AI for resolution scaling and more realistic, real-time computed shadows, reflections, and other artifacts. That was the basis that in a few months the company will use to argue the prices and, together with the performance, will be the tricks to convince the users along with some small novelties that they present.
NVIDIA, RTX+DLSS, old made new?
NVIDIA not only wants to sell graphics based on performance well above the existing one, it seems that it will not be enough and for this reason it has prepared several arguments that have to do with DLSS and RTX. They are the two cutting-edge technologies together with Image Sharpening that are dominating the gaming sector and although we don’t know exactly how they are going to power them by hardware, we do know small details that have been leaked and others that have already been revealed.
As we can see in the SM of the AD102, NVIDIA is going to promote important changes in the RT Core Y Core Tensionerto the point that he calls them 4th Generation and 3rd Generation respectively. We do not know the changes that are going to be introduced because it is confidential information, but we do have some brushstrokes as a result of the changes in DLSS and the rendering engines.
We went from 2 motors + Tensor Core to 3 engines with said 4th generation units (4 motors in total), which implies much more muscle. This has been thought with one goal: a more intelligent use of movement vectors in each game, providing not only more FPS in AI scaling, but also eliminating ghosting and producing greater sharpness.
Pipeline integration of “pure” Ray Tracing
We will not explain how NVIDIA is going to use RTX with DLSS together and scaled in this new generation, mainly because, although we know how it works so far, the changes seem to be going towards purer Ray Tracing.
To do this, the BVH algorithm has to be optimized to traverse the entire tree more efficiently so that the result is produced before the shaders start their phase within the pipeline.
In other words, we try to make a purer complete rendering, more integrated within it, where the Shading phase is executed in a simpler way, obtaining the data of the intersections with greater speed and haste.
Therefore, the NVIDIA tandem will be RTX + DLSS with a series of optimizations that will be focused in general terms on three points:
- improve the details of the scenes with more movement.
- Improved temporal stability.
- Higher performance in rendering, reflection and shadows of the textures and particles.
If we add to this the fact that the performance will increase to almost double, if not double itself, what we will have will be a greater jump in FPS than expected with a clear objective: Ray Tracing and DLSS (or equivalent technology). if it happens by hardware) they have to be a standard and not settings that have to be activated to enhance performance. This is how NVIDIA plans to sell the RTX 40 graphics cards, a very interesting technological leap that will be masked with arguments of various kinds in its official presentation, where the performance will be impressive, yes, but the rest will be more so.