As far as we know, DLSS is NVIDIA's exclusive technology that is able to improve the value of the framework in supportive games due to artificial intelligence and large sample.
The findings are a series of high quality images without the usual teeth of a saw, because the image resolution has increased to 64 times. This process was performed using a supercomputer Saturn v from NVIDIA itself, so every game that requires AI training has to go through it.
This, as well as other problems or limitations such as resolution, or lack of sharpness due to several differences, has left the DLSS very corrupt to this day. And that NVIDIA is returning to load with the DLSS 2.0 version, which has been much improved as we will see below.
DLSS 2.0: AI version due to new algorithm
NVIDIA will redefine its efforts to the power of its AI algorithm, which has been completely analyzed and enabled. The company will therefore focus on four key areas: translation, physics, photography and broadcasting.
These four components are the backbone of a new deep learning neural network and the ultimate goal remains to maximize FPS while producing sharp images for each game.
To do this, NVIDIA removes many obstacles from the original DLSS, where decisions were limited and firmness at low scales was a problem due to the low number of pixels on the screen for the AI to work properly.
The neural network is now much faster and slower, which will facilitate the implementation of this technology in any existing game. To accomplish this, NVIDIA relies on a call Temporary Response, something we already saw last week with the announcement of the DX 12 Ultimate.
What is intended to be distinguished from the already known zero values is that the data is sharper than the initial decision scale and, and improves the resilience of the structures. That is, when we scale from 1080p to 4K, NVIDIA aims to achieve much higher image quality than if we have a traditional 4K monitor and will also try to achieve better performance than we have given that decision.
An algorithm development and three types of rendering choices
The new algorithm will also use Tensor Cores to accomplish this process. Those of Huang ensure that the new artificial intelligence network of their invoice algorithm works well and will be able to do twice as much as the original version, that is, perform double-digit calculations faster.
This will help push the FPS level higher than DLSS 1.0. As we mentioned, this upgrade will be included in a large number of titles, because DLSS 2.0 is trained to use unspecified content in the game in question.
So, it works with all types of games, which would not be possible for Saturn V to enjoy the development of this technology. In addition, DLSS 2.0 will now provide us with three modes of image quality: Moderation, Quality and Performance, all of which control the internal resolution of the game.
The first titles to get these enhancements with DLSS 2.0 will be Wolfenstein: Youngblood and Delanie for us a month, which is available with its compatible platform from now on, and Command and Mechwarrrior 5 will have it this week.
Inreal Engine will soon be supported in a consistent manner.
More articles coming soon
In addition, other titles will add support to the next NVIDIA driver for their corresponding position in March, so the number of titles will increase exponentially.
With DLSS 2.0, NVIDIA closes the circle before the launch of its new facilities, where both Train and Ampere will be able to enjoy this new development in Ray Tracing and Artificial Intelligence.
Obviously, this excludes the notion that it was being circulated on the Internet that DLSS would now be speeded up by CUDA Cores, as this would make no sense. Censor Cores units based systolic setting They have higher performance than CUDA Cores because of their similarity to systematic functionality, something that has allowed NVIDIA to use them both in its Volta and Turing architecture for different functions.
Now NVIDIA is taking the appropriate step to leverage AI power across its RTX GPUs and large number of games, which AMD, meanwhile, has not been able to respond to and RDNA 2 may not be bringing any improvements like DLSS 2.0 in this case. .
It will be interesting to see how Lisa Su does in this NVIDIA move.