NVIDIA’s DLSS technology has become the innovation in the value of its RTX 20 and RTX 30, as it is not found in competitive products and is a huge advantage for an NVIDIA that is no longer a specialized company only on graphics cards and therefore GPUs, to focus on AI.
Both AMD and Intel, especially the latter, do not want to be left behind, but NVIDIA’s years of advantage have allowed them to deploy technologies such as DLSS in games. Where version 2.0 is used in several leading games, which makes AMD and Intel have to look for an answer to NVIDIA’s DLSS.
Intel shows interest in FidelityFX in Intel Xe
AMD contrary to NVIDIA has decided to launch its library collection FidelityFX in open source, which means that anyone can adopt them for their GPUs, including Intel and NVIDIA. While Jen-Hsun Huang’s company wants to keep the DLSS as an undocumented black box and exclusive to its graphics cards.
This has piqued Intel’s interest, albeit with nuances, since the words of the chief architect of the different flavors of Intel Xe architecture, Raja Koduri, are a combination of feigned interest and push towards his previous company.
We are definitely looking into it, capabilities for Deep Learning Xe-HPG architecture enable solutions that can achieve better quality and performance.
We will definitely try to align ourselves with open solutions to make the job of independent software vendors easier.
It should be noted that one of AMD’s weaknesses compared to NVIDIA is the lack of specialized units for AI in the entire portfolio of the company. While Intel has shown interest and has even developed various technologies.
Of the Intel Xe-HPG we do not know if they will have support for AI, but Koduri’s words suggest that Intel would have added units to the Tensor Core style in its Intel Xe-HPG. Which allow much more accurate solutions than AMD FSR, since these units were advanced by Raja Koduri himself in several presentations on the Intel Xe architecture.
FidelityFX’s “open” nature demonstrates AMD’s shortcomings
It’s no secret that AMD does not have the resources of Intel and NVIDIA at this time, and by this we mean financial and personnel resources. For example, AMD cannot have a team of people specialized in supporting DLSS as if NVIDIA does, hence launching FidelityFX as open source, which allows other manufacturers to adopt it.
The idea of open source in FidelityFX helps AMD to pass the ball to the roof of software developers. With the excuse that they can freely adopt such technologies and manipulate them as necessary for their programs. All this product of an AMD with less resources compared to Intel and NVIDIA.