Details about Intel’s XeSS. It is a competition for DLSS and FSR

Intel, when disclosing details about its ARC graphics cards, revealed that it was preparing a response to NVIDIA’s DLSS and AMD’s FSR. Today, thanks to an interview between Wccftech and Karthik Vaidyanathan, the company’s engineer who deals with this technology, we got to know the details about Intel’s XeSS.

The new interview provided a number of details about Intel’s XeSS

Most importantly, Intel’s real-time XeSS super-sampling does not require any in-game training. This technology is to adopt a unified library a unified library that will be compatible with different titles at the same time, which can be compared to the current version of DLSS 2.0, where libraries can be freely transferred between compatible games.

Also read: The new revision of PlayStation 5 has made its debut. How is it different from the original?

According to the engineer, XeSS will be more common than NVIDIA DLSS, because its compatibility will be similar to that known from AMD’s FSR. It will be available in two variants, a variant based on the acceleration of XMX (artificial intelligence) units, i.e. those in Intel Arc graphics processors and those on DP4a, i.e. the one supported by the NVIDIA Pascal, Turing and AMD RDNA1 / 2 architecture. However, the Tensor cores in GeForce RTX cards will not be used.

Also read: Huawei patented a smartphone with an unusual curved screen

In the case of the latter, XeSS will naturally underperform, rendering frames slower and therefore less fluid. However, what is important and which will affect the universality of the technology, the difference between these variants will not mean that developers will have to take care of both implementations, because these will not be required. There will also be many XeSS operation modes (quality / smoothness), and Intel will expand technologies in the future, which will most likely lead to the premiere of XeSS 2.0 and 3.0 using open-source licenses.