Intel XeSS 2 Explained: How It Compares to DLSS and FSR

Quick Answer

Intel XeSS 2 is Intel''s second-generation AI-based upscaling technology that renders games at a lower internal resolution and reconstructs a higher-resolution output to improve frame rates. Compared to NVIDIA DLSS and AMD FSR, XeSS 2 is competitive in image quality on Intel Arc hardware and continues to be available on non-Intel GPUs via a fallback path, though it shows its strongest performance on Intel''s own Xe-based graphics architecture.

Upscaling has become a standard tool in the PC gaming toolkit. Where once you chose between native resolution and lower resolution, today you choose between several AI-assisted reconstruction technologies that each promise the visual quality of a higher resolution with the performance of a lower one. XeSS 2 is Intel''s answer - and its position in the market is more nuanced than simply ''the Intel option''.

How XeSS 2 Works

XeSS (Xe Super Sampling) 2 uses a machine learning model to reconstruct a high-resolution image from a lower-resolution input frame, using motion vectors and temporal data from previous frames to inform the reconstruction. On Intel Arc GPUs, XeSS 2 runs its inference on dedicated XMX (Xe Matrix Extension) AI hardware, which is purpose-built for the matrix math that neural upscaling requires - similar to how DLSS uses Tensor Cores on NVIDIA RTX GPUs.

On non-Intel hardware - including AMD and NVIDIA GPUs - XeSS 2 falls back to a DP4a (dot product) path that uses standard shader units rather than dedicated AI hardware. This fallback mode is still functional and produces good results, but it is slower and image quality is slightly below what XeSS 2 achieves on native Intel Arc hardware with XMX acceleration.

XeSS 2 vs DLSS vs FSR: The Key Differences

DLSS 3 and 4 (NVIDIA) runs exclusively on RTX hardware and uses Tensor Cores for inference and Optical Flow hardware for Frame Generation. It generally delivers the highest image quality in its Quality mode among the three technologies, with Frame Generation providing substantial fps boosts in supported titles. The limitation is hardware exclusivity - you need an RTX GPU.

FSR 4 (AMD) takes a different philosophical approach: it is designed to run on any GPU, making it the most broadly accessible upscaling solution. Image quality in FSR 4 has improved substantially over FSR 1 and 2, and AMD''s open-source approach has encouraged wide game developer adoption. On RDNA 3 and RDNA 4 hardware, FSR 4 uses machine learning acceleration; on other hardware, it uses an optimised spatial reconstruction algorithm.

XeSS 2 sits between these: hardware-accelerated on Intel Arc, available via fallback on all other GPUs, with image quality that competes with DLSS Quality on Intel hardware and is comparable to FSR 4 on other hardware. For South African gamers on an Intel Arc GPU, XeSS 2 is the natural first choice. For those on NVIDIA RTX, DLSS remains the premium option. For AMD GPU owners, FSR 4 is the optimised path.

Frequently Asked Questions

Q: Can I use XeSS 2 on an NVIDIA or AMD GPU? A: Yes, through the DP4a fallback path. Image quality and performance are lower than on Intel Arc hardware with XMX acceleration, but XeSS 2 remains usable and produces better results than no upscaling.

Q: Which upscaling technology is best for competitive gaming? A: Performance mode in any upscaling technology - DLSS, FSR, or XeSS - targets the highest fps at some image quality cost. For competitive play, DLSS Performance on RTX hardware typically delivers the best frame rate with the least blurring. All three are viable depending on your GPU.

Q: Do all games support XeSS 2? A: No. XeSS 2 requires developer integration. As of mid-2026, a growing list of titles supports XeSS 2, but adoption is broader for DLSS and FSR due to their longer market presence. Check whether your specific titles support XeSS 2 before making it a purchasing factor.