AI-assisted port brings DLSS 5 neural rendering to Intel integrated GPUs
A developer used AI agents to port NVIDIA's DLSS 5 neural rendering to Intel's Xe architecture, specifically the Arc 140V integrated graphics. The implementation reimplements the U-Net model and runs via Vulkan, achieving only 10 fps at 360p. The project demonstrates the feasibility of running such models on non-NVIDIA hardware, albeit with poor performance.
The port was accomplished using AI agents that reimplemented the U-Net model at the heart of DLSS 5's neural rendering pipeline. Rather than relying on NVIDIA's proprietary stack, the developers built the implementation to run through the Vulkan graphics API, targeting Intel's Xe architecture specifically the Arc 140V integrated GPU.
Early testing shows the system manages roughly 10 frames per second at 360p resolution, a figure far below playable standards. Still, the project serves as a proof of concept, demonstrating that neural rendering techniques can be adapted to non-NVIDIA hardware even if current performance leaves significant room for improvement.
This development could broaden access to neural rendering technology, which has largely been confined to NVIDIA's ecosystem. If such ports mature, owners of Intel or AMD hardware may eventually benefit from AI-enhanced graphics without purchasing new GPUs. However, the current performance gap suggests widespread adoption remains distant. Game developers and hardware vendors could face pressure to standardize neural rendering approaches, while consumers may gain more choice in how they experience AI-driven visuals.