2026-11-04 · Fonte：Tom's Hardware
A 147MB WebGPU port brings Nvidia's DLSS 5 neural rendering to the browser, apparently running on macOS and non-Nvidia hardware — though each render takes two seconds.
A developer has demonstrated Nvidia's DLSS 5 neural rendering technology running inside a standard web browser, built on the WebGPU API. The live demo ships as a 147MB browser port and executes the neural rendering pipeline outside the native environment where the technology normally operates. Notably, the port apparently runs on macOS as well, and — per the report — it works on non-Nvidia GPUs, taking DLSS 5 beyond the hardware ecosystem it was built for.
The headline caveat is speed: each render takes two seconds, a figure that places the port far from the real-time performance associated with neural rendering features shipped by GPU vendors. The project is therefore best read as a proof of concept — DLSS 5's neural workload can be packaged, delivered and executed through a browser runtime — even though the current implementation sits well short of interactive speeds.
For the AI compute market, the significance lies in portability. WebGPU, the browser graphics and compute API, brings GPU execution to web applications, and the demo indicates a neural rendering model built for one vendor's hardware can be re-hosted to run on non-Nvidia GPUs and, apparently, on macOS machines. If browser-based delivery of neural graphics workloads matures, it could loosen the coupling between such features and specific vendor silicon — a development worth watching for anyone tracking where AI inference workloads actually run.
Caveats remain. This is a modder's port rather than an official Nvidia release, and the source report frames macOS support only as apparent, based on the demo's behavior. The two-second render time also means the browser version is not a drop-in substitute for native DLSS 5. Still, as a demonstration that a 147MB package can carry a neural rendering model from Nvidia into the browser — across GPU vendors and, seemingly, across operating systems — the project offers a concrete data point on how portable AI-driven graphics workloads are becoming.
Tags: Nvidia · DLSS 5 · WebGPU · Neural Rendering · macOS
萬安算交所 AIXX · AIXX · Exchange verificado · Ir ao mercado