Hardware

Apex Neural Rendering beta released with claims of faster frame generation

By One Percent LowsCovered by 5 outlets
★ Add us as a preferred source on Google

TL;DR

Apex Compute released a beta for Apex Neural Rendering, a tool that uses OptiScaler to provide frame generation for NVIDIA, AMD, and Intel GPUs.

Apex Compute has released a beta for Apex Neural Rendering (ANR), a new suite of AI-driven technologies including super resolution, ray reconstruction, and frame generation. The frame generation component, ANR FG, is currently available for download on GitHub and works through OptiScaler in games that support NVIDIA DLSS Frame Generation. Apex Compute says ANR FG supports 2x, 3x, and 4x settings.

The company claims ANR FG provides higher image quality and better performance than NVIDIA DLSS. In tests on an NVIDIA GeForce RTX 5090 at 4K with a 2x setting, Apex Compute reports ANR FG achieved 1.01 ms of latency per frame compared to 1.26 ms for DLSS Frame Generation. Apex Compute also says ANR FG uses 214 MB of GPU memory, while DLSS FG uses 562 MB. The neural network weights for ANR FG are reportedly 0.17 MB.

Apex Compute says ANR FG is faster than XeSS on Intel Arc and faster than FSR 3.1 and 4.0 on an AMD Radeon RX 7800 XT. On a Radeon RX 9060 XT, the company says the performance is comparable to FSR 3.1 and 4.0. The technology is compatible with NVIDIA GeForce RTX 20 series and newer, GeForce GTX 16 series, AMD Radeon RX 5000 series, Steam Deck, Intel Arc A-series and B-series, and Apple M1 or later. Supported operating systems include Windows 10, Windows 11, Linux, SteamOS with Proton, and macOS 26 or later.

The super resolution and ray reconstruction components are still in development.

More on NVIDIAAMDIntel