Apex Compute unveils NVIDIA DLSS competitor with support for graphics cards starting from the GTX 16 and RX 5000 - GameGPU
Apex Compute has unveiled Apex Neural Rendering, a cross-platform software suite offering frame generation technologies that outperform NVIDIA's DLSS while supporting a wider range of graphics cards.
The new neural rendering architecture was created from scratch and promises significantly better performance than NVIDIA's solutions. It supports NVIDIA GeForce GTX 16, RTX 20-50 series, AMD Radeon RX 5000-9000 series, Intel Arc A/B-series, and Apple Silicon chips via Metal API, making it accessible to a broad range of hardware.
Performance Advantages
Technical benchmarks show ANR frame generation takes only 1.01 ms in 2X mode on an RTX 5090 at 4K, compared to 1.26 ms for DLSS. In 4X mode, ANR averages 0.61 ms versus 0.82 ms for NVIDIA. The solution also requires significantly less video memory—214 MB at 4K compared to 562 MB for NVIDIA's standard frame generator.
Visual Quality
Comparative footage from 'The Last of Us Part II' demonstrates that Apex Compute's algorithm handles complex geometry, dynamic vegetation, and fast-moving objects more cleanly, minimizing ghosting artifacts around weapons and UI elements.
Players report smooth gameplay at 120 frames per second, coupled with the FSR 4.1.2 upscaler, the absence of noticeable visual artifacts, and low input lag.
Current Status
The beta version of ANR frame generation is now available for download and benchmarking. While Super Resolution scaling and Ray Reconstruction algorithms remain in the neural network training phase, they are expected to be released later as the technology continues to develop.
Related
