Games too big? AMD has a solution for this

There’s probably no doubt that today’s games are too big. It’s not about the time it takes to play or the size of the worlds, although even in this area it’s sometimes exaggerated (e.g. Assassin’s Creed: Valhalla). It’s about how much space games take up on the disk these days and how long it takes to download them. AMD has developed a solution to partially solve this problem.

AMD Neural Texture Block Compression

This new solution is AMD Neural Texture Block Compression, a texture compression mechanism using neural networks. We will learn the details of the solution at EGSR (Eurographics Symposium on Rendering) in London, which will take place on July 2, just a few days away. The presentation will be led by S. Fujieda and T. Harada from AMD GPUOpen.

We’ll present “Neural Texture Block Compression” @ #EGSR2024 in London.

Nobody likes downloading huge game packages. Our method compresses the texture using a neural network, reducing data size.

Unchanged runtime execution allows easy game integration. https://t.co/gvj1D8bfBf pic.twitter.com/XglpPkdI8D

— AMD GPUOpen (@GPUOpen) June 25, 2024

We don’t know the details yet, but we can deduce a lot just from the name of the new solution. In fact, NVIDIA showed a similar mechanism last year. AMD Neural Texture Block Compression will likely use neural networks to compress game textures to reduce their size and then decompress them without significantly affecting graphical quality.. This would reduce the size of game downloads and perhaps their disk space requirements.

We will learn more on July 2, when the AMD GPUOpen group will present the new solution in practice. However, there is no doubt that neural networks and machine learning will have a huge impact on the future of games. They are already used today to improve the fluidity of gameplay (DLSS and FSR), and soon they may also affect other issues, including generating textures or NPC characters.