Next week we’re going to see AMD showcase a new piece of technology, to them at least. As AI tech progresses further and further, Neural Texture Block Compression seems to be AMD’s next step in the process, a compression method that helps reduce data size.
This isn’t the first of its kind though, as Nvidia revealed something similar last year. However, with AMD promising “easy game integration”, they seem to be pushing for this kind of technology to actually become commonplace. Hopefully, this will help tackle the growing problem of ridiculously large game file sizes.
Neural Texture Block Compression will be presented next week
Two representatives from AMD, S. Fujieda and T. Harada, are listed on the EGSR 2024 program – the event begins on July 2nd in London, with this particular presentation scheduled for 3:30 – 3:45 on that date. As rightly said in this post on X, “Nobody likes downloading huge game packages”, so it looks like this texture compression is a promising sign for developers and gamers that will help cut down the size of modern games.
Easy implementation would be a winner versus Nvidia
If it is truly easy to implement, then it will be a turning point – especially given that Nvidia has yet to reach the potential of this technology in a widespread way. If AMD can beat them to it, we suspect that open-sourcing the code in the same way as FSR will greatly appeal to developers. According to AMD, “Unchanged runtime execution” is what will allow for easy integration for game developers.
While no games currently make use of Nvidia’s compression tech, VideoCardz recently reported that Nvidia is indeed putting some focus on AI-generated textures in future games. It looks like this could be the next big thing in optimization, coming just short of a year since Nvidia first showed off Neural Compression, seemingly leaving it on the backburner since then.
