AMD is reportedly preparing to present a kind of new compression technology during EGSR2024, which is set to kick off next week. That technology is officially known as “Random-Access Neural Compression of Material Textures”, and on paper, should allow neural algorithms to increase resolution by up to four times, compared to Block Compression methods.
The technology is similar to one that was presented by NVIDIA last year, and AMD’s alternative works in the same way. Basically, the idea here is that the technology would use a neural network to compress huge in-game packages, reducing the data size, and run over an unchanged runtime execution. To put that in another way: the technology could theoretically make games load faster.
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
AMD says that its technology would be easy to implement, but didn’t say if it can be implemented at the driver level, or if a specific and specialised code was necessary. Also, the hardware required to execute the compression technology is still unknown.
Leveraging AI and neural networks on AMD’s own chips is par for the course; this entire year and the following years are expected to be completely dedicated to the advancement of artificial intelligence (AI), so little surprise here that the chipmaker is also focusing on it.
(Source: AMD GPUOpen)
