Games that eat up your GPU’s memory have become a problem, and leading manufacturers have a plan to tackle it. AMD and Sony have given us a glimpse into the future of gaming hardware with a new set of announcements. AMD’s Jack Huynh has been speaking with PS4 and PS5 architect Mark Cerny as part of Project Amethyst, a multi-year collaboration between the two tech giants designed to improve the PC and console gaming experience.

One aspect of the project that caught our eye is the focus on GPU memory – bandwidth usage in particular. As modern AAA games continue to get increasingly demanding on your memory (many gamers now believe that 8GB of VRAM isn’t enough anymore), a way to optimize this on the hardware side of things is desperately needed.

Universal Compression will optimize memory bandwidth usage

In a very scripted conversation, Cerny and Huynh take turns to talk about some of the goals of Project Amethyst, with some lovely AMD and PlayStation hardware decor as a backdrop. As you’d expect, AI is at the heart of many of these improvements. The pair discusses Neural Arrays, Radiance Cores, and Universal Compression.

Universal Compression – a system that evaluates every piece of data headed to memory, not just textures, and compresses it wherever possible. Only the essential bytes are sent, dramatically reducing memory bandwidth usage. This means the GPU can deliver more detail, higher frame rates, and greater efficiency.​

Source: Jack Huynh, SVP & GM, Computing & Graphics at AMD

GPU memory bandwidth limitations are recognized as a bottleneck for modern gaming, especially when it comes to high-fidelity 4K textures, ray tracing & path tracing, and other machine learning-powered tech. These “next-gen rendering techniques” require “significantly more bandwidth to handle 4K+ textures and ray tracing denoising maps for smooth asset streaming,” says Huynh.

The new texture compression method is designed to replace DCC (Delta Color Compression). As Cerny points out, “with current GPUs, including the ones in PlayStation 5 and PlayStation 5 Pro, we have something called DCC”. Essentially, what this does is reduce memory bandwidth consumption when your GPU reads and writes data such as textures or ‘render targets’.

Future GPUs and SoCs will instead be upgraded to the new Universal Compression method, taking data compression “much further,” says Huynh. It does this by analyzing every piece of data handled by your GPU’s memory – not just textures – compressing the data “whenever possible”. This is set to dramatically reduce memory bandwidth usage and, in effect, lower the load on your GPU.

This should mean that the effective memory bandwidth of your GPU should go beyond what we currently expect from the ‘paper spec’. Mark Cerny points out that these latest developments only exist in simulation right now – it’s “still very early days for these technologies,” but it looks like we’re definitely moving in the right direction. Still, we wouldn’t complain about more VRAM and higher bandwidth crammed into the physical hardware on top of these improvements.

About the Author

Jack Goodall

At PC Guide, Jack is mostly responsible for reporting on hardware deals. He also specializes in monitors, TVs, and headsets and can be found putting his findings together in a review or best-of guide.