AMD has demonstrated how the use of DGF can shape the way GPUs perform modern-day animations and ray tracing, with possible plans to integrate hardware-level capabilities into future GPUs.
AMD's Next-Gen UDNA GPUs Could Potentially Feature Much Better RT & Animation Performance, With 'Hardware-Level' DGF Support
It's often exciting to learn about future technologies by diving into technical details, as we did with AMD's Zen 6 CPU post and its new D2D communication method. Now, in a dedicated GPUOpen blog post by Team Red, they have detailed animation support with DGF (Dense Geometry Format). If you are confused about what this actually means, we'll discuss it in depth ahead. However, just for reference, DGF will allow 'future RDNA GPUs' to use less memory bandwidth and build RT acceleration structures directly from DGF blocks, leading towards a more efficient RT performance.
Let's talk about what DGF actually is. Normally, GPUs are responsible for processing animation geometry, but with the Dense Geometry Format, you get a compressed container for geometry that’s designed to be GPU-friendly. This is achieved by 'chopping up' large triangle meshes into blocks of meshlets, and then storing the data of each block in a local dense format. Once the DGF 'base block' is created, and if there's a need to animate it, instead of unpacking the whole block, AMD's DGF technique utilizes a per-frame compute shader and re-quantization to use the compressed block by updating data.
For RT in particular, DGF reduces the overhead associated with rebuilding BVHs (Bounding Volume Hierarchies) since the GPU understands the DGF block itself, which in turn reduces the resources required in ray tracing pipelines and ultimately benefits performance. As mentioned above, DGF currently happens on AMD's compute shader units, but with AMD's next-gen UDNA GPUs, it could actually shift to fixed-function hardware units for faster animation. Moreover, DGF compression requires significantly less resource overhead, allowing more geometry to fit inside a GPU cache and ultimately leading to lower latency and higher performance.
Of course, DGF is one of the elements around which next-gen UDNA GPUs could see increased RT or animation performance, and while these don't look much significant on first glance, they do contribute towards faster animation times, that too without being resource-intensive.
News Source: Videocardz
About the author: Muhammad Zuhair is a hardware and technology reporter for Wccftech, specializing in the semiconductor industry and the complex interplay between technology, manufacturing, and geopolitics. His coverage focuses on the corporate strategies and technological roadmaps of industry giants like TSMC, NVIDIA, Samsung, and Intel. Zuhair's expertise lies in deconstructing complex topics such as fabrication nodes (e.g., 2nm process), the economic impact of policies like the CHIPS Act, and the strategic development of AI infrastructure from NVIDIA, AMD and Intel.
Follow Wccftech on Google to get more of our news coverage in your feeds.
