DirectX Raytracing accelerated Radeon RX 9070 from 1.8 FPS to 84 FPS in Microsoft benchmarks - GameGPU
Microsoft has released a significant update to the DirectX Raytracing standard, introducing the DirectX 12 Agility SDK 1.722 Preview, which demonstrates a massive leap in performance on the Radeon RX 9070.
The update, included in the DirectX 12 Agility SDK 1.722 Preview, introduces Clustered Geometry, Partitioned TLAS, and GPU-based structure generation to eliminate bottlenecks in rendering dynamic worlds.
These innovations allow engines to update only deformed polygons or specific sectors of space, rather than rebuilding entire global spatial hierarchies.
Clustered Geometry segments source geometry into independent clusters of up to 256 triangles and 256 vertices, generating intermediate CLAS blocks.
Performance gains on Radeon RX 9070
In a specialized technical demonstration on the Radeon RX 9070, the frame rate jumped from the initial 1.8 FPS to 84 FPS.
The amount of video memory allocated to acceleration structures dropped from 12.7 GB to 774 MB.
Microsoft engineers emphasize that these measurements were obtained in an isolated synthetic DXR benchmark and serve as a demonstration of the removal of system limitations, so in commercial games the average performance gain will be more modest.
The toolkit has received basic hardware support among leading manufacturers of modern GPUs.
Compatibility with major GPU manufacturers
The new algorithms are officially available on all generations of NVIDIA GeForce RTX series graphics cards, AMD Radeon RX 9000, as well as Intel Arc B-series graphics accelerators and supported integrated cores of Intel Core processors Ultra.
Related
