Microsoft's new DirectX ray tracing features, like clustered geometry and partitioned TLAS, improve GPU efficiency and speed up rendering in games. Learn how these innovations reduce CPU load.
© Microsoft
Microsoft has unveiled new ray tracing capabilities in DirectX that significantly enhance graphics performance in games. The key innovations include clustered geometry, partitioned TLAS, and GPU-based indirect acceleration operations. These features enable GPUs to work more efficiently by processing groups of triangles and scene elements without constant reliance on the CPU.
Clustered geometry combines adjacent triangles into shared blocks, simplifying the creation, movement, and rendering of large volumes of objects. As a result, the GPU doesn't need to update each vertex individually, and objects like foliage, crowds, or in-game details can be rendered once and easily moved within the scene. This reduces the load on the graphics processor and speeds up ray tracing.
Partitioned TLAS divides the game environment into separate blocks that the GPU can process independently. In large open worlds, this means only the necessary scene elements are traced, rather than the entire world at once. This approach cuts down on unnecessary computations, boosts rendering speed, and optimizes handling of dynamic content.
The new indirect acceleration structure operations shift some computations from the CPU to the GPU. Creating, moving, and compressing templates is now handled directly by the graphics processor, reducing latency and increasing data processing parallelism. These changes are particularly noticeable in complex scenes with numerous objects and dynamic lighting.
Overall, the next-generation DirectX Ray Tracing updates make ray tracing more efficient and accessible for modern games. Game engines will be able to leverage new GPU capabilities to improve graphics quality while reducing CPU load and accelerating the processing of complex scenes, which is crucial for realistic worlds and large open levels.