Microsoft has released a second DirectX Ray Tracing functional specification that details the expected shape of its ray tracing pipeline, the goals behind the design, and how key parts of the technology work behind the scenes. The earlier specification covered the ray tracing pipeline from ray shader generation and scheduling through acceleration structures and final shading. The new material focuses on clustered geometry, partitioned top-level acceleration structures, and indirect acceleration structure operations.
Clustered geometry is presented as a way to treat groups of nearby triangles as shared building blocks instead of handling triangles individually. That lets the GPU build, move, and instantiate geometry in bulk rather than issuing multiple separate calls for triangle data. Microsoft also defines compact vertex encodings and predefined template formats to keep the GPU memory and bandwidth needed for bulk geometry building and movement under control. The approach is designed to avoid updating or duplicating existing geometry and to make it easier to render foliage, crowds, and in-game props once and then move them around efficiently.
The practical effect is a lighter workload for the GPU and better ray tracing performance in games. By reducing repeated geometry handling and shifting more work into bulk operations, the specification points to a more efficient rendering path for scenes filled with many similar or reusable objects. The broader update also signals that Microsoft is refining the structure and management of ray tracing data through partitioned top-level acceleration structures and GPU-driven acceleration structure operations, with the overall aim of making the pipeline more scalable and performant.
