Microsoft's DirectX Raytracing Trick Could Slash VRAM Use By Up To 90% - HotHardware

Microsoft has introduced what it calls "the most significant update to DXR since its debut in 2018," bringing performance optimizations that could revolutionize real-time ray-tracing in gaming. The new features standardize technologies previously developed by hardware vendors, making them accessible to all game developers regardless of their target hardware.

Key Performance Improvements

The three main additions to DirectX Raytracing are Clustered Geometry (CLAS), Partitioned TLAS (PTLAS), and GPU-driven acceleration structure builds. These features work together to optimize how acceleration structures—smart approximations of 3D scenes—are managed, dramatically reducing the computational workload required for ray-triangle intersection tests.

CLAS accelerates these tests by using triangle clustering, while ExecuteIndirectRTASOperations() allows the GPU to handle batches of operations directly from GPU memory, bypassing the CPU. PTLAS is particularly valuable for open-world games, as it enables engines to update only affected scene partitions when objects move, leaving static portions intact.

Benchmark Results and Real-World Impact

Testing on a Radeon RX 9070 XT, Microsoft demonstrated a 7.83x speed-up in per-frame GPU ray-tracing performance and a 89.7% reduction in video memory usage for acceleration structures. While the company acknowledges that real games won't see gains of this magnitude, the improvements are still substantial and could significantly enhance the visual fidelity of ray-traced graphics.

Future Availability

These technologies are already supported on GPU hardware from all three major vendors, with Qualcomm's Adreno GPUs expected to receive support next year. The widespread hardware compatibility suggests these optimizations could become standard in future game development, provided developers implement them in their titles.