PROCESSOR AND METHOD FOR PAST-HIT-DRIVEN TRACE SEQUENCE FOR BEAM TRACKING
The GPU with programmable units and SIMT architecture optimizes ray tracing in complex scenes by dynamically distributing workloads and leveraging ray coherence, addressing performance bottlenecks and inefficiencies in BVH traversal.
Patent Information
- Application Number
- DE102025147456
- Authority / Receiving Office
- DE · DE
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2025-01-03
- Filing Date
- 2025-11-17
- Publication Date
- 2026-07-09
AI Technical Summary
Ray tracing in complex scenes with millions of geometric primitives faces performance bottlenecks due to excessive intersection tests and inefficient traversal of Bounding Volume Hierarchies (BVH) trees, lacking adaptability and leveraging spatial and temporal coherence of rays for optimization.
Implementing a graphics processing unit (GPU) with programmable functional units and SIMT architecture, utilizing a scheduler for dynamic workload distribution across processing clusters, and employing a memory matrix for efficient data transfer and processing.
Enhances ray tracing performance by optimizing traversal through adaptive scene analysis and leveraging ray coherence, reducing redundant computations and improving frame rates in real-time applications.
Smart Images

Figure 00000000_0000_ABST
