Bounding Volume Hierarchy Compression via Relative Precision Storage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for compressing bounding volume hierarchies (BVH) in graphics processors are inefficient, leading to high memory usage and complex ray tracing implementations due to the need for full precision bounding boxes and extra stack storage during traversal.
Innovation Solution
The proposed solution involves storing a parent bounding box with full precision and child bounding boxes relative to the parent with lower precision, allowing for efficient compression and reduced memory requirements without the need for extra stack storage or re-traversal of the node path during ray tracing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If full precision bounding boxes are used for all nodes in the BVH, then measurement precision is improved, but memory usage increases
Solution Approach 1:
The patent applies local quality by using different precision levels for different parts of the BVH structure. Specifically, ancestor nodes store full precision bounding box data, while descendant nodes use reduced precision (e.g., 8-bit or 16-bit quantized) bounding box data. This allows the system to maintain high measurement precision where it is most needed (in ancestor nodes that are traversed more frequently) while reducing memory usage in descendant nodes that require less precision for intersection testing.
2Measurement precision
If full precision bounding boxes are stored for all nodes, then measurement precision is improved, but device complexity increases due to extra stack storage requirements
Solution Approach 1:
The patent extracts the full precision bounding box data from descendant nodes and stores it only in the ancestor nodes. During ray traversal, when an ancestor node is visited, its full precision bounding box is used to determine intersection. If the ray intersects the ancestor, the traversal continues to descendant nodes which only store reduced precision data. This extraction eliminates the need for descendant nodes to store full precision data, thereby reducing device complexity and stack storage requirements while maintaining measurement precision where it matters most.
3Quantity of substance
If reduced precision is used for child bounding boxes, then memory usage is reduced, but measurement precision deteriorates
Solution Approach 1:
The patent applies preliminary action by performing intersection testing with ancestor nodes using their full precision bounding boxes before proceeding to descendant nodes. This preliminary test filters out rays that do not intersect the ancestor's bounding box, preventing unnecessary traversal into descendant nodes. For rays that do intersect the ancestor, the system then uses the reduced precision descendant bounding boxes, which are sufficient for accurate intersection testing at that level. This two-stage approach ensures measurement precision is maintained for the critical filtering operation while allowing reduced precision for subsequent operations.
Data Source
AI summary
Apparatus and method for compressing an acceleration data structure such as a bounding volume hierarchy (BVH). For example, one embodiment of a graphics processing apparatus comprises: one or more cores to execute graphics instructions including instructions to perform ray tracing operations; and compression circuitry to compress lowest level nodes of a hierarchical acceleration data structure comprising a plurality of hierarchically arranged nodes, each of the lowest level nodes comprising pointers to leaf data; the compression circuitry to quantize the lowest level nodes to generate quantized lowest level nodes and to store each quantized lowest level node and associated leaf data without the pointers to the leaf data.