15 February 2017 in IEEE Transactions on Visualization and Computer Graphics
We propose a novel massively parallel construction algorithm for Bounding Volume Hierarchies (BVHs) based on locally-ordered agglomerative clustering. Our method builds the BVH iteratively from bottom to top by merging a batch of cluster pairs in each iteration. To efficiently find the neighboring clusters, we keep the clusters ordered along the Morton curve. This ordering allows us to identify approximate nearest neighbors very efficiently and in parallel. We implemented our algorithm in CUDA and evaluated it in the context of GPU ray tracing. For complex scenes, our method achieves up to a twofold reduction of build times while providing up to 17 percent faster trace times compared with the state-of-the-art methods.
9 August 2024 in Proceedings of the ACM on Computer Graphics and Interactive Techniques
We propose a novel GPU-oriented approach for constructing binary bounding volume hierarchies (BVHs) based on the parallel locally-ordered clustering (PLOC/PLOC++) algorithm. Compared to competing high-performance GPU BVH build algorithms (PLOC++ or ATRBVH), our method provides similar BVH quality in just a single kernel launch while achieving 1.1-3.6× lower construction times for the entire BVH build and 1.6-13× lower for just the binary BVH construction phase. Additionally, we propose an efficient algorithm to convert a binary BVH to an n-wide BVH with just a single kernel launch. Besides being extremely efficient, our proposed algorithms are simple to implement, allowing easy integration into existing frameworks.
7 May 2016 in The Visual Computer
9 August 2024 in Proceedings of the ACM on Computer Graphics and Interactive Techniques
We present HIPRT, an open-source ray tracing framework in HIP. HIPRT provides a versatile, cross-platform solution for professional rendering on contemporary many-core architectures. The core of the framework relies on the bounding volume hierarchy (BVH) with scalable construction algorithms and efficient ray traversal, employing hardware acceleration on AMD GPUs. From a user perspective, we aim at minimalist and user-friendly API design, allowing a user to write ray tracing applications only in a few lines of code. Unlike other graphics APIs that couple ray tracing and shading together, HIPRT provides only ray tracing functionality and thus can be seamlessly integrated into existing rendering environments. To demonstrate advanced features of HIPRT, we integrated it into the three rendering systems: Blender Cycles, PBRT-v4, and Radeon ProRender.
1 June 2021 in Computers & Graphics
1 October 2021 in Computer Graphics Forum