Shallow Bounding Volume Hierarchies for Fast SIMD Ray Tracing of Incoherent Rays

Abstract

Photorealistic image synthesis is a computationally demanding task that relies on ray tracing for the evaluation of integrals. Rendering time is dominated by tracing long paths that are very incoherent by construction. We therefore investigate the use of SIMD instructions to accelerate incoherent rays. SIMD is used in the hierarchy construction, the tree traversal and the leaf intersection. This is achieved by increasing the arity of acceleration structures, which also reduces memory requirements. We show that the resulting hierarchies can be built quickly and are smaller than acceleration structures known so far while at the same time outperforming them for incoherent rays. Our new acceleration structure speeds up ray tracing by a factor of 1.6 to 2.0 compared to a highly optimized bounding interval hierarchy implementation, and 1.3 to 1.6 compared to an efficient kd-tree. At the same time, the memory requirements are reduced by 10-50%. Additionally we show how a caching mechanism in conjunction with this memory efficient hierarchy can be used to speed up shadow rays in a global illumination algorithm without increasing the memory footprint. This optimization decreased the number of traversal steps up to 50%.

Getting rid of packets - Efficient SIMD single-ray traversal using multi-branching BVHs -

“…Concurrently to our work, the use of shallow BVHs has also been investigated by Dammertz et al [4]. In fact, both approaches are surprisingly similar in the data structure and algorithms they propose as well as in the conclusions they draw.…”

“…Concurrently to this paper, the idea of BVHs with branching factors higher than two has also been investigated by Dammertz et al [4]; we will come back to their approach-and contrast it with ours-in Section 6.4.…”

“…Current SIMD width for CPUs is 4 [10] (increasing to 8 in the near feature [11]) and recent GPUs provide an even higher SIMD width using their parallel floating point units. For the remainder of this paper, we will assume a SIMD width of 16, though the same approach would also fit other SIMD widths (a similar approach for a 4-wide SIMD architecture has been concurrently investigated in [4]). …”

Fast and Exact Fiber Surfaces for Tetrahedral Meshes

“…As discussed above, querying in the range for the set of cells that project on the edges of the FSCP can be reduced to an intersection test in 2-D. We therefore apply one of the most efficient strategies, based on bounding volume hierarchies (BVH) [20], [11].…”

Simulation of fluorescent concentrators

“…the angular experiment and use advances in fast ray tracing [5] research which make a quick simulation possible and even allow for interactive previews for dynamically changing material parameters.…”