Light tracing parallel optimization method based on Intel many-core framework

Abstract

The invention discloses a light tracing parallel optimization method based on an Intel many-core framework. The light tracing parallel optimization method based on the Intel many-core framework comprises the steps of 1, a spatial index BVH space acceleration structure for scene data is established; 2, primary light is generated, wherein a screen is divided into a plurality of screen subspaces according to the pixel and sampling is conducted, so that the primary light is generated; 3, light is traced, wherein the BVH space acceleration structure and scene geometric data are uploaded, communication between the Intel many-core framework and a CPU is conducted through asynchronous transmission, a plurality of sub-threads are started by a main thread, each sub-thread traces the primary light generated by each screen subspace according to the light beam, so that the point of intersection between each light ray and the scene is obtained; 4, the points of intersection are colored, wherein the points of intersection obtained during light tracking are colored, the step 2 is executed if secondary light is generated during coloring, the color of the primary light is obtained finally through coloring, and the light rays overlap on the screen, so that a final image is synthesized.

Description

technical field [0001] The invention relates to the field of realistic graphics rendering, in particular to a ray tracing parallel optimization method based on Intel many-core architecture. Background technique [0002] Ray tracing is an important algorithm for realistic rendering, which presents a three-dimensional space scene as a two-dimensional plane image by tracing light rays that travel in the opposite direction to the real world (that is, from the eyes to the light source). Ray tracing can achieve a series of realistic rendering results such as reflection, refraction, scattering, and shadows, and can simulate global illumination effects. It is widely used in virtual reality, advertising, film and television, animation games and other fields. [0003] Although the image generated by the ray tracing algorithm has high fidelity, the algorithm needs to emit a large amount of light for calculation. In addition to calculating the influence of the light source on the object...

AI Technical Summary

Problems solved by technology

[0007] The disadvantage of this method is that when the rays in the ray package lack consistency, the performance will drop significantly, and the number of active rays will decrease rapidly

The problem with this approach is that SIMD is not very efficient when different SIMD channels execute different code paths

Moreover, this method requires multiple independent traversal stacks and temporary variables, which requires a lot of registers and local storage space, resulting in thread scheduling delays and reduced efficiency.

In addition, if the rays are consistent, then the addressing operations generated by all SIMD channels are adjacent, but once the rays are severely dispersed, multiple completely inconsistent memory access requests will be generated, which is not conducive to the optimization of the storage system

Images