Quantum Computing May Make Ray Tracing Easier

Quantum Computing May Make Ray Tracing Easier

Posted on

An international team of researchers across the UK, the US and Portugal think they’ve found an answer to ray tracing’s steep performance requirements. And according to them, the answer lies in a hybrid of classical ray tracing algorithms with quantum computing. According to the research paper (currently in preprint), ray tracing workloads aided by quantum computing can offer an up to 190% performance improvement by slashing the number of calculations required by each ray, significantly reducing the requirements of the technology.

The introduction of ray tracing in graphics technologies has marked a significant evolution in the way we render games. And yet, its adoption and performance have been relatively limited compared to how groundbreaking this technology is. Part of the reason stems from ray tracing’s deep hardware and computational requirements, which can bring even the world’s most powerful GPUs to their knees. In addition, the need for specialized hardware locks most users out of the technology, barring a discrete GPU upgrade that can handle such workloads.

The current surge in upscaling technologies from all GPU vendors: Nvidia’s DLSS, AMD’s FSR 1.0 and FSR 2.0, and Intel’s upcoming XeSS were primarily built to offset the extreme performance penalties that come from enabling ray tracing. These technologies work by lowering the amount of rendered pixels to reduce the computational complexity of a given scene before applying an algorithm that reconstructs the image towards its targeted output resolution. This approach is not without caveats, despite the image quality improvements that have been continuously built into these software suites since their introduction.

(Image credit: Towards Quantum Ray Tracing paper)

The research paper offers yet another way to significantly reduce the computational expense of ray tracing. The researchers ultimately demonstrated their claims by rendering a small, 128×128 ray traced image in three approaches: classical rendering, non-optimized quantum rendering, and optimized quantum rendering. The results speak for themselves: the classical rendering technique required computing 2,678 million ray intersections on that tiny 3D image (64 per ray). The unoptimized quantum technique nearly halved that number, requiring only 33.6 intersection evaluations per ray (for a total of 1,366 million ray intersections). Finally, the optimized quantum-classical hybrid algorithm managed to render the same image with only 896 thousand intersection evaluations, averaging out at 22.1 per ray – a far cry from the 64 per ray achieved with current rendering techniques.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *