Evaluating Stroke Expansion Efficiency Across GPUs and CPUs

This study benchmarks a new stroke rendering method implemented on both CPU and GPU. The GPU version—written in WGSL and tested on devices from mobile to RTX 4090—demonstrates up to 14× faster performance than CPU stroking, especially in curve-heavy scenes. By comparing line, arc, and Bézier primitives, the results show how GPU pipelines can drastically reduce execution time and memory use, paving the way for real-time, scalable vector rendering across devices.