GPU Programming Primitives for Computer Graphics
Daniel Meister, Atsushi Yoshimura, Chih-Chen Kao · 2023
Various parallel algorithms can be decomposed into programming primitives that share similar patterns. This course focuses on studying these programming primitives and their applicability in computer graphics, specifically in the context of massively parallel processing on GPUs. The course begins by establishing a theoretical foundation, followed by practical examples and real-world applications. We explain two pivotal algorithms: parallel reduction and parallel prefix scan in detail, discussing their variants and different implementations. Afterward, we provide a collection of more advanced techniques and tricks applicable across various domains. At the end of the course, we also briefly discuss code optimization.