Publicación

Tipo de publicación: Artículos científicos (Publicados en revistas de divulgación o científicas)

Título del artículo: CPU–GPU bufer communication using compute shader to fll volumes with spheres

Registrado por: The Journal of Supercomputing

Año de publicación: 2022

Resumen: This paper describes the usage of shaders to make parallel operations by improving
the CPU–GPU communication, using both rendering and compute shaders. When
the number of spheres is large, the execution becomes slow and requires a lot of
space to store particles. We parallelized the frozen method using an efcient interprocess GPU communication to reduce the operations required. We propose to handle two bufers, one for operations and the other for rendering. While the rendering
bufer increases as the number of spheres is required, the operational bufer maintains its size and the number of operations is hold. Experimental results demonstrate
that the proposed method shows up 100x throughput improvement over the sequential version. We defne a confguration by a 4-tuple as input to the algorithm, and we
found a pattern to choose better confgurations.

Editorial:

Número de páginas:

Autor(es): F. A. Madera?Ramirez, J. L. Lopez?Martinez, F. Moo?Mena, J. Gomez?Montalvo

Campo: Computación

Disciplina: Ciencias de la Computación

Subdisciplina: Procesamiento en Paralelo

Número de registro: The Journal of Supercomputing https://doi.org/10.1007/s11227-021-04136-1