Paralelização de Tarefas de Codificação de Vídeo VVC utilizando GPGPUs

• Iago Storch UFRGS
• Daniel Palomino UFPel
• Sergio Bampi UFRGS

DOI: https://doi.org/10.5753/webmedia_estendido.2024.244383

Resumo

The compression required in video-based applications imposes a significant computational workload. Moreover, computing systems are becoming increasingly heterogeneous, and GPUs have gained popularity. This Ph.D. Thesis proposes a GPU acceleration methodology that provides guidelines for efficiently leveraging heterogeneous CPU+GPU systems to accelerate video coding applications. Experimental results demonstrate that the solutions developed following the proposed methodology accelerate the encoding and improve its energy efficiency while posing minor coding efficiency losses.

Palavras-chave: sistemas heterogêneos, computação em GPU, codificação de vídeo

Referências

Bitmovin. 2023. Bitmovin and GAIA Project: Streaming Sustainability Progress Report. Technical Report. Bitmovin.

Benjamin Bross et. al. 2021. Overview of the Versatile Video Coding (VVC) Standard and its Applications. IEEE Trans. on Circ. and Syst. for Video Technol. 31, 10 (2021), 3736–3764.

Sandvine. 2024. The Global Internet Phenomena Report. Technical Report. Sandvine.