"Growing on the Cloud" Webinar | Using Heterogeneous Hardware such as GPUs and VPUs to Accelerate Video Transcoding

With the maturity of AI technology and the popularization of new applications such as cloud games and live broadcasting, the form of traffic has changed from text, graphics, and images to streaming media such as audio and video. In this context, the pursuit of extremely high efficiency and extremely low latency conflicts with traditional hardware bottlenecks and the status quo of idle computing power. How to efficiently combine heterogeneous hardware, fully improve hardware utilization, and reduce costs is now worth thinking about. a problem.

GPUs can provide elastic computing capabilities for many fields, and support for video transcoding is one of the most widely used fields of GPUs. The computing resources in cloud computing come from different hardware, the most representative of which are CPU and GPU, as well as professional computing hardware such as VPU, NPU, and TPU that emerged later. Each hardware has different characteristics and usage costs, which directly determine the design and implementation of the task scheduling module.

The transcoding service in Qingyun Cloud VOD service 👉 (free during the public beta period) is the practical application of GPU in video transcoding. After it is equipped with a GPU, it can make full use of the huge full product service capabilities of the public cloud through algorithm scheduling, provide users with efficient processing of massive videos, and meet users’ common media processing needs such as video transcoding, format conversion, cropping, and video watermarking. Reduce the user's difficulty in processing multimedia files.

In this seminar, we invited Li Xiaowen, an audio and video technology expert from Qingyun Technology, to share the application and practice of GPU, VPU and other heterogeneous hardware in video transcoding services.

Li Xiaowen Qingyun Technology Consulting R&D engineer, leading the technical architecture design of Qingyun audio and video processing. He has been engaged in the design and development of distributed systems for a long time, and has worked in the field of audio and video for many years, and is committed to continuously improving the product capabilities and service quality of Qingyun audio and video.

Thursday, September 1st at 3:00 pm
Scan the QR code to sign up for the first phase of Qingyun's "Growing on the Cloud" series of webinars to have the opportunity to get popular technical books!