@article{2125,
  author = {Ghofrane El-Haj-Ahmed, Felipe Gil-Castineira, Enrique Costa-Montenegro, Pablo Counago-Soto},
  title = {System-on-Chip Evaluation for the Implementation of Video Processing Servers},
  journal = {Journal of Multimedia Processing and Technologies},
  year = {2016},
  volume = {7},
  number = {4},
  doi = {},
  url = {},
  abstract = {Nowadays, users are demanding an increasing amount of multimedia content (viewing videos, streaming contents in real time, using video-conferences for communication, etc.), but it is usually necessary to perform advanced operations in those video streams to adapt them to their different requirements. For example, it is necessary to transcode or to rescale those streams to adapt them to the display of the users device or to the available bandwidth, as well as mixing the different media streams for multiuser video-conferences. 

In the world of telephony, Media Servers perform this kind of operations for large amounts of users. To improve their performance, hardware acceleration devices, such as Graphics Processing Unit (GPUs), are typically used. Nevertheless, thanks to the high graphic computing power that require modern smartphones, current Systems-on-Chip (SoC) devices include high performance embedded GPUs. In order to evaluate the viability to use off the shelf embedded devices, that include different SoCs, to implement a Media Server, it is necessary to evaluate if such devices are able to perform different video operations in real time. 

In this paper, we evaluate and compare the performance of different embedded GPUs in order to know if they are suitable as low cost Media Servers.
},
}