The new version of the Intellect software package (Axxon Intellect 4.7.6) includes the updated video decoding system which uses the computational resources of the graphic processor (GPU) of the video card. The graphic processors based on parallel computations CUDA architecture from NVIDIA are now supported. Using GPU resources considerably increases the performance of the video system and decreases the CPU load.
Video decoding is one of the main problems in current distributed video surveillance systems that use IP-cameras or analog cameras with IP-videoservers. The reason is that video from multiple sources comes compressed. On the one hand, it saves computing power during archiving, but, on the other hand, video streams should be decompressed for playback or analysis, which requires a lot of resources.
The AxxonSoft solution is conceptually new – using the video card graphic processor for video decompression. This increases the performance of the video system several times over and frees up the CPU which can then be used for other tasks, such as resource-intensive video analysis algorithms. The currently used GPUs even in medium-priced NVIDIA GeForce video cards can decompress up to 4 Full HD video streams of up to 30 fps frame rate in H.264 format.
These developments are revolutionary for video processing. Using the GPU will allow for more video channels per computer and slash the number of computers required in the system. This results in increased reliability of the system, while its cost of ownership (purchase and running costs ) falls due to lower hardware and software costs, lower power consumption and maintenance costs.
Axxon Intellect 4.7.6 supports all NVIDIA video cards using CUDA-based GPUs. They allow video decompression in H.264 format. The next version is anticipated to support NVIDIA Tesla processors and MJPEG format. GPUs are likely to be used not only for video decompression, but for video analysis and post-processing before playback, as well.