ZimaCube N100 prototype V2.0 Video Encoding Performance Test
Hardware and Environment Information:
Hardware: ZimaCube N100 prototype V2.0
Operating System and Version: ZimaOS V0.4.9
Transcoding Tool: ffmpeg
Test File: ZimaCube.m4v
Test Scheme:
Use the same test file and test environment, call the computing power of the CPU and IGPU to do the same work, and output encoding files in the same format. Compare soft coding and hard coding in coding speed, CPU utilization and output quality.
Test Process:
CPU：
Encode the input video file ZimaCube.m4v into H.264 format video file out.mp4 using libx264 encoder, and limit the bitrate to 2 Mbps.
1. Run command:
root@zima:/DATA/Media/Movies# ffmpeg -y -vsync 0 -i ZimaCube.m4v -vcodec libx264 -preset medium -b:v 2000000 out.mp4
2. CPU usage during encoding:
3. Run results:
4. Transcoding information:
- Progress: frame=1917, fps=30, time=00:01:16.67, speed=1.2x
IGPU：
Use VA-API for hardware acceleration to decode the input video file ZimaCube.m4v, then use VA-API for H.264 encoding, and finally save the output as output.mp4.
1. Run command:
root@c2985d912a60:/# /usr/lib/jellyfin-ffmpeg/ffmpeg -hwaccel vaapi -hwaccel_output_format vaapi -hwaccel_device /dev/dri/renderD128 -i /Media/Movies/ZimaCube.m4v -c:v h264_vaapi output.mp4
2. CPU usage during encoding:
3. Run results:
4. Transcoding information:
- Progress: frame=1917, fps=173, time=00:01:16.67, speed=6.93x
Conclusion:
In the video encoding performance test, we compared the effects of CPU-only software encoding and VAAPI hardware accelerated encoding. The test results show that enabling VAAPI hardware encoding on the same processor can increase the video encoding speed by nearly 6 times, from 1.2x real-time encoding to nearly 7x real-time encoding, while also reducing CPU utilization.
