The rk3588 platform uses Rockchip’s official ffmpeg（ GitHub - nyanmisaka/ffmpeg-rockchip: FFmpeg with async and zero-copy Rockchip MPP & RGA support ）The compiled ffplay can play 1080p videos with a CPU usage of up to 400%. How to confirm that the system hardware supports rkmpp hardware decoding and how to configure it?
Will high CPU usage occur when using our official Debian firmware?
Using Rockhip’s mpp program mpi_dec_test for testing, we found 4k H264 decoders with CPU usage ranging from 80% to 90%, but not exceeding 90%. I tested 1080P again, and it took up 60% -95% in the first minute and 30% -50% in the following minutes. Is this CPU usage normal? I feel like the CPU usage is still a bit high, this is just a video decoding without rendering. Do we need to configure the hardware layer?
The CPU usage of the 4k H264 decoder is between 80% and 90%, but not more than 90%. This should be normal. What is the current frame rate? In addition, rk3588 is 800%
If you only use MPP:
Scenario: A customized video surveillance system that needs to perform hardware decoding and display of H.265 video streams.
Developers directly call the underlying API of MPP to communicate with the hardware to decode the video stream.
If you use Rockit:
Scenario: Develop a media player that supports multi-format playback and integrates GStreamer.
Developers use the GStreamer plug-in provided by Rockit to quickly access Rockchip hardware acceleration functions without having to pay attention to the underlying details of MPP.
Thank you for your reply. Currently, the solution is implemented directly using the MPP interface. However, based on the testing results of the mpi_dec_test program, it takes up a lot of CPU. If we only use the MPP library API for decoding, do we need to configure the relevant hardware? Does using the API directly indicate hardware decoding
Yes, hardware encoding and decoding is already supported using the mpp framework