FFmpeg low CPU usage

cpuffmpegvideovideo conversion

I'm using FFmpeg to convert a number of files from MP4(H264 & AC3) to MP4(H264 & AAC) using the command ffmpeg -i in.mp4 -c:v copy -c:a aac out.mp4

I have a 4 core CPU with hyper-threading that shows around 15% usage when doing the conversion. The files are located on an SSD, and disk activity is around 3MB/s, so disk does not seem to be a bottleneck.

When the encoding starts the CPU hits its maximum frequency, but I have not noticed the load on any of the cores to go over about 50%.

Does FFmpeg make poor use of multiple cores? Is there any way to make the encode faster?

Best Answer

The only thing you're encoding is audio, and most audio encoding libraries in existence are single-threaded. This is most likely because audio encoding is already blazing fast as a single-threaded application (when compared to video encoding anyway), and it doesn't use too much memory so it's actually feasible to encode each file using a single thread and just start up as many separate processes as necessary to fully saturate the CPU. Factor in the fact that multi-threading also doesn't necessarily result in linear performance improvements and you probably have the reason why the developers of most audio encoders don't think multi-threading is a high priority. I know only of two audio encoders that implement multi-threading - LAME MT for MP3 and pflac for FLAC - and both are separate modifications that aren't part of the main codebases of the projects they derived from.

As for your CPU usage, with hyper-threading you have 8 logical cores and one eight of 100% is 12.5% which isn't too far from your 15% utilisation figure. I'm not really sure why you system isn't showing a 100% load on any of the cores, perhaps the OS is moving the process between cores to even out the load or something like that.

If you need to encode a large number of files you might want to consider writing a script that spins up multiple FFmpeg processes to encode multiple files at the same time. I have very little scripting/programming experience but I know of an open source tool that applies the same logic for image optimisation: picopt. So if you need a pointer on how to do it in Python you could take a look at picopt's source code.

Related Solutions

Linux – ffmpeg not using full CPU power

Like most code, this code is not doing a good job of taking advantage of multiple cores. It may switch which core it's placing load on very rapidly (around 100 times per second), but it is only able to take full advantage of one core at a time.

FFMPEG multiple outputs performance (Single instance vs Multiple instances)

A less obvious problem is that depending on your input/output or filters ffmpeg might need to do pixel format conversion internally and in certain cases this becomes a bottleneck when using parallel outputs if done on each stream separately.

The idea is to do the pixel format conversion once if possible, like:

-filter_complex '[0:v]format=yuv420p, split=3[s1][s2][s3]' \
-map '[s1]' ... \
-map '[s2]' ... \
-map '[s3]' ... \

Same filters applied to all outputs should also be used only once. Some filters might need a specific pixel format.

For other causes see the small note at the bottom of the wiki:

Parallel encoding

Outputting and re encoding multiple times in the same FFmpeg process will typically slow down to the "slowest encoder" in your list. Some encoders (like libx264) perform their encoding "threaded and in the background" so they will effectively allow for parallel encodings, however audio encoding may be serial and become the bottleneck, etc. It seems that if you do have any encodings that are serial, it will be treated as "real serial" by FFmpeg and thus your FFmpeg may not use all available cores.

Best Answer

Related Solutions

Linux – ffmpeg not using full CPU power

FFMPEG multiple outputs performance (Single instance vs Multiple instances)

Related Question