[FFmpeg-trac] #3208(FFmpeg:new): Very long CPU-bound loop with -fflags +genpts and -af volumedetect

FFmpeg trac at avcodec.org
Sun Dec 8 19:18:31 CET 2013

#3208: Very long CPU-bound loop with -fflags +genpts and -af volumedetect
             Reporter:  lelegard  |                     Type:  defect
               Status:  new       |                 Priority:  normal
            Component:  FFmpeg    |                  Version:  2.1.1
             Keywords:            |               Blocked By:
             Blocking:            |  Reproduced by developer:  0
Analyzed by developer:  0         |
 This report describes a problem with memory usage and processing time of
 ffmpeg 2.1.1 on Windows 7 using the "volumedetect" audio filter in
 combination with the "-fflags +genpts" option.

 I know that this combination is probably useless. But the ffmpeg commands
 was automatically generated by a tool which added "-fflags +genpts" on all
 ffmpeg commands.

 $ ffmpeg -fflags +genpts -i input.avi -af volumedetect -vn -f null -y NUL
 ffmpeg version 2.1.1 Copyright (c) 2000-2013 the FFmpeg developers
   built on Nov 20 2013 21:13:48 with gcc 4.8.2 (GCC)
   configuration: --enable-gpl --enable-version3 --disable-w32threads
 --enable-avisynth --enable-bzlib --enable-fontconfig --enable-frei0r
 --enable-gnutls --enable-iconv --enable-libass --enable-libbluray
 --enable-libcaca --enable-libfreetype --enable-libgsm --enable-libilbc
 --enable-libmodplug --enable-libmp3lame --enable-libopencore-amrnb
 --enable-libopencore-amrwb --enable-libopenjpeg --enable-libopus --enable-
 librtmp --enable-libschroedinger --enable-libsoxr --enable-libspeex
 --enable-libtheora --enable-libtwolame --enable-libvidstab --enable-libvo-
 aacenc --enable-libvo-amrwbenc --enable-libvorbis --enable-libvpx
 --enable-libwavpack --enable-libx264 --enable-libxavs --enable-libxvid
   libavutil      52. 48.101 / 52. 48.101
   libavcodec     55. 39.101 / 55. 39.101
   libavformat    55. 19.104 / 55. 19.104
   libavdevice    55.  5.100 / 55.  5.100
   libavfilter     3. 90.100 /  3. 90.100
   libswscale      2.  5.101 /  2.  5.101
   libswresample   0. 17.104 /  0. 17.104
   libpostproc    52.  3.100 / 52.  3.100
 Input #0, avi, from 'input.avi':
   Duration: 01:26:59.64, start: 0.000000, bitrate: 1469 kb/s
     Stream #0:0: Video: mpeg4 (Advanced Simple Profile) (XVID /
 0x44495658), yuv420p, 696x452 [SAR 1:1 DAR 174:113], 25 tbr, 25 tbn, 25
     Stream #0:1: Audio: mp3 (U[0][0][0] / 0x0055), 48000 Hz, stereo, s16p,
 128 kb/s
 Output #0, null, to 'NUL':
     encoder         : Lavf55.19.104
     Stream #0:0: Audio: pcm_s16le, 48000 Hz, stereo, s16, 1536 kb/s
 Stream mapping:
   Stream #0:1 -> #0:0 (mp3 -> pcm_s16le)
 Press [q] to stop, [?] for help

 All the above is quickly displayed. Then, there is no more display for
 several minutes. During this period, ffmpeg uses 100% CPU and the virtual
 memory of the process slowly increases up to approximately 180 MB.

 After several minutes, the following display restarts. The "time" value
 increases at the usual speed. And the virtual memory decreases quite

 size=N/A time=01:26:59.11 bitrate=N/A
 video:0kB audio:978584kB subtitle:0 global headers:0kB muxing overhead
 [Parsed_volumedetect_0 @ 000000000282f4a0] n_samples: 501034752
 [Parsed_volumedetect_0 @ 000000000282f4a0] mean_volume: -20.7 dB
 [Parsed_volumedetect_0 @ 000000000282f4a0] max_volume: 0.0 dB
 [Parsed_volumedetect_0 @ 000000000282f4a0] histogram_0db: 462
 [Parsed_volumedetect_0 @ 000000000282f4a0] histogram_1db: 3951
 [Parsed_volumedetect_0 @ 000000000282f4a0] histogram_2db: 19706
 [Parsed_volumedetect_0 @ 000000000282f4a0] histogram_3db: 71897
 [Parsed_volumedetect_0 @ 000000000282f4a0] histogram_4db: 209764
 [Parsed_volumedetect_0 @ 000000000282f4a0] histogram_5db: 496839

 Using the same command without the "-fflags +genpts" option, there is no
 CPU-bound loop in the middle. The "time" line starts to be updated
 immediately. The virtual memory of the process is stable at 11 MB.

 But the  "-fflags +genpts" option is not responsible alone for this
 problem. When this option is used in a conversion command, there is no
 such very long CPU-bound loop. The "time" line starts to be updated
 immediately. The virtual memory keeps increasing during the execution up
 to more than 220 MB but this is probably the required memory for the
 conversion process.


 The "-fflags +genpts" option, when used in combination with the audio
 filter "volumedetect" only, creates a several minutes long CPU-bound loop
 with an increasing virtual memory.

 The "-fflags +genpts" option, when used in another context such as an
 audio/video conversion, does not create this artefact.


 Obvious. The "-fflags +genpts" option is useless with the audio filter
 "volumedetect", so do not use it.

Ticket URL: <https://trac.ffmpeg.org/ticket/3208>
FFmpeg <http://ffmpeg.org>
FFmpeg issue tracker

More information about the FFmpeg-trac mailing list