diff options
| author | Michael Niedermayer <michaelni@gmx.at> | 2011-05-21 03:11:50 +0200 |
|---|---|---|
| committer | Michael Niedermayer <michaelni@gmx.at> | 2011-05-21 03:11:50 +0200 |
| commit | 0424e052f83adc422d8a746e3cdc5ab6bc28679e (patch) | |
| tree | 80437095d9912fb844ec2773afe862d54f08c80d /mt-work | |
| parent | 5c08c7b2151409c3b7a05e54fe46834deedff119 (diff) | |
| download | ffmpeg-0424e052f83adc422d8a746e3cdc5ab6bc28679e.tar.gz | |
Merge remote-tracking branch 'ffmpeg-mt/master'
* ffmpeg-mt/master:
Update todo.
h264: add an assert that copied pictures are valid picture pointers
valgrind-check: run with 1 and 3 threads
h264: When decoding a packet with multiple PPS/SPS, don't start the next thread until all of them have been read
Allow some pictures to be released earlier after 51ead6d2c40c5defdd211f435aec49b19f5f6a18
h264: fix slice threading MC reading uninitialized frame edges.
Please see ffmpeg-mt for a list of authors of these changes.
Conflicts:
libavcodec/h264.c
mt-work/valgrind-check.sh
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Diffstat (limited to 'mt-work')
| -rw-r--r-- | mt-work/todo.txt | 14 | ||||
| -rw-r--r-- | mt-work/valgrind-check.sh | 4 |
2 files changed, 11 insertions, 7 deletions
diff --git a/mt-work/todo.txt b/mt-work/todo.txt index 013853e3ae..678d213d8c 100644 --- a/mt-work/todo.txt +++ b/mt-work/todo.txt @@ -1,7 +1,7 @@ Todo -- For other people -- Multithread vp8 or vc1. +- Multithread vc1. - Multithread an intra codec like mjpeg (trivial). - Fix mpeg1 (see below). - Try the first three items under Optimization. @@ -18,11 +18,13 @@ work.) In general testing error paths should be done more. bugs in vsync in ffmpeg.c, which are currently obscuring real failures. h264: -- Files split at the wrong NAL unit don't (and can't) -be decoded with threads (e.g. TS split so PPS is after -the frame, PAFF with two fields in a packet). Scan the -packet at the start of decode and don't finish setup -until all PPS/SPS have been encountered. +- Files that aren't parsed (e.g. mp4) and contain PAFF with two +field pictures in the same packet are not optimal. Modify the +nals_needed check so that the second field's first slice is +considered as needed, then uncomment the FIXME code in decode_postinit. +Ex: http://astrange.ithinksw.net/ffmpeg/mt-samples/PAFF-Chalet-Tire.mp4 +- The conformance sample MR3_TANDBERG_B.264 has problems (allocated picture overflow). +- One 10-bit sample has problems. mpeg4: - Packed B-frames need to be explicitly split up diff --git a/mt-work/valgrind-check.sh b/mt-work/valgrind-check.sh index dc3833abb6..276327a76a 100644 --- a/mt-work/valgrind-check.sh +++ b/mt-work/valgrind-check.sh @@ -1,3 +1,5 @@ #!/bin/bash -valgrind --leak-check=full ./ffmpeg_g -threads 3 -vsync 0 -y -t 30 -i "$1" -an -f framecrc /dev/null
\ No newline at end of file +valgrind --track-origins=yes --leak-check=full ./ffmpeg_g -threads 1 -vsync 0 -y -t 30 -i "$1" -an -f null /dev/null + +valgrind --track-origins=yes --leak-check=full ./ffmpeg_g -threads 3 -vsync 0 -y -t 30 -i "$1" -an -f null /dev/null |