This change improves the performance and multicore scalability of the vp9
codec for streaming single-pass encoded videos by taking advantage of up
to 64 cores in the system. The current thread limit for ffmpeg codecs is 16
(MAX_AUTO_THREADS in pthread_internal.h) due to a limitation in H.264 codec
that prevents more than 16 threads being used.
Experiments show that increasing the thread limit to 64 for vp9 improves
the performance for encoding 4K raw videos for streaming by up to 47%
compared to 16 threads, and from 20-30% for 32 threads, with the same quality
as measured by the VMAF score.
Rationale for this change:
Vp9 uses tiling to split the video frame into multiple columns; tiles must
be at least 256 pixels wide, so there is a limit to how many tiles can be
used. The tiles can be processed in parallel, and more tiles mean more CPU
threads can be used. 4K videos can make use of 16 threads, and 8K videos
can use 32. Row-mt can double the number of threads so 64 threads can be used.
Signed-off-by: James Zern <jzern@google.com>
Up until now, codec.h contains both public and private parts
of AVCodec. This exposes the internals of AVCodec to users
and leads them into the temptation of actually using them
and forces us to forward-declare structures and types that
users can't use at all.
This commit changes this by adding a new structure FFCodec to
codec_internal.h that extends AVCodec, i.e. contains the public
AVCodec as first member; the private fields of AVCodec are moved
to this structure, leaving codec.h clean.
Reviewed-by: Anton Khirnov <anton@khirnov.net>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Bump the minimum libvpx version to 1.3.0 and rework the configure logic
to fail only if no decoders and encoders are found.
Based on the original patch from Vittorio.
Signed-off-by: Vittorio Giovara <vittorio.giovara@gmail.com>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
This one is tricky. Consider a graph with two sink links, A and B.
request_oldest() requests on A, and A returns EOF. This EOF causes
a filter to flush pending frames to B; they are unrequested.
Then request_oldest() moves on to B, and B returns EOF.
This is required for letting applications to create and destroy
AVFilterInOut structs in a convenient way.
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Unlike avfilter_graph_parse(), it returns unlinked inputs and outputs
to the caller, which allows parsing of graphs where inputs/outputs are
not known in advance.
Make avfilter_graph_free() free not only the internal structures, but
also the allocated graph, and set the graph pointer to NULL for
increased safety.
Simplify usage.
Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit 4359288c56)
Make avfilter_graph_free() free not only the internal structures, but
also the allocated graph, and set the graph pointer to NULL for
increased safety.
Simplify usage.
Signed-off-by: Mans Rullgard <mans@mansr.com>
ff_avfilter_graph_check_validity()
ff_avfilter_graph_config_links()
ff_avfilter_graph_config_formats()
and move their declaration to internal.h. These functions are never
used in application code, so it is better to consider them internal
functions, this can be changed later if necessary. Simplify API.
Originally committed as revision 25737 to svn://svn.ffmpeg.org/ffmpeg/trunk
avfilter_graph_parse() declaration to libavfilter/avfiltergraph.h.
Simplify, and less confusing for the user.
Originally committed as revision 25705 to svn://svn.ffmpeg.org/ffmpeg/trunk
The new name is shorter and more consistent with the rest of the API.
This change breaks libavfilter API/ABI.
Originally committed as revision 25674 to svn://svn.ffmpeg.org/ffmpeg/trunk
James Almer
OvchinnikovDmitrii
Andreas Rheinhardt
Vittorio Giovara
Luca Barbato
Michael Niedermayer
Guillaume Martres
Clément Bœsch
Anton Khirnov
Justin Ruggles
Stefano Sabatini
Nicolas George
Stefano Sabatini
Diego Biurrun
Mans Rullgard