[FFmpeg-devel] [PATCH 2/2] avcodec/mjpegdec: add support for frame threading
Paul B Mahol
onemda at gmail.com
Thu Sep 7 20:28:08 EEST 2023
On Thu, Sep 7, 2023 at 7:17 PM Paul B Mahol <onemda at gmail.com> wrote:
>
>
> On Tue, Dec 6, 2022 at 12:02 AM Timo Rothenpieler <timo at rothenpieler.org>
> wrote:
>
>> On 05.12.2022 15:15, Andreas Rheinhardt wrote:
>> > Timo Rothenpieler:
>> >> In my tests, this lead to a notable speed increase with the amount
>> >> of threads used. Decoding a 720p sample gave the following results:
>> >>
>> >> 1 Thread: 1428 FPS
>> >> 2 Threads: 2501 FPS
>> >> 8 Threads: 7575 FPS
>> >> Automatic: 11326 FPS (On a 16 Core/32 Threads system)
>> >> ---
>> >> libavcodec/jpeglsdec.c | 2 +-
>> >> libavcodec/mjpegdec.c | 13 +++++++------
>> >> libavcodec/sp5xdec.c | 4 ++--
>> >> 3 files changed, 10 insertions(+), 9 deletions(-)
>> >>
>>
>
> I made almost same patch, can you apply this one?
> Thanks.
>
Actually, on better look, this one is more complicated, and does not apply
anymore.
So ignore my 'request'.
>
>
>> >> diff --git a/libavcodec/jpeglsdec.c b/libavcodec/jpeglsdec.c
>> >> index 2e6d018ea6..c0642e8e30 100644
>> >> --- a/libavcodec/jpeglsdec.c
>> >> +++ b/libavcodec/jpeglsdec.c
>> >> @@ -559,7 +559,7 @@ const FFCodec ff_jpegls_decoder = {
>> >> .init = ff_mjpeg_decode_init,
>> >> .close = ff_mjpeg_decode_end,
>> >> FF_CODEC_RECEIVE_FRAME_CB(ff_mjpeg_receive_frame),
>> >> - .p.capabilities = AV_CODEC_CAP_DR1,
>> >> + .p.capabilities = AV_CODEC_CAP_DR1 | AV_CODEC_CAP_FRAME_THREADS,
>> >> .caps_internal = FF_CODEC_CAP_INIT_CLEANUP |
>> >> FF_CODEC_CAP_SETS_PKT_DTS,
>> >> };
>> >> diff --git a/libavcodec/mjpegdec.c b/libavcodec/mjpegdec.c
>> >> index 9b7465abe7..d30d722398 100644
>> >> --- a/libavcodec/mjpegdec.c
>> >> +++ b/libavcodec/mjpegdec.c
>> >> @@ -54,6 +54,7 @@
>> >> #include "exif.h"
>> >> #include "bytestream.h"
>> >> #include "tiff_common.h"
>> >> +#include "thread.h"
>> >>
>> >>
>> >> static int init_default_huffman_tables(MJpegDecodeContext *s)
>> >> @@ -713,7 +714,7 @@ int ff_mjpeg_decode_sof(MJpegDecodeContext *s)
>> >> s->avctx->pix_fmt,
>> >> AV_PIX_FMT_NONE,
>> >> };
>> >> - s->hwaccel_pix_fmt = ff_get_format(s->avctx, pix_fmts);
>> >> + s->hwaccel_pix_fmt = ff_thread_get_format(s->avctx,
>> pix_fmts);
>> >> if (s->hwaccel_pix_fmt < 0)
>> >> return AVERROR(EINVAL);
>> >>
>> >> @@ -729,7 +730,7 @@ int ff_mjpeg_decode_sof(MJpegDecodeContext *s)
>> >> }
>> >>
>> >> av_frame_unref(s->picture_ptr);
>> >> - if (ff_get_buffer(s->avctx, s->picture_ptr,
>> AV_GET_BUFFER_FLAG_REF) < 0)
>> >> + if (ff_thread_get_buffer(s->avctx, s->picture_ptr,
>> AV_GET_BUFFER_FLAG_REF) < 0)
>> >> return -1;
>> >> s->picture_ptr->pict_type = AV_PICTURE_TYPE_I;
>> >> s->picture_ptr->key_frame = 1;
>> >> @@ -2388,7 +2389,7 @@ static int mjpeg_get_packet(AVCodecContext
>> *avctx)
>> >> int ret;
>> >>
>> >> av_packet_unref(s->pkt);
>> >> - ret = ff_decode_get_packet(avctx, s->pkt);
>> >> + ret = ff_thread_decode_get_packet(avctx, s->pkt);
>> >> if (ret < 0)
>> >> return ret;
>> >>
>> >> @@ -3020,7 +3021,7 @@ const FFCodec ff_mjpeg_decoder = {
>> >> .close = ff_mjpeg_decode_end,
>> >> FF_CODEC_RECEIVE_FRAME_CB(ff_mjpeg_receive_frame),
>> >> .flush = decode_flush,
>> >> - .p.capabilities = AV_CODEC_CAP_DR1,
>> >> + .p.capabilities = AV_CODEC_CAP_DR1 | AV_CODEC_CAP_FRAME_THREADS,
>> >> .p.max_lowres = 3,
>> >> .p.priv_class = &mjpegdec_class,
>> >> .p.profiles = NULL_IF_CONFIG_SMALL(ff_mjpeg_profiles),
>> >> @@ -3050,7 +3051,7 @@ const FFCodec ff_thp_decoder = {
>> >> .close = ff_mjpeg_decode_end,
>> >> FF_CODEC_RECEIVE_FRAME_CB(ff_mjpeg_receive_frame),
>> >> .flush = decode_flush,
>> >> - .p.capabilities = AV_CODEC_CAP_DR1,
>> >> + .p.capabilities = AV_CODEC_CAP_DR1 | AV_CODEC_CAP_FRAME_THREADS,
>> >> .p.max_lowres = 3,
>> >> .caps_internal = FF_CODEC_CAP_INIT_CLEANUP |
>> >> FF_CODEC_CAP_SETS_PKT_DTS,
>> >> @@ -3068,7 +3069,7 @@ const FFCodec ff_smvjpeg_decoder = {
>> >> .close = ff_mjpeg_decode_end,
>> >> FF_CODEC_RECEIVE_FRAME_CB(ff_mjpeg_receive_frame),
>> >> .flush = decode_flush,
>> >> - .p.capabilities = AV_CODEC_CAP_DR1,
>> >> + .p.capabilities = AV_CODEC_CAP_DR1 | AV_CODEC_CAP_FRAME_THREADS,
>> >> .caps_internal = FF_CODEC_CAP_EXPORTS_CROPPING |
>> >> FF_CODEC_CAP_SETS_PKT_DTS |
>> FF_CODEC_CAP_INIT_CLEANUP,
>> >> };
>> >> diff --git a/libavcodec/sp5xdec.c b/libavcodec/sp5xdec.c
>> >> index 394448c5a9..8b08dc672a 100644
>> >> --- a/libavcodec/sp5xdec.c
>> >> +++ b/libavcodec/sp5xdec.c
>> >> @@ -101,7 +101,7 @@ const FFCodec ff_sp5x_decoder = {
>> >> .init = ff_mjpeg_decode_init,
>> >> .close = ff_mjpeg_decode_end,
>> >> FF_CODEC_RECEIVE_FRAME_CB(ff_mjpeg_receive_frame),
>> >> - .p.capabilities = AV_CODEC_CAP_DR1,
>> >> + .p.capabilities = AV_CODEC_CAP_DR1 | AV_CODEC_CAP_FRAME_THREADS,
>> >> .p.max_lowres = 3,
>> >> .caps_internal = FF_CODEC_CAP_INIT_CLEANUP |
>> >> FF_CODEC_CAP_SETS_PKT_DTS,
>> >> @@ -118,7 +118,7 @@ const FFCodec ff_amv_decoder = {
>> >> .close = ff_mjpeg_decode_end,
>> >> FF_CODEC_RECEIVE_FRAME_CB(ff_mjpeg_receive_frame),
>> >> .p.max_lowres = 3,
>> >> - .p.capabilities = AV_CODEC_CAP_DR1,
>> >> + .p.capabilities = AV_CODEC_CAP_DR1 | AV_CODEC_CAP_FRAME_THREADS,
>> >> .caps_internal = FF_CODEC_CAP_INIT_CLEANUP |
>> >> FF_CODEC_CAP_SETS_PKT_DTS,
>> >> };
>> >
>> > Can you test the sample that Michael mentioned here:
>> >
>> https://patchwork.ffmpeg.org/project/ffmpeg/patch/AS8PR01MB7944E105BE990A5D01EF89208FEF9@AS8PR01MB7944.eurprd01.prod.exchangelabs.com/
>> ?
>> > (I never got around to analyzing this, but if I am not mistaken, it
>> > shows that this decoder is not a simple one-in, one-out decoder, so that
>> > making it multithreaded will be more complicated than just adding the
>> > flag and using ff_thread_get_buffer/format.)
>>
>> Assuming this is the right file:
>> https://trac.ffmpeg.org/attachment/ticket/1915/not_interleaved.avi
>>
>> It does not decode for me at all, with or without this patch applied:
>>
>> > [mjpeg @ 0x562e81877c80] No JPEG data found in image
>> > Error while decoding stream #0:0: Invalid data found when processing
>> input
>>
>>
>> _______________________________________________
>> ffmpeg-devel mailing list
>> ffmpeg-devel at ffmpeg.org
>> https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>>
>> To unsubscribe, visit link above, or email
>> ffmpeg-devel-request at ffmpeg.org with subject "unsubscribe".
>>
>
More information about the ffmpeg-devel
mailing list