Git Inbox Mirror of the ffmpeg-devel mailing list - see https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
 help / color / mirror / Atom feed
  • * Re: [FFmpeg-devel] [PATCH 1/3] libavcodec/vaapi_encode: Change the way to call async to increase performance
       [not found] <20211027085705.4114165-1-wenbin.chen@intel.com>
       [not found] ` <20211027085705.4114165-3-wenbin.chen@intel.com>
@ 2021-12-27  7:22 ` Xiang, Haihao
  2021-12-27  7:41   ` Chen, Wenbin
  2021-12-27 18:31 ` Mark Thompson
  2 siblings, 1 reply; 6+ messages in thread
From: Xiang, Haihao @ 2021-12-27  7:22 UTC (permalink / raw)
  To: ffmpeg-devel; +Cc: Chen, Wenbin

On Wed, 2021-10-27 at 16:57 +0800, Wenbin Chen wrote:
> Fix: #7706. After commit 5fdcf85bbffe7451c2, vaapi encoder's performance
> decrease. The reason is that vaRenderPicture() and vaSyncSurface() are
> called at the same time (vaRenderPicture() always followed by a
> vaSyncSurface()). When we encode stream with B frames, we need buffer to
> reorder frames, so we can send serveral frames to HW at once to increase
> performance. Now I changed them to be called in a
> asynchronous way, which will make better use of hardware.
> 1080p transcoding increases about 17% fps on my environment.

Could you provide your command ? I'd like to have a try. 

Thanks
Haihao

> 
> Signed-off-by: Wenbin Chen <wenbin.chen@intel.com>
> ---
>  libavcodec/vaapi_encode.c | 41 ++++++++++++++++++++++++++++-----------
>  libavcodec/vaapi_encode.h |  3 +++
>  2 files changed, 33 insertions(+), 11 deletions(-)
> 
> diff --git a/libavcodec/vaapi_encode.c b/libavcodec/vaapi_encode.c
> index ec054ae701..5927849233 100644
> --- a/libavcodec/vaapi_encode.c
> +++ b/libavcodec/vaapi_encode.c
> @@ -951,8 +951,10 @@ static int vaapi_encode_pick_next(AVCodecContext *avctx,
>      if (!pic && ctx->end_of_stream) {
>          --b_counter;
>          pic = ctx->pic_end;
> -        if (pic->encode_issued)
> +        if (pic->encode_complete)
>              return AVERROR_EOF;
> +        else if (pic->encode_issued)
> +            return AVERROR(EAGAIN);
>      }
>  
>      if (!pic) {
> @@ -1177,20 +1179,31 @@ int ff_vaapi_encode_receive_packet(AVCodecContext
> *avctx, AVPacket *pkt)
>              return AVERROR(EAGAIN);
>      }
>  
> -    pic = NULL;
> -    err = vaapi_encode_pick_next(avctx, &pic);
> -    if (err < 0)
> -        return err;
> -    av_assert0(pic);
> +    while (av_fifo_size(ctx->encode_fifo) <= MAX_PICTURE_REFERENCES *
> sizeof(VAAPIEncodePicture *)) {
> +        pic = NULL;
> +        err = vaapi_encode_pick_next(avctx, &pic);
> +        if (err < 0)
> +            break;
> +        av_assert0(pic);
>  
> -    pic->encode_order = ctx->encode_order++;
> +        pic->encode_order = ctx->encode_order +
> +                            (av_fifo_size(ctx->encode_fifo) /
> sizeof(VAAPIEncodePicture *));
>  
> -    err = vaapi_encode_issue(avctx, pic);
> -    if (err < 0) {
> -        av_log(avctx, AV_LOG_ERROR, "Encode failed: %d.\n", err);
> -        return err;
> +        err = vaapi_encode_issue(avctx, pic);
> +        if (err < 0) {
> +            av_log(avctx, AV_LOG_ERROR, "Encode failed: %d.\n", err);
> +            return err;
> +        }
> +
> +        av_fifo_generic_write(ctx->encode_fifo, &pic, sizeof(pic), NULL);
>      }
>  
> +    if (!av_fifo_size(ctx->encode_fifo))
> +        return err;
> +
> +    av_fifo_generic_read(ctx->encode_fifo, &pic, sizeof(pic), NULL);
> +    ctx->encode_order = pic->encode_order + 1;
> +
>      err = vaapi_encode_output(avctx, pic, pkt);
>      if (err < 0) {
>          av_log(avctx, AV_LOG_ERROR, "Output failed: %d.\n", err);
> @@ -2520,6 +2533,11 @@ av_cold int ff_vaapi_encode_init(AVCodecContext *avctx)
>          }
>      }
>  
> +    ctx->encode_fifo = av_fifo_alloc((MAX_PICTURE_REFERENCES + 1) *
> +                                      sizeof(VAAPIEncodePicture *));
> +    if (!ctx->encode_fifo)
> +        return AVERROR(ENOMEM);
> +
>      return 0;
>  
>  fail:
> @@ -2552,6 +2570,7 @@ av_cold int ff_vaapi_encode_close(AVCodecContext *avctx)
>  
>      av_freep(&ctx->codec_sequence_params);
>      av_freep(&ctx->codec_picture_params);
> +    av_fifo_freep(&ctx->encode_fifo);
>  
>      av_buffer_unref(&ctx->recon_frames_ref);
>      av_buffer_unref(&ctx->input_frames_ref);
> diff --git a/libavcodec/vaapi_encode.h b/libavcodec/vaapi_encode.h
> index b41604a883..89fe8de466 100644
> --- a/libavcodec/vaapi_encode.h
> +++ b/libavcodec/vaapi_encode.h
> @@ -29,6 +29,7 @@
>  
>  #include "libavutil/hwcontext.h"
>  #include "libavutil/hwcontext_vaapi.h"
> +#include "libavutil/fifo.h"
>  
>  #include "avcodec.h"
>  #include "hwconfig.h"
> @@ -345,6 +346,8 @@ typedef struct VAAPIEncodeContext {
>      int             roi_warned;
>  
>      AVFrame         *frame;
> +
> +    AVFifoBuffer *encode_fifo;
>  } VAAPIEncodeContext;
>  
>  enum {
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 6+ messages in thread
  • * Re: [FFmpeg-devel] [PATCH 1/3] libavcodec/vaapi_encode: Change the way to call async to increase performance
       [not found] <20211027085705.4114165-1-wenbin.chen@intel.com>
       [not found] ` <20211027085705.4114165-3-wenbin.chen@intel.com>
  2021-12-27  7:22 ` [FFmpeg-devel] [PATCH 1/3] libavcodec/vaapi_encode: Change the way to call async " Xiang, Haihao
@ 2021-12-27 18:31 ` Mark Thompson
  2021-12-28  2:05   ` Chen, Wenbin
  2 siblings, 1 reply; 6+ messages in thread
From: Mark Thompson @ 2021-12-27 18:31 UTC (permalink / raw)
  To: ffmpeg-devel

On 27/10/2021 09:57, Wenbin Chen wrote:
> Fix: #7706. After commit 5fdcf85bbffe7451c2, vaapi encoder's performance
> decrease. The reason is that vaRenderPicture() and vaSyncSurface() are
> called at the same time (vaRenderPicture() always followed by a
> vaSyncSurface()). When we encode stream with B frames, we need buffer to
> reorder frames, so we can send serveral frames to HW at once to increase
> performance. Now I changed them to be called in a
> asynchronous way, which will make better use of hardware.
> 1080p transcoding increases about 17% fps on my environment.
> 
> Signed-off-by: Wenbin Chen <wenbin.chen@intel.com>
> ---
>   libavcodec/vaapi_encode.c | 41 ++++++++++++++++++++++++++++-----------
>   libavcodec/vaapi_encode.h |  3 +++
>   2 files changed, 33 insertions(+), 11 deletions(-)

The API does not allow this behaviour.

For some bizarre reason (I think a badly-written example combined with the Intel driver being synchronous in vaEndPicture() for a long time), the sync to a surface is to the /input/ surface of an encode rather than the output surface.

That means you can't have multiple encodes outstanding on the same surface and expect to sync usefully, because the only argument to vaSyncSurface() is the surface to sync to without anything about the associated context.

Therefore trying to make it asynchronous like this falls down when input surfaces might appear multiple times, or might be used in the input of multiple encoders, because you can't tell whether your sync means the thing you actually wanted to finish has finished.

(The commit you point to above as having decreased performance fixed this bug, since it became much more visible with decoupled send/receive.)

So: put this change after the switch to syncing on output buffers (since that operation does make sense for this), and leave the existing behaviour for cases where you have to sync on the input surface.

- Mark
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 6+ messages in thread
    • end of thread, other threads:[~2021-12-28  2:05 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
     [not found] <20211027085705.4114165-1-wenbin.chen@intel.com>
     [not found] ` <20211027085705.4114165-3-wenbin.chen@intel.com>
     [not found]   ` <DM4PR11MB52946A90C16BC74C6A6DF492F88A9@DM4PR11MB5294.namprd11.prod.outlook.com>
2021-12-24 23:22     ` [FFmpeg-devel] [PATCH 3/3] libavcodec/vaapi_encode: Add async_depth to vaapi_encoder to increase performance Ed Martin
2021-12-25  5:49       ` Dennis Mungai
2021-12-27  7:22 ` [FFmpeg-devel] [PATCH 1/3] libavcodec/vaapi_encode: Change the way to call async " Xiang, Haihao
2021-12-27  7:41   ` Chen, Wenbin
2021-12-27 18:31 ` Mark Thompson
2021-12-28  2:05   ` Chen, Wenbin

    Git Inbox Mirror of the ffmpeg-devel mailing list - see https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

This inbox may be cloned and mirrored by anyone:

	git clone --mirror https://master.gitmailbox.com/ffmpegdev/0 ffmpegdev/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 ffmpegdev ffmpegdev/ https://master.gitmailbox.com/ffmpegdev \
		ffmpegdev@gitmailbox.com
	public-inbox-index ffmpegdev

Example config snippet for mirrors.


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git