Git Inbox Mirror of the ffmpeg-devel mailing list - see https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
 help / color / mirror / Atom feed
From: cgutman via ffmpeg-devel <ffmpeg-devel@ffmpeg.org>
To: ffmpeg-devel@ffmpeg.org
Cc: cgutman <code@ffmpeg.org>
Subject: [FFmpeg-devel] [PATCH] avcodec/amfenc: avoid unnecessary output delay in low delay mode (PR #20473)
Date: Tue, 09 Sep 2025 05:09:19 -0000
Message-ID: <175739456014.25.2012851308462728431@463a07221176> (raw)

PR #20473 opened by cgutman
URL: https://code.ffmpeg.org/FFmpeg/FFmpeg/pulls/20473
Patch URL: https://code.ffmpeg.org/FFmpeg/FFmpeg/pulls/20473.patch

88a8ba5c991a0ffd00c8ce34d127b1b201b1ee19 introduced an output latency regression making it impossible to acheive lower than 1 frame of latency from AMF. I assume the new behavior was intentional to improve throughput, so I only avoid it if the user explicitly asks for low delay.


>From f3083931ad8b432ecd87dee4bef6895e9e20f9d9 Mon Sep 17 00:00:00 2001
From: Cameron Gutman <aicommander@gmail.com>
Date: Mon, 8 Sep 2025 23:41:43 -0500
Subject: [PATCH] avcodec/amfenc: avoid unnecessary output delay in low delay
 mode

The code optimizes throughput by letting the encoder work on frame N
until frame N+1 is ready for submission, but this hurts low-delay uses
by delaying output by one frame. Don't delay output beyond what is
necessary when AV_CODEC_FLAG_LOW_DELAY is used.

Signed-off-by: Cameron Gutman <aicommander@gmail.com>
---
 libavcodec/amfenc.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/libavcodec/amfenc.c b/libavcodec/amfenc.c
index b16b642e4c..f363192000 100644
--- a/libavcodec/amfenc.c
+++ b/libavcodec/amfenc.c
@@ -497,7 +497,7 @@ static int amf_submit_frame(AVCodecContext *avctx, AVFrame    *frame, AMFSurface
     AMF_RESULT              res;
     int                     ret;
     int                     hw_surface = 0;
-    int                     max_b_frames = ctx->max_b_frames < 0 ? 0 : ctx->max_b_frames;
+    int                     output_delay = FFMAX(ctx->max_b_frames, 0) + ((avctx->flags & AV_CODEC_FLAG_LOW_DELAY) ? 0 : 1);
 
 // prepare surface from frame
     switch (frame->format) {
@@ -634,8 +634,8 @@ static int amf_submit_frame(AVCodecContext *avctx, AVFrame    *frame, AMFSurface
         ret = av_fifo_write(ctx->timestamp_list, &frame->pts, 1);
             if (ret < 0)
             return ret;
-        if(ctx->submitted_frame <= ctx->encoded_frame + max_b_frames + 1)
-            return AVERROR(EAGAIN); // if frame just submiited - don't poll or wait
+        if(ctx->submitted_frame <= ctx->encoded_frame + output_delay)
+            return AVERROR(EAGAIN); // too soon to poll or wait
     }
     return 0;
 }
@@ -681,7 +681,7 @@ int ff_amf_receive_packet(AVCodecContext *avctx, AVPacket *avpkt)
     AVFrame    *frame = av_frame_alloc();
     int         block_and_wait;
     int64_t     pts = 0;
-    int max_b_frames = ctx->max_b_frames < 0 ? 0 : ctx->max_b_frames;
+    int output_delay = FFMAX(ctx->max_b_frames, 0) + ((avctx->flags & AV_CODEC_FLAG_LOW_DELAY) ? 0 : 1);
 
     if (!ctx->encoder){
         av_frame_free(&frame);
@@ -700,7 +700,7 @@ int ff_amf_receive_packet(AVCodecContext *avctx, AVPacket *avpkt)
         if(ret != AVERROR_EOF){
             av_frame_free(&frame);
             if(ret == AVERROR(EAGAIN)){
-                if(ctx->submitted_frame <= ctx->encoded_frame + max_b_frames + 1) // too soon to poll
+                if(ctx->submitted_frame <= ctx->encoded_frame + output_delay) // too soon to poll
                     return ret;
             }
         }
-- 
2.49.1

_______________________________________________
ffmpeg-devel mailing list -- ffmpeg-devel@ffmpeg.org
To unsubscribe send an email to ffmpeg-devel-leave@ffmpeg.org

                 reply	other threads:[~2025-09-09  5:09 UTC|newest]

Thread overview: [no followups] expand[flat|nested]  mbox.gz  Atom feed

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=175739456014.25.2012851308462728431@463a07221176 \
    --to=ffmpeg-devel@ffmpeg.org \
    --cc=code@ffmpeg.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

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