Git Inbox Mirror of the ffmpeg-devel mailing list - see https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
 help / color / mirror / Atom feed
From: Michael Diesel <not.michael.diesel@gmail.com>
To: ffmpeg-devel@ffmpeg.org
Cc: r.arzumanyan@visionlabs.ai, timo@rothenpieler.org
Subject: Re: [FFmpeg-devel] [PATCH] avcodec/cuviddec: update amount of decoder surfaces from within sequence decode callback
Date: Wed, 9 Aug 2023 00:55:09 +0300
Message-ID: <9d6b8529-2623-4a8c-a5d1-6f72f709b37f@gmail.com> (raw)
In-Reply-To: <4b2e5f71-f8b0-8abf-d3bc-daea4fbe3134@rothenpieler.org>

Hello ,
after this patch I get stutter in the output video when using h264_cuvid 
as decoder

With "surfaces" as -1 will result :
     CUVIDPARSERPARAMS::ulMaxNumDecodeSurfaces = 5 = 
CUVID_DEFAULT_NUM_SURFACES
     CUVIDDECODECREATEINFO::ulNumDecodeSurfaces = 8 = 
format->min_num_decode_surfaces + 3

As per NVDEC Video Decoder API Programming Guide / Creating a parser , 
the two values should be equal.

cuvid_handle_video_sequence/pfnSequenceCallback should return a value :
">1: succeeded, and driver should override 
CUVIDPARSERPARAMS::ulMaxNumDecodeSurfaces with this return value"

Something like:
     if(ctx->cuparseinfo.ulMaxNumDecodeSurfaces != 
cuinfo.ulNumDecodeSurfaces)
         return cuinfo.ulNumDecodeSurfaces;
     else
         return 1;

This solves partially the problem , there's still occasional stutter , a 
value of 8 is not sufficient.
On my test system 12 is the magic number.

Timo Rothenpieler wrote:
> applied with minor amendments.
> 
> Turned the if-statements around, we don't do them that way around in 
> ffmpeg code.
> And switched the surfaces option to have a default of -1, to not rely on 
> a magic value of 25. That could surprise users who happen to request 25 
> surfaces.

_______________________________________________
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".

      reply	other threads:[~2023-08-08 21:55 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <CAJZRZVnbFVOQaUuvA=F7qUmr+P_wfrjJnga2cq=tdNtm4Ky0sQ@mail.gmail.com>
2023-06-05  7:30 ` Roman Arzumanyan
2023-06-05 10:19   ` Anton Khirnov
2023-06-05 12:29     ` Roman Arzumanyan
2023-06-06 13:35       ` Timo Rothenpieler
2023-08-08 21:55         ` Michael Diesel [this message]

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=9d6b8529-2623-4a8c-a5d1-6f72f709b37f@gmail.com \
    --to=not.michael.diesel@gmail.com \
    --cc=ffmpeg-devel@ffmpeg.org \
    --cc=r.arzumanyan@visionlabs.ai \
    --cc=timo@rothenpieler.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