Re: [FFmpeg-devel] [PATCH] lavc/libvpxenc: prevent fifo from filling up

From: James Zern <jzern-at-google.com@ffmpeg.org>
To: FFmpeg development discussions and patches <ffmpeg-devel@ffmpeg.org>
Subject: Re: [FFmpeg-devel] [PATCH] lavc/libvpxenc: prevent fifo from filling up
Date: Fri, 7 Jul 2023 09:42:15 -0700
Message-ID: <CABWgkXKk9ZVHTqODHpCtOQfCCsTCf30DXrDeGnSCdnNThL9hNg@mail.gmail.com> (raw)
In-Reply-To: <1700748487.794928.1688610127686@privateemail.com>

On Wed, Jul 5, 2023 at 7:22 PM David Lemler <david@lemler.family> wrote:
>
> > On 07/05/2023 2:16 PM CDT James Zern <jzern-at-google.com@ffmpeg.org> wrote:
> >
> >
> > On Wed, Jul 5, 2023 at 12:15 PM James Zern <jzern@google.com> wrote:
> > >
> > > Hi,
> > >
> >
> > +ffmpeg-dev. I took the wrong email off the reply.
> >
> > > On Mon, Jul 3, 2023 at 10:08 PM David Lemler <david@lemler.family> wrote:
> > > >
> > > > Prevent the fifo used in encoding VPx videos from filling up and stopping
> > > > encode when it reaches 21845 items, which happens when the video has more
> > > > than
> > > > that number of frames.
> > > >
> > >
> > > What failure do you see, a memory allocation error? 21845 sounds
> > > somewhat arbitrary, maybe specific to your machine, so I'd leave it
> > > off the comment and commit message.
> > >
>
> When running a command like the following to perform the first pass of a
> 2-pass VP9 encode:
>
> ffmpeg -i input.mp4 -c:v libvpx-vp9 -b:v 0 -crf 31 -pass 1 -an \
> -deadline best -threads 8 -row-mt 1 -f null /dev/null
>
> ffmpeg exited with the error:
> Error submitting video frame to the encoder
> Conversion failed!
>
> and the ffmpeg2pass-0.log file existed but was empty.
>
> I've tested and reproduced the issue on Intel x86-64 on Windows
> (10 22H2, gyan.dev build) and Linux (Debian 12, own compiled build), and on
> AMD EPYC x86-64 on Linux (Debian 12, own compiled build).  I believe the
> number of frames is (1024 * 1024) / sizeof(FrameData); see lavc/libvpxenc.c
> the call to av_fifo_alloc2 around line 999 and lavutil/fifo.c the assignment
> of f->auto_grow_limit around line 72.  I'm not sure if that calculation
> would change on different systems/architectures.
>

Thanks for the explanation, I didn't look closely enough before making
my comment.
_______________________________________________
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".