From 86c30e327e1eb8ba913d74d5394ea90a87b55a69 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tomas=20H=C3=A4rdin?=
Date: Fri, 10 Jun 2022 14:12:11 +0200
Subject: [PATCH 02/13] lavc/jpeg2000dec: Reindent
---
libavcodec/jpeg2000dec.c | 36 ++--
1 file changed, 18 inserti
From 6ab67531c946ca320e49bc93f4f086835ffd2c1c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tomas=20H=C3=A4rdin?=
Date: Fri, 10 Jun 2022 17:18:14 +0200
Subject: [PATCH 03/13] lavc/jpeg2000dwt: Implement sliced transforms
lavc/tests/jpeg2000dwt tests this.
---
libavcodec/j2kenc.c| 3 +-
From d0ec602b0f61dd7f8d53efccc2c4859058a5d55d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tomas=20H=C3=A4rdin?=
Date: Mon, 13 Jun 2022 14:45:07 +0200
Subject: [PATCH 04/13] lavc/jpeg2000dec: Implement IDWT slicing
---
libavcodec/jpeg2000dec.c | 99 +++-
libavcod
From 080ebdc9bad130098bff575f9ce690b8a522c9f7 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tomas=20H=C3=A4rdin?=
Date: Mon, 13 Jun 2022 15:09:17 +0200
Subject: [PATCH 05/13] lavc/jpeg2000dec: Thread init_tile()
---
libavcodec/jpeg2000dec.c | 30 +++---
1 file changed, 15 ins
Left this as an ff_ funtion for now since it's only used by the j2k
code
/Tomas
From 5d36d431ffe4c8ba0f698d0c288ebc16b83f0bbc Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tomas=20H=C3=A4rdin?=
Date: Tue, 14 Jun 2022 13:35:18 +0200
Subject: [PATCH 06/13] lavu/mem: Add ff_fast_recalloc()
---
libavuti
From 72a5f47503338a4fff816440ad64bc62cc23a738 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tomas=20H=C3=A4rdin?=
Date: Mon, 13 Jun 2022 17:04:10 +0200
Subject: [PATCH 07/13] lavc/jpeg2000*: Use ff_fast_recalloc() to eliminate
lots of allocations
---
libavcodec/jpeg2000.c| 72 ++
From c0e00cf03f5a1fcffc90395d4b26607e1681690c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tomas=20H=C3=A4rdin?=
Date: Tue, 7 Jun 2022 16:43:40 +0200
Subject: [PATCH 08/13] lavc/jpeg2000: Switch Jpeg2000TgtNode to int32_t parent
---
libavcodec/j2kenc.c | 44
From 03b806f89453571310dcb14edbd9f51e059b7476 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tomas=20H=C3=A4rdin?=
Date: Wed, 8 Jun 2022 10:08:15 +0200
Subject: [PATCH 09/13] lavc/jpeg2000: Speed up ff_jpeg2000_tag_tree_init()
using stereotypes for sizes <= 4x4
---
libavcodec/jpeg2000.c | 35 +++
From d3aaf24ca4778e6ba280f99f9ce90cb15738699b Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tomas=20H=C3=A4rdin?=
Date: Tue, 14 Jun 2022 11:23:08 +0200
Subject: [PATCH 10/13] lavc/jpeg2000: Reindent
---
libavcodec/jpeg2000.c | 44 +--
1 file changed, 22 insert
From 5b492d4e92a11946fd7425497205b1842fa1912c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tomas=20H=C3=A4rdin?=
Date: Tue, 14 Jun 2022 10:57:45 +0200
Subject: [PATCH 11/13] lavc/jpeg2000: Minimize calls to av_codec_is_encoder()
---
libavcodec/jpeg2000.c | 19 +++
1 file changed, 11
From 15761070d1cdc622ffbc5d6aeb0a50e063361012 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tomas=20H=C3=A4rdin?=
Date: Tue, 14 Jun 2022 11:19:06 +0200
Subject: [PATCH 12/13] lavc/jpeg2000dec: Use coarser slicing for initial
reslevels
This brings -lowres 2 lossless 4K J2K on an AMD EPYC 7R32 to 52 f
Don't have access to the full machine to test this with 96 threads. On
2/3rds of an AMD EPYC 7R32 (-threads 64) it runs at 50 fps.
Specifically the decoder uses 59.2 seconds to decode a 60.0 second
clip.
/Tomas
From 19fc2413dc2bafff577c68830cde48e08138771e Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?
Tomas Härdin:
>
>
Why call it at all? Why not just add a new parameter to
ff_jpeg2000_init_component that is always set to 1 when called from the
encoder and 0 when called from the decoder?
(And is this really a bottleneck?)
- Andreas
___
ffmpeg-devel
Tomas Härdin:
>
>
> @@ -2166,12 +2163,13 @@ static int jpeg2000_mct_write_frame(AVCodecContext
> *avctx, void *td,
> return 0;
> }
>
> -static void jpeg2000_dec_cleanup(Jpeg2000DecoderContext *s)
> +static void jpeg2000_dec_cleanup(Jpeg2000DecoderContext *s, int close)
> {
> int ti
On Tue, 14 Jun 2022, Gyan Doshi wrote:
On 2022-06-14 01:51 pm, Marton Balint wrote:
On Tue, 14 Jun 2022, Gyan Doshi wrote:
Add new flag AVFMT_START_REALTIME for the couple of demuxers that set it
internally.
I don't think this is the same semantics.
start_time_realtime is the t
On Tue, Jun 14, 2022 at 04:42:06PM +0200, Tomas Härdin wrote:
> Left this as an ff_ funtion for now since it's only used by the j2k
> code
>
> /Tomas
> mem.c | 24
> mem.h | 55 +++
> 2 files changed, 79 insertions(
On Tue, Jun 14, 2022 at 04:41:14PM +0200, Tomas Härdin wrote:
>
> jpeg2000dec.c | 30 +++---
> 1 file changed, 15 insertions(+), 15 deletions(-)
> 6fa2fbf99afee36ee73459863df0527a72663f43
> 0005-lavc-jpeg2000dec-Thread-init_tile.patch
> From 080ebdc9bad130098bff575f9c
Gyan Doshi:
> Add new flag AVFMT_START_REALTIME for the couple of demuxers that set it
> internally.
> ---
> doc/APIchanges | 3 +++
> libavformat/avformat.h | 1 +
> libavformat/demux.c| 3 +++
> libavformat/rtsp.c | 3 ++-
> libavformat/version.h | 2 +-
> 5 files changed, 10 in
> > -Original Message-
> > From: ffmpeg-devel On Behalf Of
> > Xiang, Haihao
> > Sent: Thursday, June 9, 2022 8:48 AM
> > To: ffmpeg-devel@ffmpeg.org
> > Cc: Wu, Tong1 ; Chen, Wenbin
> >
> > Subject: Re: [FFmpeg-devel] [PATCH] avcodec/qsvenc: make QSV encoder
> > encode VAAPI and D3D11 fr
> Quoting Wu, Tong1 (2022-06-08 06:47:27)
> > >
> > > Quoting Tong Wu (2022-06-07 11:22:16)
> > > > QSV encoder is able to encode frames with VAAPI or D3D11 pixel
> > > > format directly. This patch adds support for qsv encoder to accept
> > > > VAAPI and
> > > > D3D11 pixel formats as input.
> > >
On 2022-06-15 03:18 am, Andreas Rheinhardt wrote:
Gyan Doshi:
Add new flag AVFMT_START_REALTIME for the couple of demuxers that set it
internally.
---
doc/APIchanges | 3 +++
libavformat/avformat.h | 1 +
libavformat/demux.c| 3 +++
libavformat/rtsp.c | 3 ++-
libavforma
Gyan Doshi:
>
>
> On 2022-06-15 03:18 am, Andreas Rheinhardt wrote:
>> Gyan Doshi:
>>> Add new flag AVFMT_START_REALTIME for the couple of demuxers that set it
>>> internally.
>>> ---
>>> doc/APIchanges | 3 +++
>>> libavformat/avformat.h | 1 +
>>> libavformat/demux.c | 3 +++
>>>
For SSE2 and SSE3, there are four states that the two flags
involved (AV_CPU_FLAG_SSE[23] and AV_CPU_FLAG_SSE[23]SLOW) can convey.
When ordered from worst to best they are:
1. both flags unset (SSE[23] unavailable)
2. the slow flag set, the ordinary flag unset (this is designed
for cases where SSE2
EXTERNAL_SSE2_SLOW is now more encompassing than EXTERNAL_SSE2.
Signed-off-by: Andreas Rheinhardt
---
libavcodec/x86/lpc.c | 2 +-
libavcodec/x86/vp8dsp_init.c | 4 ++--
2 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/libavcodec/x86/lpc.c b/libavcodec/x86/lpc.c
index 54408
Ffmpeg/ffprobe/ffplay sets scan_all_pmts to 1 when finding the streams, that
should be enough to handle files for which some early PMTs miss some streams.
Fixes ticket #9782.
Signed-off-by: Marton Balint
---
libavformat/mpegts.c | 12 ++--
1 file changed, 2 insertions(+), 10 deletions(-
It is no longer converted since mkv_write_chapters() is called
before mkv_write_tags() which happens since commit
4ebfc13c338423cf48f1a1266c890422367f7775. Given the fact that
chapters can also be written late, mkv_write_chapters() has to
convert the metadata itself.
Fixes ticket #9812.
Signed-of
Add new flag AVFMT_START_REALTIME for the couple of demuxers that set it
internally.
---
doc/APIchanges | 3 +++
libavformat/avformat.h | 1 +
libavformat/demux.c| 3 +++
libavformat/rtsp.c | 3 ++-
libavformat/version.h | 2 +-
5 files changed, 10 insertions(+), 2 deletions(-)
d
On Tue, 14 Jun 2022, Gyan Doshi wrote:
Add new flag AVFMT_START_REALTIME for the couple of demuxers that set it
internally.
I don't think this is the same semantics.
start_time_realtime is the time when the stream was originally captured,
and not when the stream was received. So you can't
On 2022-06-14 01:51 pm, Marton Balint wrote:
On Tue, 14 Jun 2022, Gyan Doshi wrote:
Add new flag AVFMT_START_REALTIME for the couple of demuxers that set it
internally.
I don't think this is the same semantics.
start_time_realtime is the time when the stream was originally
captured, an
---
doc/APIchanges | 2 ++
1 file changed, 2 insertions(+)
diff --git a/doc/APIchanges b/doc/APIchanges
index 5857e67ae6..20b944933a 100644
--- a/doc/APIchanges
+++ b/doc/APIchanges
@@ -96,6 +96,8 @@ API changes, most recent first:
2022-01-26 - af94ab7c7c0 - lavu 57.19.100 - tx.h
Add AV_TX_FL
Andreas Rheinhardt:
> This is more spec-compliant because it does not rely
> on dead-code elimination by the compiler. Especially
> MSVC has problems with this, as can be seen in
> https://ffmpeg.org/pipermail/ffmpeg-devel/2022-May/296373.html
> or
> https://ffmpeg.org/pipermail/ffmpeg-devel/2022-M
Patch 12 in this series is optional since it's just me getting the
speed up on a specific machine
/Tomas
From 115aa26c343419e81c1b5ba0bfdb1615cbec27e9 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tomas=20H=C3=A4rdin?=
Date: Fri, 10 Jun 2022 14:10:02 +0200
Subject: [PATCH 01/13] lavc/jpeg2000dec: Fine
32 matches
Mail list logo
