On Mon, 25 Jul 2022, Hubert Mazur wrote:
Provide neon implementation for sse16 function.
Performance comparison tests are shown below.
- sse_0_c: 273.0
- sse_0_neon: 48.2
Benchmarks and tests run with checkasm tool on AWS Graviton 3.
Signed-off-by: Hubert Mazur
---
libavcodec/aarch64/me_cmp_
On Wed, 13 Jul 2022, Martin Storsjö wrote:
The height is hardcoded in some of the me_cmp functions, but not
in all of them. But in the case of all other functions, it's hardcoded
in the same place in SIMD functions as in the C reference functions,
while this one function differs from the behavio
On Wed, 13 Jul 2022, Martin Storsjö wrote:
Previously, the checkasm test always passed h=8, so no other cases
were tested.
Out of the me_cmp functions, in practice, some functions are hardcoded
to always assume a 8x8 block (ignoring the h parameter), while others
do use the parameter. For those
On Mon, 25 Jul 2022, Hubert Mazur wrote:
Provide neon implementation for sse4 function.
Performance comparison tests are shown below.
- sse_2_c: 74.0
- sse_2_neon: 24.0
Benchmarks and tests are run with checkasm tool on AWS Graviton 3.
Signed-off-by: Hubert Mazur
---
libavcodec/aarch64/me_cm
On Mon, 25 Jul 2022, Hubert Mazur wrote:
Provide optimized implementation of sse8 function for arm64.
Performance comparison tests are shown below.
- sse_1_c: 133.0
- sse_1_neon: 36.7
Benchmarks and tests run with checkasm tool on AWS Graviton 3.
Signed-off-by: Hubert Mazur
---
libavcodec/aa
On Mon, 25 Jul 2022, Hubert Mazur wrote:
Provide optimized implementation of pix_abs16_y2 function for arm64.
Performance comparison tests are shown below.
pix_abs_0_2_c: 308.5
pix_abs_0_2_neon: 39.2
Benchmarks and tests run with checkasm tool on AWS Graviton 3.
Signed-off-by: Hubert Mazur
-
On Mon, 25 Jul 2022, Hubert Mazur wrote:
Provide optimized implementation of pix_abs8 function for arm64.
Performance comparison tests are shown below.
- pix_abs_1_0_c: 105.2
- pix_abs_1_0_neon: 21.4
- sad_1_c: 107.2
- sad_1_neon: 20.9
Benchmarks and tests are run with checkasm tool on AWS Gra
On Mon, 25 Jul 2022, Hubert Mazur wrote:
Provide optimized implementation of pix_abs16_y2 function for arm64.
Performance comparison tests are shown below.
pix_abs_0_2_c: 308.5
pix_abs_0_2_neon: 39.2
Benchmarks and tests run with checkasm tool on AWS Graviton 3.
Signed-off-by: Hubert Mazur
-
Quoting Andreas Rheinhardt (2022-08-03 20:47:36)
> Missing const on st
src/fftools/ffmpeg_demux.c:41:46: warning: passing argument 3 of
‘av_ts_make_time_string’ discards ‘const’ qualifier from pointer target type
[-Wdiscarded-qualifiers]
41 |pkt->pos, av_ts2timestr(pkt->dts, &st->
Anton Khirnov (12022-08-04):
> src/fftools/ffmpeg_demux.c:41:46: warning: passing argument 3 of
> ‘av_ts_make_time_string’ discards ‘const’ qualifier from pointer target type
> [-Wdiscarded-qualifiers]
>41 |pkt->pos, av_ts2timestr(pkt->dts, &st->time_base));
av_ts_make_time_strin
Nicolas George:
> Anton Khirnov (12022-08-04):
>> src/fftools/ffmpeg_demux.c:41:46: warning: passing argument 3 of
>> ‘av_ts_make_time_string’ discards ‘const’ qualifier from pointer target type
>> [-Wdiscarded-qualifiers]
>>41 |pkt->pos, av_ts2timestr(pkt->dts, &st->time_base));
Andreas Rheinhardt (12022-08-04):
> Or the timebase should be passed by value (as is commonly done elsewhere).
That would have been the best initially. But it is a public function,
changing it would be an API break, probably not worth the deprecation
dance.
Regards,
--
Nicolas George
signat
DNN OpenVINO backend would not report missing model file if it does not
exist. It would corrupt directly with out any error infomation. This commit
would check both .xml and .bin file existance before loading model.
Signed-off-by: Ting Fu
---
libavfilter/dnn/dnn_backend_openvino.c | 20 +
Fix the break when used libavcodec/Makefile
Signed-off-by: Jun Zhao
---
libavcodec/Makefile | 2 ++
1 file changed, 2 insertions(+)
diff --git a/libavcodec/Makefile b/libavcodec/Makefile
index 836132f430..5ba7fb6b86 100644
--- a/libavcodec/Makefile
+++ b/libavcodec/Makefile
@@ -1234,6 +1234,7 @
Jun Zhao:
> Fix the break when used libavcodec/Makefile
>
> Signed-off-by: Jun Zhao
> ---
> libavcodec/Makefile | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/libavcodec/Makefile b/libavcodec/Makefile
> index 836132f430..5ba7fb6b86 100644
> --- a/libavcodec/Makefile
> +++ b/libavcod
Quoting Ting Fu (2022-08-04 11:31:01)
> DNN OpenVINO backend would not report missing model file if it does not
> exist. It would corrupt directly with out any error infomation. This commit
"corrupt"?
The patch looks completely wrong. Testing for file existence explicitly
is known to be a bad pat
Looks reasonable, will push soonish if nobody objects.
--
Anton Khirnov
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
To unsubscribe, visit link above, or email
ffmpeg-devel-requ...@ffmpeg.org wi
From: Zhao Zhili
Signed-off-by: Zhao Zhili
---
libavutil/hwcontext_videotoolbox.h | 1 +
1 file changed, 1 insertion(+)
diff --git a/libavutil/hwcontext_videotoolbox.h
b/libavutil/hwcontext_videotoolbox.h
index b6bdcf243d..25dde85df5 100644
--- a/libavutil/hwcontext_videotoolbox.h
+++ b/libav
replace the unicode left/right signal quotation mark with normal
signal quotation.
Signed-off-by: Jun Zhao
---
libavformat/hevc.c | 10 +-
1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/libavformat/hevc.c b/libavformat/hevc.c
index df141aec26..37d97941d5 100644
--- a/libav
On Thu, 14 Jul 2022, Martin Storsjö wrote:
The AArch64 assembly accesses those symbols directly, without
indirection via e.g. the GOT on ELF. In order for this not to
require text relocations, those symbols need to be resolved fully
at link time, i.e. those symbols can't be interposable.
Normal
On Wed, 27 Jul 2022, Thomas Guillem wrote:
DECLARE_ALIGNED, DECLARE_ASM_ALIGNED, and DECLARE_ASM_CONST will include
attribute_visibility_hidden.
Hmm, I'm not entirely sure that we should do that - if we should add such
extra meaning to those macros.
How many symbols would it need to be appl
On Fri, 22 Jul 2022, Swinney, Jonathan wrote:
This specialization handles the case where filtersize is 4 mod 8, e.g.
12, 20, etc. Aarch64 was previously using the c function for this case.
This implementation speeds up that case significantly.
hscale_8_to_15__fs_12_dstW_512_c: 6234.1
hscale_8_t
On Wed, 27 Jul 2022, Swinney, Jonathan wrote:
- added a test for yuv2plane1
- fixed test for yuv2planeX for aarch64 which was previously not working
at all
- updated the test for yuv2planeX to check exact results or approximated
results
Signed-off-by: Jonathan Swinney
---
libswscale/x86/swsc
On Wed, 27 Jul 2022, Swinney, Jonathan wrote:
This commit adds new code paths for vscale when filterSize is 2, 4, or
8. By using specialized code with unrolling to match the filterSize we
can improve performance.
This patch also corrects the behavor for
filterSize 1 which was previously faili
Dear FFmpeg development community,
My name is Tudor Zaharia and I am sending this message after reading this page:
https://ffmpeg.org/consulting.html
For a project that I am implementing, I would like to have H264 hardware
encoding capabilities in libwebrtc through Nvidia's nvenc. My understadi
I don't know if the mailing list includes my contact details, so here is my
email address: tudor(at)swisstch[dot]com
Best regards,
Tudor Zaharia
From: ffmpeg-devel on behalf of Tudor Zaharia
Sent: Thursday, August 4, 2022 2:22 PM
To: ffmpeg-devel@ffmpeg.org
Su
On Wed, Aug 03, 2022 at 03:58:26PM +0200, Anton Khirnov wrote:
> There are currently three possible modes for an output stream:
> 1) The stream is produced by encoding output from some filtergraph. This
>is true when ost->enc_ctx != NULL, or equivalently when
>ost->encoding_needed != 0.
> 2
Thomas Guillem:
> DECLARE_ALIGNED, DECLARE_ASM_ALIGNED, and DECLARE_ASM_CONST will include
> attribute_visibility_hidden.
>
> ld: error: relocation R_X86_64_PC32 cannot be used against symbol
> 'ff_h264_cabac_tables'; recompile with -fPIC
defined in
/home/tom/work/git/vlc-android/vlc/c
Michael Niedermayer:
> On Wed, Aug 03, 2022 at 03:58:26PM +0200, Anton Khirnov wrote:
>> There are currently three possible modes for an output stream:
>> 1) The stream is produced by encoding output from some filtergraph. This
>>is true when ost->enc_ctx != NULL, or equivalently when
>>ost
On Tue, Aug 02, 2022 at 06:54:15PM +0200, Nicolas George wrote:
> Signed-off-by: Nicolas George
> ---
> libavutil/pixfmt.h | 17 +
> 1 file changed, 17 insertions(+)
>
>
> Identical to the one I posted last week.
>
>
> diff --git a/libavutil/pixfmt.h b/libavutil/pixfmt.h
> ind
On Tue, Aug 02, 2022 at 06:54:20PM +0200, Nicolas George wrote:
> Signed-off-by: Nicolas George
> ---
> libavutil/pixdesc.c | 6 ++
> 1 file changed, 6 insertions(+)
probably ok
thx
[...]
--
Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
The smallest minority on
Michael Niedermayer (12022-08-04):
> You seem to describe samples as rectangular areas of constant value IIUC.
> If you look at the ITU/ISO specs (mpeg2, h264 or others) they are described
> by point samples. The desity of samples matches. While the default locations
> do not.
> What you list above
Quoting Andreas Rheinhardt (2022-08-04 14:54:41)
> Michael Niedermayer:
> > On Wed, Aug 03, 2022 at 03:58:26PM +0200, Anton Khirnov wrote:
> >> There are currently three possible modes for an output stream:
> >> 1) The stream is produced by encoding output from some filtergraph. This
> >>is tru
Anton Khirnov:
> Quoting Andreas Rheinhardt (2022-08-04 14:54:41)
>> Michael Niedermayer:
>>> On Wed, Aug 03, 2022 at 03:58:26PM +0200, Anton Khirnov wrote:
There are currently three possible modes for an output stream:
1) The stream is produced by encoding output from some filtergraph. T
On Thu, Aug 04, 2022 at 03:57:27PM +0200, Nicolas George wrote:
> Michael Niedermayer (12022-08-04):
> > You seem to describe samples as rectangular areas of constant value IIUC.
> > If you look at the ITU/ISO specs (mpeg2, h264 or others) they are described
> > by point samples. The desity of samp
On Thu, Aug 04, 2022 at 10:47:34AM +0300, Martin Storsjö wrote:
> On Wed, 13 Jul 2022, Martin Storsjö wrote:
>
> > The height is hardcoded in some of the me_cmp functions, but not
> > in all of them. But in the case of all other functions, it's hardcoded
> > in the same place in SIMD functions as
p...@sandflow.com:
> From: Pierre-Anthony Lemieux
>
> As discussed at
> https://ffmpeg.org/pipermail/ffmpeg-devel/2022-July/298491.html.
> Note that ff_stream_params_copy() does not copy:
> * the index field
> * the attached_pic if its size is 0
>
> Addresses http://ffmpeg.org/pipermail/ffmpe
On Thu, Aug 4, 2022 at 9:53 AM Andreas Rheinhardt
wrote:
>
> p...@sandflow.com:
> > From: Pierre-Anthony Lemieux
> >
> > As discussed at
> > https://ffmpeg.org/pipermail/ffmpeg-devel/2022-July/298491.html.
> > Note that ff_stream_params_copy() does not copy:
> > * the index field
> > * the att
Pierre-Anthony Lemieux:
> On Thu, Aug 4, 2022 at 9:53 AM Andreas Rheinhardt
> wrote:
>>
>> p...@sandflow.com:
>>> From: Pierre-Anthony Lemieux
>>>
>>> As discussed at
>>> https://ffmpeg.org/pipermail/ffmpeg-devel/2022-July/298491.html.
>>> Note that ff_stream_params_copy() does not copy:
>>> *
These buffers are not shared in any way.
Signed-off-by: Andreas Rheinhardt
---
libavcodec/error_resilience.c | 18 +-
libavcodec/error_resilience.h | 4 ++--
2 files changed, 11 insertions(+), 11 deletions(-)
diff --git a/libavcodec/error_resilience.c b/libavcodec/error_resilie
Signed-off-by: Andreas Rheinhardt
---
libavcodec/cbs.c | 74
1 file changed, 25 insertions(+), 49 deletions(-)
diff --git a/libavcodec/cbs.c b/libavcodec/cbs.c
index 8c0a5b5719..f715e463a1 100644
--- a/libavcodec/cbs.c
+++ b/libavcodec/cbs.c
@@ -9
The code just creates new references without allocating
new buffers for the subobjects; therefore the actual data pointer
stays valid and need not be updated.
Also remove an assert that ensured that the calculation
for updating the pointer makes sense.
Signed-off-by: Andreas Rheinhardt
---
liba
And also avoid an unnecessary indirection for src_buf.
Signed-off-by: Andreas Rheinhardt
---
libavcodec/cbs.c | 19 +--
1 file changed, 9 insertions(+), 10 deletions(-)
diff --git a/libavcodec/cbs.c b/libavcodec/cbs.c
index 43329a14a4..c81297ec93 100644
--- a/libavcodec/cbs.c
++
Signed-off-by: Andreas Rheinhardt
---
libavcodec/cbs_h2645.c| 56 ---
libavcodec/cbs_internal.h | 31 +-
2 files changed, 35 insertions(+), 52 deletions(-)
diff --git a/libavcodec/cbs_h2645.c b/libavcodec/cbs_h2645.c
index 12e38c80b5..1
Signed-off-by: Andreas Rheinhardt
---
libavcodec/cbs.c | 26 +++
libavcodec/cbs_internal.h | 69 ++-
libavcodec/cbs_mpeg2.c| 10 +++---
3 files changed, 57 insertions(+), 48 deletions(-)
diff --git a/libavcodec/cbs.c b/libavcodec/cbs.c
It is no longer used.
Also rename ff_cbs_alloc_unit_content2 to ff_cbs_alloc_unit_content.
Signed-off-by: Andreas Rheinhardt
---
libavcodec/cbs.c | 24 ++--
libavcodec/cbs.h | 13 ++---
libavcodec/cbs_av1.c | 2 +-
libavcodec/cbs_h2645.c | 4 ++--
liba
The functions to replace parameter sets are only called
after the respective parameter set has just been read or
has just been written; all of these functions check
that the id field is within the appropriate range.
So the checks in the replace-functions can be removed.
Signed-off-by: Andreas Rhei
cbs_jpeg was the last user of CBS that didn't use
CodedBitstreamUnitTypeDescriptors.
Signed-off-by: Andreas Rheinhardt
---
libavcodec/cbs_internal.h | 7
libavcodec/cbs_jpeg.c | 76 ++-
2 files changed, 34 insertions(+), 49 deletions(-)
diff --git a
On Wed, Aug 03, 2022 at 01:02:47PM -0300, James Almer wrote:
> Signed-off-by: James Almer
> ---
> libavcodec/pthread_frame.c | 30 ++
> libavcodec/threadframe.h | 3 +++
> libavcodec/utils.c | 21 +
> 3 files changed, 54 insertions(+)
[..
On 8/4/2022 3:33 PM, Michael Niedermayer wrote:
On Wed, Aug 03, 2022 at 01:02:47PM -0300, James Almer wrote:
Signed-off-by: James Almer
---
libavcodec/pthread_frame.c | 30 ++
libavcodec/threadframe.h | 3 +++
libavcodec/utils.c | 21 ++
On Thu, Aug 4, 2022 at 10:15 AM Andreas Rheinhardt
wrote:
>
> Pierre-Anthony Lemieux:
> > On Thu, Aug 4, 2022 at 9:53 AM Andreas Rheinhardt
> > wrote:
> >>
> >> p...@sandflow.com:
> >>> From: Pierre-Anthony Lemieux
> >>>
> >>> As discussed at
> >>> https://ffmpeg.org/pipermail/ffmpeg-devel/2022
On Thu, 4 Aug 2022, Michael Niedermayer wrote:
On Thu, Aug 04, 2022 at 10:47:34AM +0300, Martin Storsjö wrote:
On Wed, 13 Jul 2022, Martin Storsjö wrote:
The height is hardcoded in some of the me_cmp functions, but not
in all of them. But in the case of all other functions, it's hardcoded
in
Signed-off-by: James Almer
---
libswscale/input.c | 35 +++
libswscale/utils.c | 1 +
2 files changed, 36 insertions(+)
diff --git a/libswscale/input.c b/libswscale/input.c
index 750367b28b..68abc4d62c 100644
--- a/libswscale/input.c
+++ b/libswscale/input.c
@@ -
Pierre-Anthony Lemieux:
> On Thu, Aug 4, 2022 at 10:15 AM Andreas Rheinhardt
> wrote:
>>
>> Pierre-Anthony Lemieux:
>>> On Thu, Aug 4, 2022 at 9:53 AM Andreas Rheinhardt
>>> wrote:
p...@sandflow.com:
> From: Pierre-Anthony Lemieux
>
> As discussed at
> https://ffmpeg.o
Some NLVEs may insert a KLV packet for EIA-608 data even though
the number of encapsulated ANC packets is zero.
---
libavformat/mxfdec.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/libavformat/mxfdec.c b/libavformat/mxfdec.c
index 400941c348..f3d2cf4efa 100644
--- a/libavfo
On Thu, Aug 4, 2022 at 3:22 PM Andreas Rheinhardt
wrote:
>
> Pierre-Anthony Lemieux:
> > On Thu, Aug 4, 2022 at 10:15 AM Andreas Rheinhardt
> > wrote:
> >>
> >> Pierre-Anthony Lemieux:
> >>> On Thu, Aug 4, 2022 at 9:53 AM Andreas Rheinhardt
> >>> wrote:
>
> p...@sandflow.com:
> > F
Pierre-Anthony Lemieux:
> On Thu, Aug 4, 2022 at 3:22 PM Andreas Rheinhardt
> wrote:
>>
>> Pierre-Anthony Lemieux:
>>> On Thu, Aug 4, 2022 at 10:15 AM Andreas Rheinhardt
>>> wrote:
Pierre-Anthony Lemieux:
> On Thu, Aug 4, 2022 at 9:53 AM Andreas Rheinhardt
> wrote:
>>
>
On Thu, Aug 4, 2022 at 4:13 PM Andreas Rheinhardt
wrote:
>
> Pierre-Anthony Lemieux:
> > On Thu, Aug 4, 2022 at 3:22 PM Andreas Rheinhardt
> > wrote:
> >>
> >> Pierre-Anthony Lemieux:
> >>> On Thu, Aug 4, 2022 at 10:15 AM Andreas Rheinhardt
> >>> wrote:
>
> Pierre-Anthony Lemieux:
> >>
Andreas Rheinhardt:
> It has been deprecated in 94d68a41fabb55dd8c7e59b88fe4a28a637d1e5f
> and can't be set via AVOptions. The only codecs that use it
> (the MPEG-1/2 encoders) have private options for this.
> So remove it.
>
> Signed-off-by: Andreas Rheinhardt
> ---
> libavcodec/avcodec.h | 5
Hi Anton,
Thank you for comment.
After double checked the OpenVINO, it is true that the code would corrupt if
the binary file does not exist.
We would have nothing to do in this case, that's why I code to check the file
existence explicitly.
Yes, you are right, it is not a proper way to do like
On Thu, 2022-08-04 at 20:59 -0700, Philip Langdale wrote:
> vaapi_decode_find_best_format currently does not set the
> VA_SURFACE_ATTRIB_SETTABLE flag on the pixel format attribute that it
> returns.
>
> Without this flag, the attribute will be ignored by vaCreateSurfaces,
> meaning that the drive
Signed-off-by: Fei Wang
---
update:
1. refine commit message and content in doc file.
doc/filters.texi | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/doc/filters.texi b/doc/filters.texi
index 4a5412c91d..a66b750d57 100644
--- a/doc/filters.texi
+++ b/doc/filters.texi
@@
The present default value of 0 will render the overlay video invisible.
A default of 1.0 is consistent with most common use cases.
Signed-off-by: Fei Wang
---
doc/filters.texi | 2 +-
libavfilter/vf_overlay_vaapi.c | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --gi
On Wed, 2022-07-27 at 18:51 +0530, Gyan Doshi wrote:
> Typo in title.
>
>
>
> On 2022-07-27 06:06 pm, Fei Wang wrote:
> > Previous default value 0.0 means 100% transparency for overlaid
> > video,
> > which make overlaid invisible. Change to 1.0 will let output video
> > much
> > more clear and
On Wed, 27 Jul 2022, Andreas Rheinhardt wrote:
Swinney, Jonathan:
This patch looks good to me. I would appreciate its merging.
} while (0)
#define PERF_STOP(t) do { \
+int ret;\
ioctl(sysfd, PERF_EVENT_IOC_
65 matches
Mail list logo
