2015-10-26 12:44 GMT+01:00 wm4 :
> On Mon, 26 Oct 2015 11:22:38 +0100
> Gwenole Beauchesne wrote:
>
>
>> >> /**
>> >> * Hardware Accelerator identifier.
>> >> *
>> >> * @note
>> >> * A hardware accelerator can be device-l
Hi,
2015-10-23 16:52 GMT+02:00 wm4 :
> On Fri, 23 Oct 2015 16:12:07 +0200
> Gwenole Beauchesne wrote:
>
>> Hi,
>>
>> 2015-10-23 15:41 GMT+02:00 wm4 :
>> > On Fri, 23 Oct 2015 14:54:56 +0200
>> > Gwenole Beauchesne wrote:
>> >
>> >&g
Hi,
2015-10-25 14:47 GMT+01:00 wm4 :
> On Sun, 25 Oct 2015 14:28:06 +0100
> Nicolas George wrote:
>
>> Le quartidi 4 brumaire, an CCXXIV, Gwenole Beauchesne a écrit :
>> > Allow for av_malloc() to allocate memory that could be mapped into
>> > the GPU address sp
, and avcodec changes.
Option 2: use a finer decision model
- mem: add av_malloc_aligned()
- buffer: add av_buffer_alloc2() with align and/or flags
- frame/avcodec: use new APIs
Signed-off-by: Gwenole Beauchesne
---
configure | 1 +
libavcodec/internal.h | 4 +++-
libavutil/mem.c
Hi,
2015-10-23 15:41 GMT+02:00 wm4 :
> On Fri, 23 Oct 2015 14:54:56 +0200
> Gwenole Beauchesne wrote:
>
>> Hi,
>>
>> 2015-10-07 18:40 GMT+02:00 wm4 :
>> > On Wed, 7 Oct 2015 19:20:56 +0300
>> > Ivan Uskov wrote:
>> >
>> >> Hell
cept your current patch,
> though. Global mutable variables are a no in new code.
> ___
> ffmpeg-devel mailing list
> ffmpeg-devel@ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
Regards,
--
Gwenole Beauchesne
Intel Corporation SAS / 2 rue de Paris, 92196 Meudon Cedex, France
Registration Number (RCS): Nanterre B 302 456 199
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
t;> Its really not a lot of work, and keeps the code sane and the decoding
>> reliable.
>>
>
> No further comments? No thoughts? No-one that would object if I were
> to push the proposed patch?
Fine with me.
Probably add a link to your ffmpeg-devel message since it provides all
the necessary rationale behind this?
Thanks,
--
Gwenole Beauchesne
Intel Corporation SAS / 2 rue de Paris, 92196 Meudon Cedex, France
Registration Number (RCS): Nanterre B 302 456 199
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
you can help to clarify these points.
>
> I'll happily provide a few benchmarks and update documentation
> accordingly if I manage to make it work.
> --
> FFmpeg = Fancy and Forgiving Mean Problematic Excellent Gymnast
> ___
>
Hi,
2015-09-22 14:21 GMT+02:00 Clément Bœsch :
> On Mon, Sep 21, 2015 at 08:27:47PM +0100, Kieran Kunhya wrote:
>> > This "decision" thingy, is intended to confirm that
>> > * Stuff should be moved to the telepoint server, possibly duplicates
>> > should be left at/moved to the hetzner server fo
YY" is just enough
information, IMHO.
I'd say, it depends on the amount of tickets fixed per release. I
don't think users are much interested in reading through 50+ ticket
numbers.
Regards,
--
Gwenole Beauchesne
Intel Corporation SAS / 2 rue de Paris, 92196 Meudon Cedex, France
Regist
slice_segment_address = sh->slice_segment_addr;
> +
> +slice_param->LongSliceFlags.value = 0;
> +slice_param->LongSliceFlags.fields.dependent_slice_segment_flag =
> sh->dependent_slice_segment_flag;
> +slice_param->LongSliceFlags.fields.slice_type = sh->slice_type;
Hi,
2015-08-19 21:57 GMT+02:00 wm4 :
> On Wed, 19 Aug 2015 19:32:27 +0200
> Gwenole Beauchesne wrote:
>
>> 2015-08-19 19:19 GMT+02:00 Gwenole Beauchesne :
>> > Hi,
>> >
>> > 2015-08-19 18:50 GMT+02:00 wm4 :
>> >> On Wed, 19 A
2015-08-19 19:19 GMT+02:00 Gwenole Beauchesne :
> Hi,
>
> 2015-08-19 18:50 GMT+02:00 wm4 :
>> On Wed, 19 Aug 2015 18:01:36 +0200
>> Gwenole Beauchesne wrote:
>>
>>> Add av_vaapi_set_pipeline_params() interface to initialize the internal
>>> FFVA
Hi,
2015-08-19 18:50 GMT+02:00 wm4 :
> On Wed, 19 Aug 2015 18:01:36 +0200
> Gwenole Beauchesne wrote:
>
>> Add av_vaapi_set_pipeline_params() interface to initialize the internal
>> FFVAContext with a valid VA display, and optional parameters including
>> a
.
This is an internal infrastructure change/addition only.
Signed-off-by: Gwenole Beauchesne
---
libavcodec/internal.h | 14 ++
libavcodec/utils.c| 2 ++
2 files changed, 16 insertions(+)
diff --git a/libavcodec/internal.h b/libavcodec/internal.h
index f93a196..413e49b 100644
used, the user shall
provide the VA context id and a valid AVCodecContext.get_buffer2() hook.
Otherwise, the older vaapi_context approach is still supported.
This is an API change. The whole vaapi_context structure is no longer
needed, and thus deprecated.
Signed-off-by: Gwenole Beauchesne
- ff_vaapi_get_chroma_format():
Converts FFmpeg pixel format to VA chroma format
- ff_vaapi_get_pixel_format():
Converts FFmpeg pixel format to VA fourcc
At some point, they could be migrated to some libavutil_vaapi and
exposed from there.
Signed-off-by: Gwenole Beauchesne
---
libavcodec/Makefile | 4
implementation and return AVERROR(ENOSYS) in the
.alloc_frame() hook so that to notify get_buffer_internal() to try
to call into .get_buffer*() hooks.
This is an internal semantical change only, and does not affect any
existing hwaccel implementation.
Signed-off-by: Gwenole Beauchesne
restting id to VA_INVALID_ID after
release for instance), converting a VAStatus to some FFmpeg error,
etc.
I have attached two additional patches so that to give a glimpse at
how it could look like.
Regards,
Gwenole Beauchesne (4):
hwaccel: add infrastructure to hold accelerator config options.
Hi,
2015-08-19 16:12 GMT+02:00 Michael Niedermayer :
> On Wed, Aug 19, 2015 at 11:03:05AM +0200, Gwenole Beauchesne wrote:
>> Move libavcodec managed objects from the public struct vaapi_context
>> to a new privately owned FFVAContext. This is done so that to clean up
>> and
: Gwenole Beauchesne
---
doc/APIchanges | 4
libavcodec/vaapi.c | 40 +++-
libavcodec/vaapi.h | 16
libavcodec/vaapi_h264.c | 10 +++---
libavcodec/vaapi_internal.h | 42
drivers that strictly conform to
the API whereby an invalid buffer id is VA_INVALID_ID and the first
valid buffer id can actually be zero.
Signed-off-by: Gwenole Beauchesne
---
libavcodec/vaapi.c | 20
1 file changed, 12 insertions(+), 8 deletions(-)
diff --git a/libavcodec
reach lavu major+2, amend doc/APIchanges.
Signed-off-by: Gwenole Beauchesne
---
doc/APIchanges | 6 ++
libavcodec/h263dec.c | 2 +-
libavcodec/h264_slice.c | 2 +-
libavcodec/mpeg12dec.c | 2 +-
libavcodec/vaapi_h264.c | 2 +-
libavcodec/vaapi_mpeg2.c | 2
Hi,
This is preparatory and minimal work in view to supporting other (WIP)
changes. I will push those by tomorrow if there is no other strong
objections.
For convenience, I have placed the code here:
<https://github.com/gbeauchesne/FFmpeg/tree/10.vaapi.lavc.fixes>
Regards,
Gwenole Beauche
Hi,
2015-08-19 0:02 GMT+02:00 Andreas Cadhalpun :
> On 18.08.2015 17:26, Gwenole Beauchesne wrote:
>> Deprecate older VA pixel formats (MOCO, IDCT) as it is now very unlikely
>> to ever be useful in the future. Only keep plain AV_PIX_FMT_VAAPI format
>> that is aliased to
.
This is an internal infrastructure change/addition only.
Signed-off-by: Gwenole Beauchesne
---
libavcodec/internal.h | 14 ++
libavcodec/utils.c| 2 ++
2 files changed, 16 insertions(+)
diff --git a/libavcodec/internal.h b/libavcodec/internal.h
index 0daf669..866ed12 100644
ructure is now not needed, and thus deprecated.
Signed-off-by: Gwenole Beauchesne
---
doc/APIchanges | 6
libavcodec/vaapi.c | 66
libavcodec/vaapi.h | 73 +++--
libavcodec/vaapi_inte
AVCodecContext.hwaccel_context, that holds the public vaapi_context,
shall no longer be accessed from within vaapi_*.c codec support files.
Signed-off-by: Gwenole Beauchesne
---
libavcodec/vaapi.c | 34 +-
libavcodec/vaapi.h | 16
libavcodec
drivers that strictly conform to
the API whereby an invalid buffer id is VA_INVALID_ID and the first
valid buffer id can actually be zero.
Signed-off-by: Gwenole Beauchesne
---
libavcodec/vaapi.c | 20
1 file changed, 12 insertions(+), 8 deletions(-)
diff --git a/libavcodec
Deprecate older VA pixel formats (MOCO, IDCT) as it is now very unlikely
to ever be useful in the future. Only keep plain AV_PIX_FMT_VAAPI format
that is aliased to the older VLD variant.
Signed-off-by: Gwenole Beauchesne
---
libavcodec/h263dec.c | 2 +-
libavcodec/h264_slice.c | 2
Hi,
2015-08-18 10:43 GMT+02:00 Hendrik Leppkes :
> On Tue, Aug 18, 2015 at 10:35 AM, Gwenole Beauchesne
> wrote:
>> 2015-08-17 21:53 GMT+02:00 wm4 :
>>> On Mon, 17 Aug 2015 19:17:50 +0200
>>> Gwenole Beauchesne wrote:
>>>
>>>> Deprecate ol
2015-08-17 21:53 GMT+02:00 wm4 :
> On Mon, 17 Aug 2015 19:17:50 +0200
> Gwenole Beauchesne wrote:
>
>> Deprecate older VA pixel formats (MOCO, IDCT) as it is now very unlikely
>> to ever be useful in the future. Only keep plain AV_PIX_FMT_VAAPI format
>> that is alias
Hi,
2015-08-17 22:26 GMT+02:00 wm4 :
> On Mon, 17 Aug 2015 19:17:53 +0200
> Gwenole Beauchesne wrote:
>
>> Add av_vaapi_context_init() and av_vaapi_context_alloc() helper functions
>> so that to properly initialize the vaapi_context structure, and allow for
>> future
Hi,
2015-08-17 19:17 GMT+02:00 Gwenole Beauchesne :
> Add av_vaapi_context_init() and av_vaapi_context_alloc() helper functions
> so that to properly initialize the vaapi_context structure, and allow for
> future extensions without breaking the API/ABI.
>
> The new version and f
libavcodec.
Besides, it is now required by design, and actual usage, that the vaapi
context structure be completely initialized once and for all before the
AVCodecContext.get_format() function ever completes.
Signed-off-by: Gwenole Beauchesne
---
libavcodec/vaapi.c | 30
drivers that strictly conform to
the API whereby an invalid buffer id is VA_INVALID_ID and the first
valid buffer id can actually be zero.
Signed-off-by: Gwenole Beauchesne
---
libavcodec/vaapi.c | 20
1 file changed, 12 insertions(+), 8 deletions(-)
diff --git a/libavcodec
Deprecate older VA pixel formats (MOCO, IDCT) as it is now very unlikely
to ever be useful in the future. Only keep plain AV_PIX_FMT_VAAPI format
that is aliased to the older VLD variant.
Signed-off-by: Gwenole Beauchesne
---
libavcodec/h263dec.c | 2 +-
libavcodec/h264_slice.c | 2
AVCodecContext.hwaccel_context, that holds the public vaapi_context,
shall no longer be accessed from within vaapi_*.c codec support files.
Signed-off-by: Gwenole Beauchesne
---
libavcodec/vaapi.c | 34 +-
libavcodec/vaapi.h | 16
libavcodec
ree/10.vaapi.lavc.fixes>
Regards,
Gwenole Beauchesne (4):
vaapi: define a single pixel format for VA-API (AV_PIX_FMT_VAAPI).
vaapi: streamline public context structure.
vaapi: fix usage of invalid buffer ids.
vaapi: add interfaces to properly initialize context.
libavcodec/h26
ine IMHO before
ff_qsv_close_internal_session() and after QSVSession definition.
Note it would also be interesting to provide a means to import a
VADisplay from the user through an additional function. e.g. in view
to interop VA decode to MSDK encode and/or video processing.
Regards,
--
Gwenole Beauchesne
Intel Corporat
2015-06-23 13:10 GMT+02:00 Michael Niedermayer :
> On Tue, Jun 23, 2015 at 11:50:52AM +0200, Gwenole Beauchesne wrote:
>> Use new H264Ref.reference field to track field picture flags. The
>> H264Picture.reference flag in DPB is now irrelevant here.
>>
>> This is a
Use new H264Ref.reference field to track field picture flags. The
H264Picture.reference flag in DPB is now irrelevant here.
This is a regression from git commit d8151a7, and that affected
multiple interlaced video streams.
Signed-off-by: Gwenole Beauchesne
---
libavcodec/vaapi_h264.c | 3
store " %%xmm3, 32(%[dst])\n" \
>> +store " %%xmm4, 48(%[dst])\n" \
>> +: : [dst]"r"(dstp), [src]"r"(srcp) : "memory", "xmm1", "xmm2",
>> "xmm3", "xmm4")
>> +#endif
> > +memcpy(dst, src, size, cpu_flags);
>> > > +#endif
>> > > +}
>> >
>> > Alternatively, what about something like:
>> >
>> > av_memcpynt_fn av_memcpynt_get_fn(void);
>> >
>> &g
* symbols into a particular namespace and make
all other symbols private, internal. e.g. SOMETHING_FFMPEG { global:
av*; local: *; };
4. Link the rest of your project with those built-in FFmpeg headers
and libraries, ship with the FFmpeg DSO.
Of course, you still have the option to link against syst
45 matches
Mail list logo
