Re: [libav-devel] [PATCH] vaapi_encode: Move quality option in common code
On 12/05/17 16:59, Anton Khirnov wrote:
> Quoting Mark Thompson (2017-04-30 20:43:34)
>> Use AVCodecContext.compression_level rather than a private option,
>> replacing the H.264-specific quality option (which stays only for
>> compatibility).
>>
>> This now works with the H.265 encoder in the i965 driver, as well as
>> the existing cases with the H.264 encoder.
>> ---
>> doc/encoders.texi | 9 ++---
>> libavcodec/vaapi_encode.c | 33 +
>> libavcodec/vaapi_encode.h | 6 ++
>> libavcodec/vaapi_encode_h264.c | 25 ++---
>> 4 files changed, 47 insertions(+), 26 deletions(-)
>>
>> diff --git a/doc/encoders.texi b/doc/encoders.texi
>> index 7cebe39c1..c369e03bf 100644
>> --- a/doc/encoders.texi
>> +++ b/doc/encoders.texi
>> @@ -948,7 +948,13 @@ The following standard libavcodec options are used:
>> @item
>> @option{rc_init_occupancy} / @option{rc_initial_buffer_occupancy}
>> @item
>> +@option{compression_level}
>> +
>> +Speed / quality tradeoff: higher values are faster / worse quality.
>> +@item
>> @option{q} / @option{global_quality}
>> +
>> +Size / quality tradeoff: higher values are smaller / worse quality.
>> @item
>> @option{qmin}
>> (only: @option{qmax} is not supported)
>> @@ -969,9 +975,6 @@ The following standard libavcodec options are used:
>> @option{level} sets the value of @emph{level_idc}.
>>
>> @table @option
>> -@item quality
>> -Set the local encoding quality/speed tradeoff (range 1-8, higher values are
>> faster; not all
>> -systems implement all levels).
>> @item low_power
>> Use low-power encoding mode.
>> @end table
>> diff --git a/libavcodec/vaapi_encode.c b/libavcodec/vaapi_encode.c
>> index 620518419..5e5177a5d 100644
>> --- a/libavcodec/vaapi_encode.c
>> +++ b/libavcodec/vaapi_encode.c
>> @@ -1423,6 +1423,39 @@ av_cold int ff_vaapi_encode_init(AVCodecContext
>> *avctx)
>> goto fail;
>> }
>>
>> +if (avctx->compression_level >= 0) {
>> +#if VA_CHECK_VERSION(0, 36, 0)
>> +VAConfigAttrib attr = { VAConfigAttribEncQualityRange };
>> +
>> +vas = vaGetConfigAttributes(ctx->hwctx->display,
>> +ctx->va_profile,
>> +ctx->va_entrypoint,
>> +, 1);
>> +if (vas != VA_STATUS_SUCCESS) {
>> +av_log(avctx, AV_LOG_WARNING, "Failed to query quality "
>> + "attribute: will use default compression level.\n");
>> +} else if (avctx->compression_level > attr.value) {
>> +av_log(avctx, AV_LOG_ERROR, "Invalid compression level: "
>> + "valid range is 0-%d.\n", attr.value);
>> +err = AVERROR(EINVAL);
>
> It looks a bit strange to me that this branch fails, while the case
> where vaapi doesn't support this option continues.
Yeah; changed to warn and bound above in this case.
> Otherwise looks fine.
And pushed (somewhat belatedly).
Thanks,
- Mark
___
libav-devel mailing list
libav-devel@libav.org
https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH] aarch64: Fix negative movrel offsets for windows
On 06/08/2017 12:38, Martin Storsjö wrote: > On windows, the offset for the relocation doesn't get stored in > the relocation itself, but as an unsigned immediate in the opcode. > Therefore, negative offsets has to be handled via a separate sub > instruction, just as on MachO. > --- > libavutil/aarch64/asm.S | 9 + > 1 file changed, 9 insertions(+) > > diff --git a/libavutil/aarch64/asm.S b/libavutil/aarch64/asm.S > index 86d36cd..7e2af4d 100644 > --- a/libavutil/aarch64/asm.S > +++ b/libavutil/aarch64/asm.S > @@ -82,6 +82,15 @@ ELF .size \name, . - \name > adrp\rd, \val+(\offset)@PAGE > add \rd, \rd, \val+(\offset)@PAGEOFF > .endif > +#elif CONFIG_PIC && defined(_WIN32) > +.if \offset < 0 > +adrp\rd, \val > +add \rd, \rd, :lo12:\val > +sub \rd, \rd, -(\offset) > +.else > +adrp\rd, \val+(\offset) > +add \rd, \rd, :lo12:\val+(\offset) > +.endif > #elif CONFIG_PIC > adrp\rd, \val+(\offset) > add \rd, \rd, :lo12:\val+(\offset) > Seems fine ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
Re: [libav-devel] [PATCH 10/10] RFC: generic hwaccel everywhere
On 06/08/2017 00:53, Mark Thompson wrote: > --- > This is several patches squashed together to invite thoughts. > > The idea here is that we add a new array of possible hardware configurations > to codec definitions, expressed as a hardware device type and the matching > pixel format. We can then use that to set up hwaccels without any additional > knowledge at all, creating the appropriate device at startup and then > matching the format in get_format in order to use it. > > This is implemented here in lavc with the H.264 decoder and then used in > avconv. There is quite a bit of churn in avconv hardware setup as the choice > of device all moves around, but it is hopefully a bit clearer now (and less > likely to try to give devices to things not needing them). "-hwaccel auto" > is reimplemented completely as using the first device from the codec config > list which can be instantiated. > I like the idea. ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
[libav-devel] [PATCH] aarch64: Fix negative movrel offsets for windows
On windows, the offset for the relocation doesn't get stored in the relocation itself, but as an unsigned immediate in the opcode. Therefore, negative offsets has to be handled via a separate sub instruction, just as on MachO. --- libavutil/aarch64/asm.S | 9 + 1 file changed, 9 insertions(+) diff --git a/libavutil/aarch64/asm.S b/libavutil/aarch64/asm.S index 86d36cd..7e2af4d 100644 --- a/libavutil/aarch64/asm.S +++ b/libavutil/aarch64/asm.S @@ -82,6 +82,15 @@ ELF .size \name, . - \name adrp\rd, \val+(\offset)@PAGE add \rd, \rd, \val+(\offset)@PAGEOFF .endif +#elif CONFIG_PIC && defined(_WIN32) +.if \offset < 0 +adrp\rd, \val +add \rd, \rd, :lo12:\val +sub \rd, \rd, -(\offset) +.else +adrp\rd, \val+(\offset) +add \rd, \rd, :lo12:\val+(\offset) +.endif #elif CONFIG_PIC adrp\rd, \val+(\offset) add \rd, \rd, :lo12:\val+(\offset) -- 2.7.4 ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel
