Le sunnuntaina 9. kesäkuuta 2024, 12.27.06 EEST Rémi Denis-Courmont a écrit :
> To preserve the alignment promise of the block base address, we pass
> the start offset separately (it is either 0 or 1).
>
> Note that optimised implementations of this function will be taken into
> actual use if MpegEncContext.dct_unquantize_h263_{inter,intra} are *not*
> overloaded by existing optimisations.
> ---
> libavcodec/h263dsp.c | 17 +++++++++++++++++
> libavcodec/h263dsp.h | 2 ++
> 2 files changed, 19 insertions(+)
>
> diff --git a/libavcodec/h263dsp.c b/libavcodec/h263dsp.c
> index 6a13353499..eb990f27bd 100644
> --- a/libavcodec/h263dsp.c
> +++ b/libavcodec/h263dsp.c
> @@ -23,6 +23,22 @@
> #include "config.h"
> #include "h263dsp.h"
>
> +static void h263_dct_unquantize_c(int16_t *block, size_t start, size_t end,
> + int qmul, int qadd)
> +{
> + for (size_t i = start; i <= end; i++) {
C and R-V V don't care, but for the sake of more constrained SIMD extensions,
it should be easier to keep intra and inter DCT functions separate.
--
Rémi Denis-Courmont
http://www.remlab.net/
_______________________________________________
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 with subject "unsubscribe".
