[FFmpeg-devel] [Patch]Hevc idct avx2

[Pierre Edouard Lepere]

I fixed it because the function uses 6 mmx registers, not 0 as described previously.

----- Mail original -----
De: "Michael Niedermayer" <michaelni at gmx.at>
À: "FFmpeg development discussions and patches" <ffmpeg-devel at ffmpeg.org>
Envoyé: Mardi 24 Juin 2014 23:23:56
Objet: Re: [FFmpeg-devel] [Patch]Hevc idct avx2

On Fri, Jun 20, 2014 at 10:19:12AM +0200, Pierre Edouard Lepere wrote:
[...]

> @@ -60,17 +72,18 @@ SECTION .text
>      %1         [%2+%4  ], m5
>  %endmacro
>  
> +
>  INIT_MMX mmxext
>  ; void ff_hevc_idct_dc_add_8_mmxext(uint8_t *dst, int16_t *coeffs, ptrdiff_t stride)
>  %if ARCH_X86_64
> -cglobal hevc_idct4_dc_add_8, 3, 4, 0
> +cglobal hevc_idct4_dc_add_8, 3, 4, 6

why are these mmxext functions changed ?


[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

The educated differ from the uneducated as much as the living from the
dead. -- Aristotle 

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel at ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel