Re: [FFmpeg-devel] [PATCH 3/9] x86: simple_idct_put: 10bits versions

2015-10-10 Thread Michael Niedermayer
On Fri, Oct 09, 2015 at 11:53:40PM +0200, Christophe Gisquet wrote: > Modeled from the prores version. Clips to [0;1023] and is bitexact. > Bitexactness requires to add an offset in a different place compared > to prores or C, and makes the function approximately 2% slower. > > For 16 frames of a

Re: [FFmpeg-devel] [PATCH 3/9] x86: simple_idct_put: 10bits versions

2015-10-09 Thread Michael Niedermayer
On Thu, Oct 08, 2015 at 08:22:50AM +0200, Christophe Gisquet wrote: > Modeled from the prores version. Clips to [0;1023] and is bitexact. > Bitexactness requires to add an offset in a different place compared > to prores or C, and makes the function approximately 2% slower. > > For 16 frames of a

[FFmpeg-devel] [PATCH 3/9] x86: simple_idct_put: 10bits versions

2015-10-08 Thread Christophe Gisquet
Modeled from the prores version. Clips to [0;1023] and is bitexact. Bitexactness requires to add an offset in a different place compared to prores or C, and makes the function approximately 2% slower. For 16 frames of a DNxHD 4:2:2 10bits test sequence: C:60861 decicycles in idct, 1048205