Re: [FFmpeg-devel] [PATCH v3] aacenc: add SIMD optimizations for abs_pow34 and quantization

2016-10-18 Thread Rostislav Pehlivanov
On 18 October 2016 at 21:04, Michael Niedermayer wrote: > On Tue, Oct 18, 2016 at 05:33:13PM +0100, Rostislav Pehlivanov wrote: > > On 18 October 2016 at 16:32, James Almer wrote: > > > > > On 10/18/2016 12:07 PM, Rostislav Pehlivanov wrote: > > > >

Re: [FFmpeg-devel] [PATCH v3] aacenc: add SIMD optimizations for abs_pow34 and quantization

2016-10-18 Thread Michael Niedermayer
On Tue, Oct 18, 2016 at 05:33:13PM +0100, Rostislav Pehlivanov wrote: > On 18 October 2016 at 16:32, James Almer wrote: > > > On 10/18/2016 12:07 PM, Rostislav Pehlivanov wrote: > > > diff --git a/libavcodec/aacenc.c b/libavcodec/aacenc.c > > > index ee3cbf8..622f0ba 100644 >

Re: [FFmpeg-devel] [PATCH v3] aacenc: add SIMD optimizations for abs_pow34 and quantization

2016-10-18 Thread Rostislav Pehlivanov
On 18 October 2016 at 16:32, James Almer wrote: > On 10/18/2016 12:07 PM, Rostislav Pehlivanov wrote: > > diff --git a/libavcodec/aacenc.c b/libavcodec/aacenc.c > > index ee3cbf8..622f0ba 100644 > > --- a/libavcodec/aacenc.c > > +++ b/libavcodec/aacenc.c > > @@ -1033,6

Re: [FFmpeg-devel] [PATCH v3] aacenc: add SIMD optimizations for abs_pow34 and quantization

2016-10-18 Thread Rostislav Pehlivanov
On 18 October 2016 at 14:51, Michael Niedermayer wrote: > On Tue, Oct 18, 2016 at 09:02:19AM +0100, Rostislav Pehlivanov wrote: > > On 17 October 2016 at 23:43, Michael Niedermayer > > > wrote: > > > > > On Mon, Oct 17, 2016 at 10:24:48PM +0100,

Re: [FFmpeg-devel] [PATCH v3] aacenc: add SIMD optimizations for abs_pow34 and quantization

2016-10-18 Thread Michael Niedermayer
On Tue, Oct 18, 2016 at 09:02:19AM +0100, Rostislav Pehlivanov wrote: > On 17 October 2016 at 23:43, Michael Niedermayer > wrote: > > > On Mon, Oct 17, 2016 at 10:24:48PM +0100, Rostislav Pehlivanov wrote: > > > Should fix segfaults on x86-32 > > > > > > Performance

Re: [FFmpeg-devel] [PATCH v3] aacenc: add SIMD optimizations for abs_pow34 and quantization

2016-10-18 Thread Rostislav Pehlivanov
On 17 October 2016 at 23:43, Michael Niedermayer wrote: > On Mon, Oct 17, 2016 at 10:24:48PM +0100, Rostislav Pehlivanov wrote: > > Should fix segfaults on x86-32 > > > > Performance improvements: > > > > quant_bands: > > with: 681 decicycles in quant_bands, 8388453

Re: [FFmpeg-devel] [PATCH v3] aacenc: add SIMD optimizations for abs_pow34 and quantization

2016-10-17 Thread James Almer
On 10/17/2016 6:24 PM, Rostislav Pehlivanov wrote: > diff --git a/libavcodec/aacenc_utils.h b/libavcodec/aacenc_utils.h > index ff9188a..f5cf77d 100644 > --- a/libavcodec/aacenc_utils.h > +++ b/libavcodec/aacenc_utils.h > @@ -37,7 +37,7 @@ > #define ROUND_TO_ZERO 0.1054f > #define C_QUANT

Re: [FFmpeg-devel] [PATCH v3] aacenc: add SIMD optimizations for abs_pow34 and quantization

2016-10-17 Thread Michael Niedermayer
On Mon, Oct 17, 2016 at 10:24:48PM +0100, Rostislav Pehlivanov wrote: > Should fix segfaults on x86-32 > > Performance improvements: > > quant_bands: > with: 681 decicycles in quant_bands, 8388453 runs,155 skips > without: 1190 decicycles in quant_bands, 8388386 runs,222 skips >

[FFmpeg-devel] [PATCH v3] aacenc: add SIMD optimizations for abs_pow34 and quantization

2016-10-17 Thread Rostislav Pehlivanov
Should fix segfaults on x86-32 Performance improvements: quant_bands: with: 681 decicycles in quant_bands, 8388453 runs,155 skips without: 1190 decicycles in quant_bands, 8388386 runs,222 skips Around 42% for the function Twoloop coder: abs_pow34: with/without: 7.82s/8.17s Around