[FFmpeg-devel] [PATCH] avcodec/vp9/x86: add ff_vp9_loop_filter_v_16_16_ssse3().

[Clément Bœsch]

On Sun, Jan 12, 2014 at 08:04:40PM +0100, Clément Bœsch wrote:
> On Sun, Jan 12, 2014 at 11:55:24AM +0100, Clément Bœsch wrote:
> > On Mon, Jan 06, 2014 at 12:12:04AM +0100, Clément Bœsch wrote:
> > [...]
> > > This might not be considered a final commit since I'll probably push along with
> > > the horizontal one. Though, of course, comments welcome.
> > 
> > New patch with horizontal, and misc improvements.
> > 
> > Note: I made vertical code assuming always aligned, but couldn't with
> > horizontal. Is this correct? (assuming aligned for horizontal doesn't work
> > at all).
> > 
> 
> And new iteration using an (aligned) stack for the horizontal (-300 cycles).
> 
> [...]
> 
> -- 
> Clément B.

> From 03b6166e7ad0413e44e4aaf246f10eab199743e7 Mon Sep 17 00:00:00 2001
> From: =?UTF-8?q?Cl=C3=A9ment=20B=C5=93sch?= <u at pkh.me>
> Date: Sun, 5 Jan 2014 21:00:40 +0100
> Subject: [PATCH] avcodec/vp9/x86: add ff_vp9_loop_filter_[vh]_16_16_ssse3().
> 
> 16662 decicycles in loop_filter_h_16_16_c, 8387355 runs, 1253 skips
> 17510 decicycles in loop_filter_v_16_16_c, 8387516 runs, 1092 skips
> 
> 4941 decicycles in ff_vp9_loop_filter_h_16_16_ssse3, 8387887 runs, 721 skips
> 3899 decicycles in ff_vp9_loop_filter_v_16_16_ssse3, 8387980 runs, 628 skips
> 
> Overall decode time goes from:
>   ./ffmpeg -v 0 -nostats -threads 1 -i ~/samples/vp9/ped1080p.webm -f null -  8.10s user 0.02s system 99% cpu 8.126 total
> to:
>   ./ffmpeg -v 0 -nostats -threads 1 -i ~/samples/vp9/ped1080p.webm -f null -  6.15s user 0.04s system 99% cpu 6.199 total
> 
> (46 to 61 fps)
> ---
>  libavcodec/x86/Makefile      |   1 +
>  libavcodec/x86/vp9dsp_init.c |   5 +
>  libavcodec/x86/vp9lpf.asm    | 666 +++++++++++++++++++++++++++++++++++++++++++
>  3 files changed, 672 insertions(+)
>  create mode 100644 libavcodec/x86/vp9lpf.asm
> 

Approved by Ronald on IRC and pushed.

[...]

-- 
Clément B.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 490 bytes
Desc: not available
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20140112/8b0f7dbd/attachment.asc>