[FFmpeg-devel] [PATCH] faster SSE FFT

[Måns Rullgård]

Loren Merritt <lorenm at u.washington.edu> writes:

> Factor a shuffle out of the inner loop and merge it into fft_permute.
> 6% faster SSE FFT on Conroe, 2.5% on Penryn.
>
> --Loren Merritt
>
> From 64d5047455d7bfc785628efdf08883983396446b Mon Sep 17 00:00:00 2001
> From: Loren Merritt <lorenm at u.washington.edu>
> Date: Sat, 12 Feb 2011 11:48:16 +0000
> Subject: [PATCH] FFT: factor a shuffle out of the inner loop and merge it into fft_permute.
>
> 6% faster SSE FFT on Conroe, 2.5% on Penryn.
> ---
>  libavcodec/arm/fft_init_arm.c |    2 +-
>  libavcodec/fft.c              |    9 ++++-
>  libavcodec/fft.h              |    5 ++-
>  libavcodec/mdct.c             |    4 +-
>  libavcodec/x86/fft.c          |    1 +
>  libavcodec/x86/fft_mmx.asm    |   62 ++++++++++++++++++++---------------------
>  6 files changed, 45 insertions(+), 38 deletions(-)

Nice.  Non-x86 parts look good.

-- 
M?ns Rullg?rd
mans at mansr.com