SBR DSP x86: implement SSE sbr_hf_gen
authorChristophe Gisquet <christophe.gisquet@gmail.com>
Fri, 24 Feb 2012 21:11:19 +0000 (22:11 +0100)
committerDiego Biurrun <diego@biurrun.de>
Fri, 7 Dec 2012 10:04:26 +0000 (11:04 +0100)
commit2aef3d66c9cdf9d9ad3ecec2fb0c6b3020e9d3b0
tree4e40bc1d24002be20fa1b0592a5b66e4cb9657d2
parent9a16359c38888af6beab7e688fc55eaa60c835ec
SBR DSP x86: implement SSE sbr_hf_gen

Start and end index are multiple of 2, therefore guaranteeing aligned access.
Also, this allows to generate 4 floats per loop, keeping the alignment all
along.

Timing:
- 32 bits: 326c -> 172c
- 64 bits: 323c -> 156c

Signed-off-by: Diego Biurrun <diego@biurrun.de>
libavcodec/x86/sbrdsp.asm
libavcodec/x86/sbrdsp_init.c