diff amr.c @ 5000:743a8b12b7de libavcodec

Faster SSE FFT/MDCT, patch by Zuxy Meng %zuxy P meng A gmail P com% unrolls some loops, utilizing all 8 xmm registers. fft-test shows ~10% speed up in (I)FFT and ~8% speed up in (I)MDCT on Dothan
author gpoirier
date Sun, 13 May 2007 16:32:32 +0000
parents 3df69e140c33
children
line wrap: on
line diff