From 3611e7a3090f06ef3b9f5060334070e122e1b406 Mon Sep 17 00:00:00 2001 From: Eli Friedman Date: Fri, 23 Jul 2010 21:46:30 +0000 Subject: Inline asm for VP56 arith coder This is a lot more reliable to get cmov rather than trying to trick gcc into generating it, useful since it's 2% faster overall. Patch by Eli Friedman Originally committed as revision 24471 to svn://svn.ffmpeg.org/ffmpeg/trunk --- libavcodec/vp56.h | 12 +++++++----- 1 file changed, 7 insertions(+), 5 deletions(-) (limited to 'libavcodec/vp56.h') diff --git a/libavcodec/vp56.h b/libavcodec/vp56.h index ad07a49e9b..1eacdc2158 100644 --- a/libavcodec/vp56.h +++ b/libavcodec/vp56.h @@ -208,23 +208,25 @@ static av_always_inline unsigned int vp56_rac_renorm(VP56RangeCoder *c) return code_word; } +#if ARCH_X86 +#include "x86/vp56_arith.h" +#endif + +#ifndef vp56_rac_get_prob +#define vp56_rac_get_prob vp56_rac_get_prob static inline int vp56_rac_get_prob(VP56RangeCoder *c, uint8_t prob) { - /* Don't put c->high in a local variable; if we do that, gcc gets - * the stupids and turns the code below into a branch again. */ unsigned int code_word = vp56_rac_renorm(c); unsigned int low = 1 + (((c->high - 1) * prob) >> 8); unsigned int low_shift = low << 8; int bit = code_word >= low_shift; - /* Incantation to convince GCC to turn these into conditional moves - * instead of branches -- faster, as this branch is basically - * unpredictable. */ c->high = bit ? c->high - low : low; c->code_word = bit ? code_word - low_shift : code_word; return bit; } +#endif // branchy variant, to be used where there's a branch based on the bit decoded static av_always_inline int vp56_rac_get_prob_branchy(VP56RangeCoder *c, int prob) -- cgit v1.2.3