xenia-project/FFmpeg
1
0
Fork 0
mirror of https://github.com/xenia-project/FFmpeg.git synced 2024-11-25 04:30:02 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

inline -> always_inline (842 -> 690 cpu cycles for dct_quantize() difference for the dct itself should be even bigger)

Browse Source 
Originally committed as revision 2422 to svn://svn.ffmpeg.org/ffmpeg/trunk
This commit is contained in:
Michael Niedermayer 2003-10-23 12:59:21 +00:00
parent 8d00c562e8
commit 1745173bc5
1 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
libavcodec/i386/fdct_mmx.c
View File

@ -131,7 +131,7 @@ static const int16_t tab_frw_01234567[] ATTR_ALIGN(8) = { // forward_dct coeff
};
static inline void fdct_col(const int16_t *in, int16_t *out, int offset)
static always_inline void fdct_col(const int16_t *in, int16_t *out, int offset)
{
movq_m2r(*(in + offset + 1 * 8), mm0);
movq_m2r(*(in + offset + 6 * 8), mm1);
@ -210,15 +210,15 @@ static inline void fdct_col(const int16_t *in, int16_t *out, int offset)
movq_r2m(mm3, *(out + offset + 7 * 8));
}
static inline void fdct_row(const int16_t *in, int16_t *out, const int16_t *table)
{
static always_inline void fdct_row(const int16_t *in, int16_t *out, const int16_t *table)
{
movd_m2r(*(in + 6), mm5);
punpcklwd_m2r(*(in + 4), mm5);
movq_r2r(mm5, mm2);
psrlq_i2r(0x20, mm5);
movq_m2r(*(in + 0), mm0);
punpcklwd_r2r(mm2, mm5);
movq_r2r(mm0, mm1);
movq_r2r(mm0, mm1);
paddsw_r2r(mm5, mm0);
psubsw_r2r(mm5, mm1);
movq_r2r(mm0, mm2);