generic-library/ffmpeg
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

offset and weights are signed, fixes some non-bitexact issues.

Browse Source 
Patch by David Conrad %lessen42 A gmail P com%

Originally committed as revision 16449 to svn://svn.ffmpeg.org/ffmpeg/trunk
This commit is contained in:
David Conrad 2009-01-06 10:35:06 +00:00 committed by Guillaume Poirier
parent 373745ae62
commit 54b9095833
1 changed files with 6 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
libavcodec/ppc/h264_altivec.c
View File

@ -942,7 +942,8 @@ void biweight_h264_WxH_altivec(uint8_t *dst, uint8_t *src, int stride, int log2_
{ {
int y, dst_aligned, src_aligned; int y, dst_aligned, src_aligned;
vec_u8 vsrc, vdst; vec_u8 vsrc, vdst;
vec_u16 vtemp, vlog2_denom, vweights, vweightd, voffset, v0, v1, v2, v3; vec_s16 vtemp, vweights, vweightd, voffset, v0, v1, v2, v3;
vec_u16 vlog2_denom;
DECLARE_ALIGNED_16(int32_t, temp[4]); DECLARE_ALIGNED_16(int32_t, temp[4]);
LOAD_ZERO; LOAD_ZERO;
@ -977,16 +978,16 @@ void biweight_h264_WxH_altivec(uint8_t *dst, uint8_t *src, int stride, int log2_
} }
if (w == 16 || dst_aligned) { if (w == 16 || dst_aligned) {
v0 = vec_mladd(v0, vweightd, zero_u16v); v0 = vec_mladd(v0, vweightd, zero_s16v);
v2 = vec_mladd(v2, vweights, zero_u16v); v2 = vec_mladd(v2, vweights, zero_s16v);
v0 = vec_adds(v0, voffset); v0 = vec_adds(v0, voffset);
v0 = vec_adds(v0, v2); v0 = vec_adds(v0, v2);
v0 = vec_sra(v0, vlog2_denom); v0 = vec_sra(v0, vlog2_denom);
} }
if (w == 16 || !dst_aligned) { if (w == 16 || !dst_aligned) {
v1 = vec_mladd(v1, vweightd, zero_u16v); v1 = vec_mladd(v1, vweightd, zero_s16v);
v3 = vec_mladd(v3, vweights, zero_u16v); v3 = vec_mladd(v3, vweights, zero_s16v);
v1 = vec_adds(v1, voffset); v1 = vec_adds(v1, voffset);
v1 = vec_adds(v1, v3); v1 = vec_adds(v1, v3);