har0ke/FFmpeg
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Port pred8x8l_horizontal_mmxext/ssse3 (H.264 intra prediction) from x264 to

Browse Source 
FFmpeg. Original authors: Holger Lubitz <holger lubitz org>, Jason Garrett-
Glaser <darkshikari gmail com> (approves LGPL relicensing for this code) and
Loren Merritt <lorenm at u dot washington dot edu> (approves LGPL relicensing
for this code). Patch by Daniel Kang <daniel dot d dot kang at gmail com>, as
part of Google's GCI 2010.

Originally committed as revision 26139 to svn://svn.ffmpeg.org/ffmpeg/trunk
This commit is contained in:
Daniel Kang 2010-12-29 18:40:53 +00:00 committed by Ronald S. Bultje
parent abab14eac0
commit 04207ef353
2 changed files with 83 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

79
libavcodec/x86/h264_intrapred.asm
View File

@ -1243,6 +1243,85 @@ PRED8x8L_DC mmxext
PRED8x8L_DC ssse3 PRED8x8L_DC ssse3
%endif %endif
;-----------------------------------------------------------------------------
; void pred8x8l_horizontal(uint8_t *src, int has_topleft, int has_topright, int stride)
;-----------------------------------------------------------------------------
%ifdef CONFIG_GPL
%macro PRED8x8L_HORIZONTAL 1
cglobal pred8x8l_horizontal_%1, 4,4
sub r0, r3
lea r2, [r0+r3*2]
movq mm0, [r0+r3*1-8]
punpckhbw mm0, [r0+r3*0-8]
movq mm1, [r2+r3*1-8]
punpckhbw mm1, [r0+r3*2-8]
mov r2, r0
punpckhwd mm1, mm0
lea r0, [r0+r3*4]
movq mm2, [r0+r3*1-8]
punpckhbw mm2, [r0+r3*0-8]
lea r0, [r0+r3*2]
movq mm3, [r0+r3*1-8]
punpckhbw mm3, [r0+r3*0-8]
punpckhwd mm3, mm2
punpckhdq mm3, mm1
lea r0, [r0+r3*2]
movq mm0, [r0+r3*0-8]
movq mm1, [r2]
mov r0, r2
movq mm4, mm3
movq mm2, mm3
PALIGNR mm4, mm0, 7, mm0
PALIGNR mm1, mm2, 1, mm2
test r1, r1 ; top_left
jnz .do_left
.fix_lt_1:
movq mm5, mm3
pxor mm5, mm4
psrlq mm5, 56
psllq mm5, 48
pxor mm1, mm5
.do_left:
movq mm0, mm4
PRED4x4_LOWPASS mm2, mm1, mm4, mm3, mm5
movq mm4, mm0
movq mm7, mm2
PRED4x4_LOWPASS mm1, mm3, mm0, mm4, mm5
psllq mm1, 56
PALIGNR mm7, mm1, 7, mm3
movq mm3, mm7
lea r1, [r0+r3*2]
movq mm7, mm3
punpckhbw mm3, mm3
punpcklbw mm7, mm7
pshufw mm0, mm3, 0xff
pshufw mm1, mm3, 0xaa
lea r2, [r1+r3*2]
pshufw mm2, mm3, 0x55
pshufw mm3, mm3, 0x00
pshufw mm4, mm7, 0xff
pshufw mm5, mm7, 0xaa
pshufw mm6, mm7, 0x55
pshufw mm7, mm7, 0x00
movq [r0+r3*1], mm0
movq [r0+r3*2], mm1
movq [r1+r3*1], mm2
movq [r1+r3*2], mm3
movq [r2+r3*1], mm4
movq [r2+r3*2], mm5
lea r0, [r2+r3*2]
movq [r0+r3*1], mm6
movq [r0+r3*2], mm7
RET
%endmacro
INIT_MMX
%define PALIGNR PALIGNR_MMX
PRED8x8L_HORIZONTAL mmxext
%define PALIGNR PALIGNR_SSSE3
PRED8x8L_HORIZONTAL ssse3
%endif
;----------------------------------------------------------------------------- ;-----------------------------------------------------------------------------
; void pred4x4_dc_mmxext(uint8_t *src, const uint8_t *topright, int stride) ; void pred4x4_dc_mmxext(uint8_t *src, const uint8_t *topright, int stride)
;----------------------------------------------------------------------------- ;-----------------------------------------------------------------------------

4
libavcodec/x86/h264_intrapred_init.c
View File

@ -63,6 +63,8 @@ void ff_pred8x8l_top_dc_mmxext (uint8_t *src, int has_topleft, int has_topri
void ff_pred8x8l_top_dc_ssse3 (uint8_t *src, int has_topleft, int has_topright, int stride); void ff_pred8x8l_top_dc_ssse3 (uint8_t *src, int has_topleft, int has_topright, int stride);
void ff_pred8x8l_dc_mmxext (uint8_t *src, int has_topleft, int has_topright, int stride); void ff_pred8x8l_dc_mmxext (uint8_t *src, int has_topleft, int has_topright, int stride);
void ff_pred8x8l_dc_ssse3 (uint8_t *src, int has_topleft, int has_topright, int stride); void ff_pred8x8l_dc_ssse3 (uint8_t *src, int has_topleft, int has_topright, int stride);
void ff_pred8x8l_horizontal_mmxext (uint8_t *src, int has_topleft, int has_topright, int stride);
void ff_pred8x8l_horizontal_ssse3 (uint8_t *src, int has_topleft, int has_topright, int stride);
void ff_pred4x4_dc_mmxext (uint8_t *src, const uint8_t *topright, int stride); void ff_pred4x4_dc_mmxext (uint8_t *src, const uint8_t *topright, int stride);
void ff_pred4x4_down_left_mmxext (uint8_t *src, const uint8_t *topright, int stride); void ff_pred4x4_down_left_mmxext (uint8_t *src, const uint8_t *topright, int stride);
void ff_pred4x4_tm_vp8_mmx (uint8_t *src, const uint8_t *topright, int stride); void ff_pred4x4_tm_vp8_mmx (uint8_t *src, const uint8_t *topright, int stride);
@ -103,6 +105,7 @@ void ff_h264_pred_init_x86(H264PredContext *h, int codec_id)
#if CONFIG_GPL #if CONFIG_GPL
h->pred8x8l [TOP_DC_PRED ] = ff_pred8x8l_top_dc_mmxext; h->pred8x8l [TOP_DC_PRED ] = ff_pred8x8l_top_dc_mmxext;
h->pred8x8l [DC_PRED ] = ff_pred8x8l_dc_mmxext; h->pred8x8l [DC_PRED ] = ff_pred8x8l_dc_mmxext;
h->pred8x8l [HOR_PRED ] = ff_pred8x8l_horizontal_mmxext;
#endif #endif
h->pred4x4 [DC_PRED ] = ff_pred4x4_dc_mmxext; h->pred4x4 [DC_PRED ] = ff_pred4x4_dc_mmxext;
#if CONFIG_GPL #if CONFIG_GPL
@ -159,6 +162,7 @@ void ff_h264_pred_init_x86(H264PredContext *h, int codec_id)
#if CONFIG_GPL #if CONFIG_GPL
h->pred8x8l [TOP_DC_PRED ] = ff_pred8x8l_top_dc_ssse3; h->pred8x8l [TOP_DC_PRED ] = ff_pred8x8l_top_dc_ssse3;
h->pred8x8l [DC_PRED ] = ff_pred8x8l_dc_ssse3; h->pred8x8l [DC_PRED ] = ff_pred8x8l_dc_ssse3;
h->pred8x8l [HOR_PRED ] = ff_pred8x8l_horizontal_ssse3;
#endif #endif
if (codec_id == CODEC_ID_VP8) { if (codec_id == CODEC_ID_VP8) {
h->pred8x8 [PLANE_PRED8x8] = ff_pred8x8_tm_vp8_ssse3; h->pred8x8 [PLANE_PRED8x8] = ff_pred8x8_tm_vp8_ssse3;