Add x264 SSE2 iDCT functions to H.264 decoder.

Originally committed as revision 16409 to svn://svn.ffmpeg.org/ffmpeg/trunk
This commit is contained in:
Jason Garrett-Glaser 2009-01-03 00:46:17 +00:00
parent c96da3edc3
commit 37fed10087
5 changed files with 379 additions and 2 deletions

View File

@ -407,6 +407,7 @@ MMX-OBJS-$(CONFIG_VP6F_DECODER) += x86/vp3dsp_mmx.o x86/vp3dsp_sse2.o
MMX-OBJS-$(CONFIG_WMV3_DECODER) += x86/vc1dsp_mmx.o
MMX-OBJS-$(HAVE_YASM) += x86/dsputil_yasm.o \
x86/h264_deblock_sse2.o \
x86/h264_idct_sse2.o \
OBJS-$(HAVE_MMX) += x86/cpuid.o \
x86/dnxhd_mmx.o \

View File

@ -2872,14 +2872,17 @@ void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx)
c->h264_v_loop_filter_luma_intra = ff_x264_deblock_v_luma_intra_mmxext;
c->h264_h_loop_filter_luma_intra = ff_x264_deblock_h_luma_intra_mmxext;
#endif
#if defined(ARCH_X86_64) || !defined(__ICC) || __ICC > 1100
if( mm_flags&FF_MM_SSE2 ){
#if defined(ARCH_X86_64) || !defined(__ICC) || __ICC > 1100
c->h264_v_loop_filter_luma = ff_x264_deblock_v_luma_sse2;
c->h264_h_loop_filter_luma = ff_x264_deblock_h_luma_sse2;
c->h264_v_loop_filter_luma_intra = ff_x264_deblock_v_luma_intra_sse2;
c->h264_h_loop_filter_luma_intra = ff_x264_deblock_h_luma_intra_sse2;
}
#endif
c->h264_idct_add16 = ff_h264_idct_add16_sse2;
c->h264_idct_add8 = ff_h264_idct_add8_sse2;
c->h264_idct_add16intra = ff_h264_idct_add16intra_sse2;
}
}
#endif

View File

@ -0,0 +1,61 @@
;*****************************************************************************
;* dct-a.asm: h264 encoder library
;*****************************************************************************
;* Copyright (C) 2003-2008 x264 project
;*
;* Authors: Laurent Aimar <fenrir@via.ecp.fr>
;* Loren Merritt <lorenm@u.washington.edu>
;* Holger Lubitz <hal@duncan.ol.sub.de>
;* Min Chen <chenm001.163.com>
;*
;* This program is free software; you can redistribute it and/or modify
;* it under the terms of the GNU General Public License as published by
;* the Free Software Foundation; either version 2 of the License, or
;* (at your option) any later version.
;*
;* This program is distributed in the hope that it will be useful,
;* but WITHOUT ANY WARRANTY; without even the implied warranty of
;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
;* GNU General Public License for more details.
;*
;* You should have received a copy of the GNU General Public License
;* along with this program; if not, write to the Free Software
;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111, USA.
;*****************************************************************************
%include "x86inc.asm"
%include "x86util.asm"
SECTION_RODATA
pw_32: times 8 dw 32
SECTION .text
%macro IDCT4_1D 6
SUMSUB_BA m%3, m%1
SUMSUBD2_AB m%2, m%4, m%6, m%5
SUMSUB_BADC m%2, m%3, m%5, m%1
SWAP %1, %2, %5, %4, %3
%endmacro
INIT_XMM
cglobal x264_add8x4_idct_sse2, 3,3
movq m0, [r1+ 0]
movq m1, [r1+ 8]
movq m2, [r1+16]
movq m3, [r1+24]
movhps m0, [r1+32]
movhps m1, [r1+40]
movhps m2, [r1+48]
movhps m3, [r1+56]
IDCT4_1D 0,1,2,3,4,5
TRANSPOSE2x4x4W 0,1,2,3,4
paddw m0, [pw_32 GLOBAL]
IDCT4_1D 0,1,2,3,4,5
pxor m7, m7
STORE_DIFF m0, m4, m7, [r0]
STORE_DIFF m1, m4, m7, [r0+r2]
lea r0, [r0+r2*2]
STORE_DIFF m2, m4, m7, [r0]
STORE_DIFF m3, m4, m7, [r0+r2]
RET

View File

@ -472,6 +472,78 @@ static void ff_h264_idct_add8_mmx2(uint8_t **dest, const int *block_offset, DCTE
}
}
#if defined(CONFIG_GPL) && defined(HAVE_YASM)
static void ff_h264_idct_dc_add8_mmx2(uint8_t *dst, int16_t *block, int stride)
{
__asm__ volatile(
"movd %0, %%mm0 \n\t" // 0 0 X D
"punpcklwd %1, %%mm0 \n\t" // x X d D
"paddsw %2, %%mm0 \n\t"
"psraw $6, %%mm0 \n\t"
"punpcklwd %%mm0, %%mm0 \n\t" // d d D D
"pxor %%mm1, %%mm1 \n\t" // 0 0 0 0
"psubw %%mm0, %%mm1 \n\t" // -d-d-D-D
"packuswb %%mm1, %%mm0 \n\t" // -d-d-D-D d d D D
"pshufw $0xFA, %%mm0, %%mm1 \n\t" // -d-d-d-d-D-D-D-D
"punpcklwd %%mm0, %%mm0 \n\t" // d d d d D D D D
::"m"(block[ 0]),
"m"(block[16]),
"m"(ff_pw_32)
);
__asm__ volatile(
"movq %0, %%mm2 \n\t"
"movq %1, %%mm3 \n\t"
"movq %2, %%mm4 \n\t"
"movq %3, %%mm5 \n\t"
"paddusb %%mm0, %%mm2 \n\t"
"paddusb %%mm0, %%mm3 \n\t"
"paddusb %%mm0, %%mm4 \n\t"
"paddusb %%mm0, %%mm5 \n\t"
"psubusb %%mm1, %%mm2 \n\t"
"psubusb %%mm1, %%mm3 \n\t"
"psubusb %%mm1, %%mm4 \n\t"
"psubusb %%mm1, %%mm5 \n\t"
"movq %%mm2, %0 \n\t"
"movq %%mm3, %1 \n\t"
"movq %%mm4, %2 \n\t"
"movq %%mm5, %3 \n\t"
:"+m"(*(uint64_t*)(dst+0*stride)),
"+m"(*(uint64_t*)(dst+1*stride)),
"+m"(*(uint64_t*)(dst+2*stride)),
"+m"(*(uint64_t*)(dst+3*stride))
);
}
extern void ff_x264_add8x4_idct_sse2(uint8_t *dst, int16_t *block, int stride);
static void ff_h264_idct_add16_sse2(uint8_t *dst, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]){
int i;
for(i=0; i<16; i+=2)
if(nnzc[ scan8[i+0] ]|nnzc[ scan8[i+1] ])
ff_x264_add8x4_idct_sse2 (dst + block_offset[i], block + i*16, stride);
}
static void ff_h264_idct_add16intra_sse2(uint8_t *dst, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]){
int i;
for(i=0; i<16; i+=2){
if(nnzc[ scan8[i+0] ]|nnzc[ scan8[i+1] ])
ff_x264_add8x4_idct_sse2 (dst + block_offset[i], block + i*16, stride);
else if(block[i*16]|block[i*16+16])
ff_h264_idct_dc_add8_mmx2(dst + block_offset[i], block + i*16, stride);
}
}
static void ff_h264_idct_add8_sse2(uint8_t **dest, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]){
int i;
for(i=16; i<16+8; i++){
if(nnzc[ scan8[i+0] ]|nnzc[ scan8[i+1] ])
ff_x264_add8x4_idct_sse2 (dest[(i&4)>>2] + block_offset[i], block + i*16, stride);
else if(block[i*16]|block[i*16+16])
ff_h264_idct_dc_add8_mmx2(dest[(i&4)>>2] + block_offset[i], block + i*16, stride);
}
}
#endif
/***********************************/
/* deblocking */

240
libavcodec/x86/x86util.asm Normal file
View File

@ -0,0 +1,240 @@
;*****************************************************************************
;* x86inc.asm
;*****************************************************************************
;* Copyright (C) 2008 Loren Merritt <lorenm@u.washington.edu>
;*
;* This program is free software; you can redistribute it and/or modify
;* it under the terms of the GNU General Public License as published by
;* the Free Software Foundation; either version 2 of the License, or
;* (at your option) any later version.
;*
;* This program is distributed in the hope that it will be useful,
;* but WITHOUT ANY WARRANTY; without even the implied warranty of
;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
;* GNU General Public License for more details.
;*
;* You should have received a copy of the GNU General Public License
;* along with this program; if not, write to the Free Software
;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02111, USA.
;*****************************************************************************
%macro SBUTTERFLY 4
mova m%4, m%2
punpckl%1 m%2, m%3
punpckh%1 m%4, m%3
SWAP %3, %4
%endmacro
%macro TRANSPOSE4x4W 5
SBUTTERFLY wd, %1, %2, %5
SBUTTERFLY wd, %3, %4, %5
SBUTTERFLY dq, %1, %3, %5
SBUTTERFLY dq, %2, %4, %5
SWAP %2, %3
%endmacro
%macro TRANSPOSE2x4x4W 5
SBUTTERFLY wd, %1, %2, %5
SBUTTERFLY wd, %3, %4, %5
SBUTTERFLY dq, %1, %3, %5
SBUTTERFLY dq, %2, %4, %5
SBUTTERFLY qdq, %1, %2, %5
SBUTTERFLY qdq, %3, %4, %5
%endmacro
%macro TRANSPOSE4x4D 5
SBUTTERFLY dq, %1, %2, %5
SBUTTERFLY dq, %3, %4, %5
SBUTTERFLY qdq, %1, %3, %5
SBUTTERFLY qdq, %2, %4, %5
SWAP %2, %3
%endmacro
%macro TRANSPOSE8x8W 9-11
%ifdef ARCH_X86_64
SBUTTERFLY wd, %1, %2, %9
SBUTTERFLY wd, %3, %4, %9
SBUTTERFLY wd, %5, %6, %9
SBUTTERFLY wd, %7, %8, %9
SBUTTERFLY dq, %1, %3, %9
SBUTTERFLY dq, %2, %4, %9
SBUTTERFLY dq, %5, %7, %9
SBUTTERFLY dq, %6, %8, %9
SBUTTERFLY qdq, %1, %5, %9
SBUTTERFLY qdq, %2, %6, %9
SBUTTERFLY qdq, %3, %7, %9
SBUTTERFLY qdq, %4, %8, %9
SWAP %2, %5
SWAP %4, %7
%else
; in: m0..m7, unless %11 in which case m6 is in %9
; out: m0..m7, unless %11 in which case m4 is in %10
; spills into %9 and %10
%if %0<11
movdqa %9, m%7
%endif
SBUTTERFLY wd, %1, %2, %7
movdqa %10, m%2
movdqa m%7, %9
SBUTTERFLY wd, %3, %4, %2
SBUTTERFLY wd, %5, %6, %2
SBUTTERFLY wd, %7, %8, %2
SBUTTERFLY dq, %1, %3, %2
movdqa %9, m%3
movdqa m%2, %10
SBUTTERFLY dq, %2, %4, %3
SBUTTERFLY dq, %5, %7, %3
SBUTTERFLY dq, %6, %8, %3
SBUTTERFLY qdq, %1, %5, %3
SBUTTERFLY qdq, %2, %6, %3
movdqa %10, m%2
movdqa m%3, %9
SBUTTERFLY qdq, %3, %7, %2
SBUTTERFLY qdq, %4, %8, %2
SWAP %2, %5
SWAP %4, %7
%if 0<11
movdqa m%5, %10
%endif
%endif
%endmacro
%macro ABS1_MMX 2 ; a, tmp
pxor %2, %2
psubw %2, %1
pmaxsw %1, %2
%endmacro
%macro ABS2_MMX 4 ; a, b, tmp0, tmp1
pxor %3, %3
pxor %4, %4
psubw %3, %1
psubw %4, %2
pmaxsw %1, %3
pmaxsw %2, %4
%endmacro
%macro ABS1_SSSE3 2
pabsw %1, %1
%endmacro
%macro ABS2_SSSE3 4
pabsw %1, %1
pabsw %2, %2
%endmacro
%define ABS1 ABS1_MMX
%define ABS2 ABS2_MMX
%macro ABS4 6
ABS2 %1, %2, %5, %6
ABS2 %3, %4, %5, %6
%endmacro
%macro SPLATB_MMX 3
movd %1, [%2-3] ;to avoid crossing a cacheline
punpcklbw %1, %1
%if mmsize==16
pshuflw %1, %1, 0xff
punpcklqdq %1, %1
%else
pshufw %1, %1, 0xff
%endif
%endmacro
%macro SPLATB_SSSE3 3
movd %1, [%2-3]
pshufb %1, %3
%endmacro
%macro PALIGNR_MMX 4
%ifnidn %4, %2
mova %4, %2
%endif
%if mmsize == 8
psllq %1, (8-%3)*8
psrlq %4, %3*8
%else
pslldq %1, 16-%3
psrldq %4, %3
%endif
por %1, %4
%endmacro
%macro PALIGNR_SSSE3 4
palignr %1, %2, %3
%endmacro
%macro SUMSUB_BA 2
paddw %1, %2
paddw %2, %2
psubw %2, %1
%endmacro
%macro SUMSUB_BADC 4
paddw %1, %2
paddw %3, %4
paddw %2, %2
paddw %4, %4
psubw %2, %1
psubw %4, %3
%endmacro
%macro HADAMARD8_1D 8
SUMSUB_BADC %1, %5, %2, %6
SUMSUB_BADC %3, %7, %4, %8
SUMSUB_BADC %1, %3, %2, %4
SUMSUB_BADC %5, %7, %6, %8
SUMSUB_BADC %1, %2, %3, %4
SUMSUB_BADC %5, %6, %7, %8
%endmacro
%macro SUMSUB2_AB 3
mova %3, %1
paddw %1, %1
paddw %1, %2
psubw %3, %2
psubw %3, %2
%endmacro
%macro SUMSUBD2_AB 4
mova %4, %1
mova %3, %2
psraw %2, 1
psraw %4, 1
paddw %1, %2
psubw %4, %3
%endmacro
%macro LOAD_DIFF 5
%ifidn %3, none
movh %1, %4
movh %2, %5
punpcklbw %1, %2
punpcklbw %2, %2
psubw %1, %2
%else
movh %1, %4
punpcklbw %1, %3
movh %2, %5
punpcklbw %2, %3
psubw %1, %2
%endif
%endmacro
%macro LOAD_DIFF_8x4P 6-8 r0,r2 ; 4x dest, 2x temp, 2x pointer
LOAD_DIFF %1, %5, none, [%7], [%8]
LOAD_DIFF %2, %6, none, [%7+r1], [%8+r3]
LOAD_DIFF %3, %5, none, [%7+2*r1], [%8+2*r3]
LOAD_DIFF %4, %6, none, [%7+r4], [%8+r5]
%endmacro
%macro STORE_DIFF 4
psraw %1, 6
movh %2, %4
punpcklbw %2, %3
paddsw %1, %2
packuswb %1, %1
movh %4, %1
%endmacro