#define COST_MV( mx, my )\
{\
int cost = h->pixf.fpelcmp[i_pixel]( p_fenc, FENC_STRIDE,\
- &p_fref[(my)*stride+(mx)], stride )\
+ &p_fref_w[(my)*stride+(mx)], stride )\
+ BITS_MVD(mx,my);\
COPY3_IF_LT( bcost, cost, bmx, mx, bmy, my );\
}
#define COST_MV_HPEL( mx, my ) \
{ \
int stride2 = 16; \
- uint8_t *src = h->mc.get_ref( pix, &stride2, m->p_fref, stride, mx, my, bw, bh ); \
+ uint8_t *src = h->mc.get_ref( pix, &stride2, m->p_fref, stride, mx, my, bw, bh, &m->weight[0] ); \
int cost = h->pixf.fpelcmp[i_pixel]( p_fenc, FENC_STRIDE, src, stride2 ) \
+ p_cost_mvx[ mx ] + p_cost_mvy[ my ]; \
COPY3_IF_LT( bpred_cost, cost, bpred_mx, mx, bpred_my, my ); \
#define COST_MV_X3_DIR( m0x, m0y, m1x, m1y, m2x, m2y, costs )\
{\
- uint8_t *pix_base = p_fref + bmx + bmy*stride;\
+ uint8_t *pix_base = p_fref_w + bmx + bmy*stride;\
h->pixf.fpelcmp_x3[i_pixel]( p_fenc,\
pix_base + (m0x) + (m0y)*stride,\
pix_base + (m1x) + (m1y)*stride,\
#define COST_MV_X4_DIR( m0x, m0y, m1x, m1y, m2x, m2y, m3x, m3y, costs )\
{\
- uint8_t *pix_base = p_fref + bmx + bmy*stride;\
+ uint8_t *pix_base = p_fref_w + bmx + bmy*stride;\
h->pixf.fpelcmp_x4[i_pixel]( p_fenc,\
pix_base + (m0x) + (m0y)*stride,\
pix_base + (m1x) + (m1y)*stride,\
#define COST_MV_X4( m0x, m0y, m1x, m1y, m2x, m2y, m3x, m3y )\
{\
- uint8_t *pix_base = p_fref + omx + omy*stride;\
+ uint8_t *pix_base = p_fref_w + omx + omy*stride;\
h->pixf.fpelcmp_x4[i_pixel]( p_fenc,\
pix_base + (m0x) + (m0y)*stride,\
pix_base + (m1x) + (m1y)*stride,\
#define COST_MV_X3_ABS( m0x, m0y, m1x, m1y, m2x, m2y )\
{\
h->pixf.fpelcmp_x3[i_pixel]( p_fenc,\
- p_fref + (m0x) + (m0y)*stride,\
- p_fref + (m1x) + (m1y)*stride,\
- p_fref + (m2x) + (m2y)*stride,\
+ p_fref_w + (m0x) + (m0y)*stride,\
+ p_fref_w + (m1x) + (m1y)*stride,\
+ p_fref_w + (m2x) + (m2y)*stride,\
stride, costs );\
costs[0] += p_cost_mvx[(m0x)<<2]; /* no cost_mvy */\
costs[1] += p_cost_mvx[(m1x)<<2];\
int bpred_mx = 0, bpred_my = 0, bpred_cost = COST_MAX;
int omx, omy, pmx, pmy;
uint8_t *p_fenc = m->p_fenc[0];
- uint8_t *p_fref = m->p_fref[0];
+ uint8_t *p_fref_w = m->p_fref_w;
ALIGNED_ARRAY_16( uint8_t, pix,[16*16] );
int i, j;
COST_MV_HPEL( bmx, bmy );
for( i = 0; i < i_mvc; i++ )
{
- if( *(uint32_t*)mvc[i] && (bmv - *(uint32_t*)mvc[i]) )
+ if( M32( mvc[i] ) && (bmv - M32( mvc[i] )) )
{
int mx = x264_clip3( mvc[i][0], mv_x_min*4, mv_x_max*4 );
int my = x264_clip3( mvc[i][1], mv_y_min*4, mv_y_max*4 );
else
{
int dir = 0;
- uint8_t *pix_base = p_fref + omx + (omy-4*i)*stride;
+ uint8_t *pix_base = p_fref_w + omx + (omy-4*i)*stride;
int dy = i*stride;
#define SADS(k,x0,y0,x1,y1,x2,y2,x3,y3)\
h->pixf.fpelcmp_x4[i_pixel]( p_fenc,\
mvsad_t *mvsads = (mvsad_t *)(xs + ((width+15)&~15));
int nmvsad = 0, limit;
int sad_thresh = i_me_range <= 16 ? 10 : i_me_range <= 24 ? 11 : 12;
- int bsad = h->pixf.sad[i_pixel]( p_fenc, FENC_STRIDE, p_fref+bmy*stride+bmx, stride )
+ int bsad = h->pixf.sad[i_pixel]( p_fenc, FENC_STRIDE, p_fref_w+bmy*stride+bmx, stride )
+ BITS_MVD( bmx, bmy );
for( my = min_y; my <= max_y; my++ )
{
cost_fpel_mvx+min_x, xs, width, bsad*17/16 );
for( i=0; i<xn-2; i+=3 )
{
- uint8_t *ref = p_fref+min_x+my*stride;
+ uint8_t *ref = p_fref_w+min_x+my*stride;
int sads[3];
h->pixf.sad_x3[i_pixel]( p_fenc, ref+xs[i], ref+xs[i+1], ref+xs[i+2], stride, sads );
for( j=0; j<3; j++ )
for( ; i<xn; i++ )
{
int mx = min_x+xs[i];
- int sad = h->pixf.sad[i_pixel]( p_fenc, FENC_STRIDE, p_fref+mx+my*stride, stride )
+ int sad = h->pixf.sad[i_pixel]( p_fenc, FENC_STRIDE, p_fref_w+mx+my*stride, stride )
+ cost_fpel_mvx[xs[i]];
if( sad < bsad*sad_thresh>>3 )
{
{
/* mvsad_t is not guaranteed to be 8 bytes on all archs, so check before using explicit write-combining */
if( sizeof( mvsad_t ) == sizeof( uint64_t ) )
- *(uint64_t*)&mvsads[i] = *(uint64_t*)&mvsads[j];
+ CP64( &mvsads[i], &mvsads[j] );
else
mvsads[i] = mvsads[j];
i += mvsads[j].sad <= sad_thresh;
nmvsad--;
mvsads[bi] = mvsads[nmvsad];
if( sizeof( mvsad_t ) == sizeof( uint64_t ) )
- *(uint64_t*)&mvsads[bi] = *(uint64_t*)&mvsads[nmvsad];
+ CP64( &mvsads[bi], &mvsads[nmvsad] );
else
mvsads[bi] = mvsads[nmvsad];
}
if( m->i_pixel <= PIXEL_8x8 && h->sh.i_type == SLICE_TYPE_P )
m->cost -= m->i_ref_cost;
-
+
refine_subpel( h, m, hpel, qpel, NULL, 1 );
}
#define COST_MV_SAD( mx, my ) \
{ \
int stride = 16; \
- uint8_t *src = h->mc.get_ref( pix[0], &stride, m->p_fref, m->i_stride[0], mx, my, bw, bh ); \
+ uint8_t *src = h->mc.get_ref( pix[0], &stride, m->p_fref, m->i_stride[0], mx, my, bw, bh, &m->weight[0] ); \
int cost = h->pixf.fpelcmp[i_pixel]( m->p_fenc[0], FENC_STRIDE, src, stride ) \
+ p_cost_mvx[ mx ] + p_cost_mvy[ my ]; \
COPY3_IF_LT( bcost, cost, bmx, mx, bmy, my ); \
if( b_refine_qpel || (dir^1) != odir ) \
{ \
int stride = 16; \
- uint8_t *src = h->mc.get_ref( pix[0], &stride, m->p_fref, m->i_stride[0], mx, my, bw, bh ); \
+ uint8_t *src = h->mc.get_ref( pix[0], &stride, m->p_fref, m->i_stride[0], mx, my, bw, bh, &m->weight[0] ); \
int cost = h->pixf.mbcmp_unaligned[i_pixel]( m->p_fenc[0], FENC_STRIDE, src, stride ) \
+ p_cost_mvx[ mx ] + p_cost_mvy[ my ]; \
if( b_chroma_me && cost < bcost ) \
{ \
h->mc.mc_chroma( pix[0], 8, m->p_fref[4], m->i_stride[1], mx, my + mvy_offset, bw/2, bh/2 ); \
+ if( m->weight[1].weightfn ) \
+ m->weight[1].weightfn[x264_pixel_size[i_pixel].w>>3]( pix[0], 8, pix[0], 8, \
+ &m->weight[1], x264_pixel_size[i_pixel].h>>1 ); \
cost += h->pixf.mbcmp[i_pixel+3]( m->p_fenc[1], FENC_STRIDE, pix[0], 8 ); \
if( cost < bcost ) \
{ \
h->mc.mc_chroma( pix[0], 8, m->p_fref[5], m->i_stride[1], mx, my + mvy_offset, bw/2, bh/2 ); \
cost += h->pixf.mbcmp[i_pixel+3]( m->p_fenc[2], FENC_STRIDE, pix[0], 8 ); \
+ if( m->weight[2].weightfn ) \
+ m->weight[2].weightfn[x264_pixel_size[i_pixel].w>>3]( pix[0], 8, pix[0], 8, \
+ &m->weight[2], x264_pixel_size[i_pixel].h>>1 ); \
} \
} \
if( cost < bcost ) \
int costs[4];
int stride = 32; // candidates are either all hpel or all qpel, so one stride is enough
uint8_t *src0, *src1, *src2, *src3;
- src0 = h->mc.get_ref( pix[0], &stride, m->p_fref, m->i_stride[0], omx, omy-2, bw, bh+1 );
- src2 = h->mc.get_ref( pix[1], &stride, m->p_fref, m->i_stride[0], omx-2, omy, bw+4, bh );
+ src0 = h->mc.get_ref( pix[0], &stride, m->p_fref, m->i_stride[0], omx, omy-2, bw, bh+1, &m->weight[0] );
+ src2 = h->mc.get_ref( pix[1], &stride, m->p_fref, m->i_stride[0], omx-2, omy, bw+4, bh, &m->weight[0] );
src1 = src0 + stride;
src3 = src2 + 1;
h->pixf.fpelcmp_x4[i_pixel]( m->p_fenc[0], src0, src1, src2, src3, stride, costs );
int mvx = om##list##x+dx;\
int mvy = om##list##y+dy;\
stride##list[i] = bw;\
- src##list[i] = h->mc.get_ref( pixy_buf[list][i], &stride##list[i], m->p_fref, m->i_stride[0], mvx, mvy, bw, bh ); \
+ src##list[i] = h->mc.get_ref( pixy_buf[list][i], &stride##list[i], m->p_fref, m->i_stride[0], mvx, mvy, bw, bh, weight_none ); \
if( rd )\
{\
h->mc.mc_chroma( pixu_buf[list][i], 8, m->p_fref[4], m->i_stride[1], mvx, mvy + mv##list##y_offset, bw>>1, bh>>1 );\
if( cost < bcost * SATD_THRESH )
{
bcost = X264_MIN( cost, bcost );
- *(uint32_t*)cache0_mv = *(uint32_t*)cache0_mv2 = pack16to32_mask(m0x,m0y);
- *(uint32_t*)cache1_mv = *(uint32_t*)cache1_mv2 = pack16to32_mask(m1x,m1y);
+ M32( cache0_mv ) = pack16to32_mask(m0x,m0y);
+ M32( cache0_mv2 ) = pack16to32_mask(m0x,m0y);
+ M32( cache1_mv ) = pack16to32_mask(m1x,m1y);
+ M32( cache1_mv2 ) = pack16to32_mask(m1x,m1y);
h->mc.avg[i_pixel+3]( pixu, FDEC_STRIDE, pixu_buf[0][i0], 8, pixu_buf[1][i1], 8, i_weight );
h->mc.avg[i_pixel+3]( pixv, FDEC_STRIDE, pixv_buf[0][i0], 8, pixv_buf[1][i1], 8, i_weight );
uint64_t costrd = x264_rd_cost_part( h, i_lambda2, i8*4, m0->i_pixel );
if( !avoid_mvp || !(mx == pmx && my == pmy) ) \
{ \
int stride = 16; \
- uint8_t *src = h->mc.get_ref( pix, &stride, m->p_fref, m->i_stride[0], mx, my, bw*4, bh*4 ); \
- dst = h->pixf.mbcmp_unaligned[i_pixel]( m->p_fenc[0], FENC_STRIDE, src, stride ) \
+ uint8_t *src = h->mc.get_ref( pix, &stride, m->p_fref, m->i_stride[0], mx, my, bw*4, bh*4, &m->weight[0] ); \
+ dst = h->pixf.mbcmp[i_pixel]( m->p_fenc[0], FENC_STRIDE, src, stride ) \
+ p_cost_mvx[mx] + p_cost_mvy[my]; \
COPY1_IF_LT( bsatd, dst ); \
} \
if( satd <= bsatd * SATD_THRESH ) \
{ \
uint64_t cost; \
- *(uint32_t*)cache_mv = *(uint32_t*)cache_mv2 = pack16to32_mask(mx,my); \
+ M32( cache_mv ) = pack16to32_mask(mx,my); \
+ M32( cache_mv2 ) = pack16to32_mask(mx,my); \
cost = x264_rd_cost_part( h, i_lambda2, i4, m->i_pixel ); \
COPY4_IF_LT( bcost, cost, bmx, mx, bmy, my, dir, do_dir?mdir:dir ); \
} \
projects / x264 / blobdiff