Merge "Translates -g from LDFLAGS as --debug in armlink_adapter.sh"
diff --git a/vp8/common/arm/loopfilter_arm.c b/vp8/common/arm/loopfilter_arm.c
index a81c505..684a7f1 100644
--- a/vp8/common/arm/loopfilter_arm.c
+++ b/vp8/common/arm/loopfilter_arm.c
@@ -41,13 +41,13 @@
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
(void) simpler_lpf;
- vp8_mbloop_filter_horizontal_edge_armv6(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
+ vp8_mbloop_filter_horizontal_edge_armv6(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
if (u_ptr)
- vp8_mbloop_filter_horizontal_edge_armv6(u_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr, 1);
+ vp8_mbloop_filter_horizontal_edge_armv6(u_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
if (v_ptr)
- vp8_mbloop_filter_horizontal_edge_armv6(v_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr, 1);
+ vp8_mbloop_filter_horizontal_edge_armv6(v_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
}
void vp8_loop_filter_mbhs_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
@@ -57,7 +57,7 @@
(void) v_ptr;
(void) uv_stride;
(void) simpler_lpf;
- vp8_loop_filter_simple_horizontal_edge_armv6(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
+ vp8_loop_filter_simple_horizontal_edge_armv6(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
}
/* Vertical MB Filtering */
@@ -65,13 +65,13 @@
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
(void) simpler_lpf;
- vp8_mbloop_filter_vertical_edge_armv6(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
+ vp8_mbloop_filter_vertical_edge_armv6(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
if (u_ptr)
- vp8_mbloop_filter_vertical_edge_armv6(u_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr, 1);
+ vp8_mbloop_filter_vertical_edge_armv6(u_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
if (v_ptr)
- vp8_mbloop_filter_vertical_edge_armv6(v_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr, 1);
+ vp8_mbloop_filter_vertical_edge_armv6(v_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
}
void vp8_loop_filter_mbvs_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
@@ -81,7 +81,7 @@
(void) v_ptr;
(void) uv_stride;
(void) simpler_lpf;
- vp8_loop_filter_simple_vertical_edge_armv6(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
+ vp8_loop_filter_simple_vertical_edge_armv6(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
}
/* Horizontal B Filtering */
@@ -94,10 +94,10 @@
vp8_loop_filter_horizontal_edge_armv6(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
if (u_ptr)
- vp8_loop_filter_horizontal_edge_armv6(u_ptr + 4 * uv_stride, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr, 1);
+ vp8_loop_filter_horizontal_edge_armv6(u_ptr + 4 * uv_stride, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
if (v_ptr)
- vp8_loop_filter_horizontal_edge_armv6(v_ptr + 4 * uv_stride, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr, 1);
+ vp8_loop_filter_horizontal_edge_armv6(v_ptr + 4 * uv_stride, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
}
void vp8_loop_filter_bhs_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
@@ -122,10 +122,10 @@
vp8_loop_filter_vertical_edge_armv6(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
if (u_ptr)
- vp8_loop_filter_vertical_edge_armv6(u_ptr + 4, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr, 1);
+ vp8_loop_filter_vertical_edge_armv6(u_ptr + 4, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
if (v_ptr)
- vp8_loop_filter_vertical_edge_armv6(v_ptr + 4, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr, 1);
+ vp8_loop_filter_vertical_edge_armv6(v_ptr + 4, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
}
void vp8_loop_filter_bvs_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
@@ -148,10 +148,10 @@
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
(void) simpler_lpf;
- vp8_mbloop_filter_horizontal_edge_y_neon(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
+ vp8_mbloop_filter_horizontal_edge_y_neon(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
if (u_ptr)
- vp8_mbloop_filter_horizontal_edge_uv_neon(u_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr, v_ptr);
+ vp8_mbloop_filter_horizontal_edge_uv_neon(u_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, v_ptr);
}
void vp8_loop_filter_mbhs_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
@@ -161,7 +161,7 @@
(void) v_ptr;
(void) uv_stride;
(void) simpler_lpf;
- vp8_loop_filter_simple_horizontal_edge_neon(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
+ vp8_loop_filter_simple_horizontal_edge_neon(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
}
/* Vertical MB Filtering */
@@ -169,10 +169,10 @@
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
(void) simpler_lpf;
- vp8_mbloop_filter_vertical_edge_y_neon(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
+ vp8_mbloop_filter_vertical_edge_y_neon(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
if (u_ptr)
- vp8_mbloop_filter_vertical_edge_uv_neon(u_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr, v_ptr);
+ vp8_mbloop_filter_vertical_edge_uv_neon(u_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, v_ptr);
}
void vp8_loop_filter_mbvs_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
@@ -182,7 +182,7 @@
(void) v_ptr;
(void) uv_stride;
(void) simpler_lpf;
- vp8_loop_filter_simple_vertical_edge_neon(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
+ vp8_loop_filter_simple_vertical_edge_neon(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
}
/* Horizontal B Filtering */
@@ -195,7 +195,7 @@
vp8_loop_filter_horizontal_edge_y_neon(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
if (u_ptr)
- vp8_loop_filter_horizontal_edge_uv_neon(u_ptr + 4 * uv_stride, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr, v_ptr + 4 * uv_stride);
+ vp8_loop_filter_horizontal_edge_uv_neon(u_ptr + 4 * uv_stride, uv_stride, lfi->flim, lfi->lim, lfi->thr, v_ptr + 4 * uv_stride);
}
void vp8_loop_filter_bhs_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
@@ -220,7 +220,7 @@
vp8_loop_filter_vertical_edge_y_neon(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
if (u_ptr)
- vp8_loop_filter_vertical_edge_uv_neon(u_ptr + 4, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr, v_ptr + 4);
+ vp8_loop_filter_vertical_edge_uv_neon(u_ptr + 4, uv_stride, lfi->flim, lfi->lim, lfi->thr, v_ptr + 4);
}
void vp8_loop_filter_bvs_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
diff --git a/vp8/common/arm/vpx_asm_offsets.c b/vp8/common/arm/vpx_asm_offsets.c
index 5baf8cc..074d350 100644
--- a/vp8/common/arm/vpx_asm_offsets.c
+++ b/vp8/common/arm/vpx_asm_offsets.c
@@ -45,15 +45,6 @@
#endif
#if CONFIG_VP8_DECODER
-DEFINE(mb_diff, offsetof(MACROBLOCKD, diff));
-DEFINE(mb_predictor, offsetof(MACROBLOCKD, predictor));
-DEFINE(mb_dst_y_stride, offsetof(MACROBLOCKD, dst.y_stride));
-DEFINE(mb_dst_y_buffer, offsetof(MACROBLOCKD, dst.y_buffer));
-DEFINE(mb_dst_u_buffer, offsetof(MACROBLOCKD, dst.u_buffer));
-DEFINE(mb_dst_v_buffer, offsetof(MACROBLOCKD, dst.v_buffer));
-DEFINE(mb_up_available, offsetof(MACROBLOCKD, up_available));
-DEFINE(mb_left_available, offsetof(MACROBLOCKD, left_available));
-
DEFINE(detok_scan, offsetof(DETOK, scan));
DEFINE(detok_ptr_block2leftabove, offsetof(DETOK, ptr_block2leftabove));
DEFINE(detok_coef_tree_ptr, offsetof(DETOK, vp8_coef_tree_ptr));
diff --git a/vp8/common/findnearmv.c b/vp8/common/findnearmv.c
index cab0403..50d1b3b 100644
--- a/vp8/common/findnearmv.c
+++ b/vp8/common/findnearmv.c
@@ -11,8 +11,6 @@
#include "findnearmv.h"
-#define FINDNEAR_SEARCH_SITES 3
-
/* Predict motion vectors using those from already-decoded nearby blocks.
Note that we only consider one 4x4 subblock from each candidate 16x16
macroblock. */
diff --git a/vp8/common/loopfilter.c b/vp8/common/loopfilter.c
index 90e1d9c..37c5b77 100644
--- a/vp8/common/loopfilter.c
+++ b/vp8/common/loopfilter.c
@@ -28,13 +28,13 @@
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
(void) simpler_lpf;
- vp8_mbloop_filter_horizontal_edge_c(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
+ vp8_mbloop_filter_horizontal_edge_c(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
if (u_ptr)
- vp8_mbloop_filter_horizontal_edge_c(u_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr, 1);
+ vp8_mbloop_filter_horizontal_edge_c(u_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
if (v_ptr)
- vp8_mbloop_filter_horizontal_edge_c(v_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr, 1);
+ vp8_mbloop_filter_horizontal_edge_c(v_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
}
void vp8_loop_filter_mbhs_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
@@ -44,7 +44,7 @@
(void) v_ptr;
(void) uv_stride;
(void) simpler_lpf;
- vp8_loop_filter_simple_horizontal_edge_c(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
+ vp8_loop_filter_simple_horizontal_edge_c(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
}
/* Vertical MB Filtering */
@@ -52,13 +52,13 @@
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
(void) simpler_lpf;
- vp8_mbloop_filter_vertical_edge_c(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
+ vp8_mbloop_filter_vertical_edge_c(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
if (u_ptr)
- vp8_mbloop_filter_vertical_edge_c(u_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr, 1);
+ vp8_mbloop_filter_vertical_edge_c(u_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
if (v_ptr)
- vp8_mbloop_filter_vertical_edge_c(v_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr, 1);
+ vp8_mbloop_filter_vertical_edge_c(v_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
}
void vp8_loop_filter_mbvs_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
@@ -68,7 +68,7 @@
(void) v_ptr;
(void) uv_stride;
(void) simpler_lpf;
- vp8_loop_filter_simple_vertical_edge_c(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
+ vp8_loop_filter_simple_vertical_edge_c(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
}
/* Horizontal B Filtering */
@@ -81,10 +81,10 @@
vp8_loop_filter_horizontal_edge_c(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
if (u_ptr)
- vp8_loop_filter_horizontal_edge_c(u_ptr + 4 * uv_stride, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr, 1);
+ vp8_loop_filter_horizontal_edge_c(u_ptr + 4 * uv_stride, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
if (v_ptr)
- vp8_loop_filter_horizontal_edge_c(v_ptr + 4 * uv_stride, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr, 1);
+ vp8_loop_filter_horizontal_edge_c(v_ptr + 4 * uv_stride, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
}
void vp8_loop_filter_bhs_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
@@ -109,10 +109,10 @@
vp8_loop_filter_vertical_edge_c(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
if (u_ptr)
- vp8_loop_filter_vertical_edge_c(u_ptr + 4, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr, 1);
+ vp8_loop_filter_vertical_edge_c(u_ptr + 4, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
if (v_ptr)
- vp8_loop_filter_vertical_edge_c(v_ptr + 4, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr, 1);
+ vp8_loop_filter_vertical_edge_c(v_ptr + 4, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
}
void vp8_loop_filter_bvs_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
@@ -137,8 +137,6 @@
int block_inside_limit = 0;
int HEVThresh;
- const int yhedge_boost = 2;
- const int uvhedge_boost = 2;
/* For each possible value for the loop filter fill out a "loop_filter_info" entry. */
for (i = 0; i <= MAX_LOOP_FILTER; i++)
@@ -182,15 +180,9 @@
for (j = 0; j < 16; j++)
{
lfi[i].lim[j] = block_inside_limit;
- lfi[i].mbflim[j] = filt_lvl + yhedge_boost;
- lfi[i].mbthr[j] = HEVThresh;
+ lfi[i].mbflim[j] = filt_lvl + 2;
lfi[i].flim[j] = filt_lvl;
lfi[i].thr[j] = HEVThresh;
- lfi[i].uvlim[j] = block_inside_limit;
- lfi[i].uvmbflim[j] = filt_lvl + uvhedge_boost;
- lfi[i].uvmbthr[j] = HEVThresh;
- lfi[i].uvflim[j] = filt_lvl;
- lfi[i].uvthr[j] = HEVThresh;
}
}
@@ -249,15 +241,9 @@
for (j = 0; j < 16; j++)
{
/*lfi[i].lim[j] = block_inside_limit;
- lfi[i].mbflim[j] = filt_lvl+yhedge_boost;*/
- lfi[i].mbthr[j] = HEVThresh;
+ lfi[i].mbflim[j] = filt_lvl+2;*/
/*lfi[i].flim[j] = filt_lvl;*/
lfi[i].thr[j] = HEVThresh;
- /*lfi[i].uvlim[j] = block_inside_limit;
- lfi[i].uvmbflim[j] = filt_lvl+uvhedge_boost;*/
- lfi[i].uvmbthr[j] = HEVThresh;
- /*lfi[i].uvflim[j] = filt_lvl;*/
- lfi[i].uvthr[j] = HEVThresh;
}
}
}
diff --git a/vp8/common/loopfilter.h b/vp8/common/loopfilter.h
index e456834..2e5997c 100644
--- a/vp8/common/loopfilter.h
+++ b/vp8/common/loopfilter.h
@@ -32,12 +32,6 @@
DECLARE_ALIGNED(16, signed char, flim[16]);
DECLARE_ALIGNED(16, signed char, thr[16]);
DECLARE_ALIGNED(16, signed char, mbflim[16]);
- DECLARE_ALIGNED(16, signed char, mbthr[16]);
- DECLARE_ALIGNED(16, signed char, uvlim[16]);
- DECLARE_ALIGNED(16, signed char, uvflim[16]);
- DECLARE_ALIGNED(16, signed char, uvthr[16]);
- DECLARE_ALIGNED(16, signed char, uvmbflim[16]);
- DECLARE_ALIGNED(16, signed char, uvmbthr[16]);
} loop_filter_info;
diff --git a/vp8/common/ppc/loopfilter_altivec.c b/vp8/common/ppc/loopfilter_altivec.c
index bad3cf3..e602fee 100644
--- a/vp8/common/ppc/loopfilter_altivec.c
+++ b/vp8/common/ppc/loopfilter_altivec.c
@@ -56,10 +56,10 @@
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
(void)simpler_lpf;
- mbloop_filter_horizontal_edge_y_ppc(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr);
+ mbloop_filter_horizontal_edge_y_ppc(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr);
if (u_ptr)
- mbloop_filter_horizontal_edge_uv_ppc(u_ptr, v_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr);
+ mbloop_filter_horizontal_edge_uv_ppc(u_ptr, v_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr);
}
void loop_filter_mbhs_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
@@ -77,10 +77,10 @@
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
(void)simpler_lpf;
- mbloop_filter_vertical_edge_y_ppc(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr);
+ mbloop_filter_vertical_edge_y_ppc(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr);
if (u_ptr)
- mbloop_filter_vertical_edge_uv_ppc(u_ptr, v_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr);
+ mbloop_filter_vertical_edge_uv_ppc(u_ptr, v_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr);
}
void loop_filter_mbvs_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
@@ -104,7 +104,7 @@
loop_filter_horizontal_edge_y_ppc(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr);
if (u_ptr)
- loop_filter_horizontal_edge_uv_ppc(u_ptr + 4 * uv_stride, v_ptr + 4 * uv_stride, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr);
+ loop_filter_horizontal_edge_uv_ppc(u_ptr + 4 * uv_stride, v_ptr + 4 * uv_stride, uv_stride, lfi->flim, lfi->lim, lfi->thr);
}
void loop_filter_bhs_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
@@ -127,7 +127,7 @@
loop_filter_vertical_edge_y_ppc(y_ptr, y_stride, lfi->flim, lfi->lim, lfi->thr);
if (u_ptr)
- loop_filter_vertical_edge_uv_ppc(u_ptr + 4, v_ptr + 4, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr);
+ loop_filter_vertical_edge_uv_ppc(u_ptr + 4, v_ptr + 4, uv_stride, lfi->flim, lfi->lim, lfi->thr);
}
void loop_filter_bvs_ppc(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
diff --git a/vp8/common/x86/loopfilter_x86.c b/vp8/common/x86/loopfilter_x86.c
index 93107e1..3621fe1 100644
--- a/vp8/common/x86/loopfilter_x86.c
+++ b/vp8/common/x86/loopfilter_x86.c
@@ -45,13 +45,13 @@
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
(void) simpler_lpf;
- vp8_mbloop_filter_horizontal_edge_mmx(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
+ vp8_mbloop_filter_horizontal_edge_mmx(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
if (u_ptr)
- vp8_mbloop_filter_horizontal_edge_mmx(u_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr, 1);
+ vp8_mbloop_filter_horizontal_edge_mmx(u_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
if (v_ptr)
- vp8_mbloop_filter_horizontal_edge_mmx(v_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr, 1);
+ vp8_mbloop_filter_horizontal_edge_mmx(v_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
}
@@ -62,7 +62,7 @@
(void) v_ptr;
(void) uv_stride;
(void) simpler_lpf;
- vp8_loop_filter_simple_horizontal_edge_mmx(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
+ vp8_loop_filter_simple_horizontal_edge_mmx(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
}
@@ -71,13 +71,13 @@
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
(void) simpler_lpf;
- vp8_mbloop_filter_vertical_edge_mmx(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
+ vp8_mbloop_filter_vertical_edge_mmx(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
if (u_ptr)
- vp8_mbloop_filter_vertical_edge_mmx(u_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr, 1);
+ vp8_mbloop_filter_vertical_edge_mmx(u_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
if (v_ptr)
- vp8_mbloop_filter_vertical_edge_mmx(v_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr, 1);
+ vp8_mbloop_filter_vertical_edge_mmx(v_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, 1);
}
@@ -88,7 +88,7 @@
(void) v_ptr;
(void) uv_stride;
(void) simpler_lpf;
- vp8_loop_filter_simple_vertical_edge_mmx(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
+ vp8_loop_filter_simple_vertical_edge_mmx(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
}
@@ -102,10 +102,10 @@
vp8_loop_filter_horizontal_edge_mmx(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
if (u_ptr)
- vp8_loop_filter_horizontal_edge_mmx(u_ptr + 4 * uv_stride, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr, 1);
+ vp8_loop_filter_horizontal_edge_mmx(u_ptr + 4 * uv_stride, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
if (v_ptr)
- vp8_loop_filter_horizontal_edge_mmx(v_ptr + 4 * uv_stride, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr, 1);
+ vp8_loop_filter_horizontal_edge_mmx(v_ptr + 4 * uv_stride, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
}
@@ -132,10 +132,10 @@
vp8_loop_filter_vertical_edge_mmx(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
if (u_ptr)
- vp8_loop_filter_vertical_edge_mmx(u_ptr + 4, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr, 1);
+ vp8_loop_filter_vertical_edge_mmx(u_ptr + 4, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
if (v_ptr)
- vp8_loop_filter_vertical_edge_mmx(v_ptr + 4, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr, 1);
+ vp8_loop_filter_vertical_edge_mmx(v_ptr + 4, uv_stride, lfi->flim, lfi->lim, lfi->thr, 1);
}
@@ -159,10 +159,10 @@
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
(void) simpler_lpf;
- vp8_mbloop_filter_horizontal_edge_sse2(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
+ vp8_mbloop_filter_horizontal_edge_sse2(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
if (u_ptr)
- vp8_mbloop_filter_horizontal_edge_uv_sse2(u_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr, v_ptr);
+ vp8_mbloop_filter_horizontal_edge_uv_sse2(u_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, v_ptr);
}
@@ -173,7 +173,7 @@
(void) v_ptr;
(void) uv_stride;
(void) simpler_lpf;
- vp8_loop_filter_simple_horizontal_edge_sse2(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
+ vp8_loop_filter_simple_horizontal_edge_sse2(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
}
@@ -182,10 +182,10 @@
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
(void) simpler_lpf;
- vp8_mbloop_filter_vertical_edge_sse2(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
+ vp8_mbloop_filter_vertical_edge_sse2(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
if (u_ptr)
- vp8_mbloop_filter_vertical_edge_uv_sse2(u_ptr, uv_stride, lfi->uvmbflim, lfi->uvlim, lfi->uvmbthr, v_ptr);
+ vp8_mbloop_filter_vertical_edge_uv_sse2(u_ptr, uv_stride, lfi->mbflim, lfi->lim, lfi->thr, v_ptr);
}
@@ -196,7 +196,7 @@
(void) v_ptr;
(void) uv_stride;
(void) simpler_lpf;
- vp8_loop_filter_simple_vertical_edge_sse2(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
+ vp8_loop_filter_simple_vertical_edge_sse2(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->thr, 2);
}
@@ -210,7 +210,7 @@
vp8_loop_filter_horizontal_edge_sse2(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
if (u_ptr)
- vp8_loop_filter_horizontal_edge_uv_sse2(u_ptr + 4 * uv_stride, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr, v_ptr + 4 * uv_stride);
+ vp8_loop_filter_horizontal_edge_uv_sse2(u_ptr + 4 * uv_stride, uv_stride, lfi->flim, lfi->lim, lfi->thr, v_ptr + 4 * uv_stride);
}
@@ -237,7 +237,7 @@
vp8_loop_filter_vertical_edge_sse2(y_ptr + 12, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
if (u_ptr)
- vp8_loop_filter_vertical_edge_uv_sse2(u_ptr + 4, uv_stride, lfi->uvflim, lfi->uvlim, lfi->uvthr, v_ptr + 4);
+ vp8_loop_filter_vertical_edge_uv_sse2(u_ptr + 4, uv_stride, lfi->flim, lfi->lim, lfi->thr, v_ptr + 4);
}
diff --git a/vp8/decoder/arm/arm_dsystemdependent.c b/vp8/decoder/arm/arm_dsystemdependent.c
index e9741e2..02e45d1 100644
--- a/vp8/decoder/arm/arm_dsystemdependent.c
+++ b/vp8/decoder/arm/arm_dsystemdependent.c
@@ -14,7 +14,6 @@
#include "blockd.h"
#include "pragmas.h"
#include "postproc.h"
-#include "dboolhuff.h"
#include "dequantize.h"
#include "onyxd_int.h"
@@ -35,12 +34,6 @@
pbi->dequant.dc_idct_add_y_block = vp8_dequant_dc_idct_add_y_block_v6;
pbi->dequant.idct_add_y_block = vp8_dequant_idct_add_y_block_v6;
pbi->dequant.idct_add_uv_block = vp8_dequant_idct_add_uv_block_v6;
-#if 0 /*For use with RTCD, when implemented*/
- pbi->dboolhuff.start = vp8dx_start_decode_c;
- pbi->dboolhuff.fill = vp8dx_bool_decoder_fill_c;
- pbi->dboolhuff.debool = vp8dx_decode_bool_c;
- pbi->dboolhuff.devalue = vp8dx_decode_value_c;
-#endif
}
#endif
@@ -54,12 +47,6 @@
pbi->dequant.dc_idct_add_y_block = vp8_dequant_dc_idct_add_y_block_neon;
pbi->dequant.idct_add_y_block = vp8_dequant_idct_add_y_block_neon;
pbi->dequant.idct_add_uv_block = vp8_dequant_idct_add_uv_block_neon;
-#if 0 /*For use with RTCD, when implemented*/
- pbi->dboolhuff.start = vp8dx_start_decode_c;
- pbi->dboolhuff.fill = vp8dx_bool_decoder_fill_c;
- pbi->dboolhuff.debool = vp8dx_decode_bool_c;
- pbi->dboolhuff.devalue = vp8dx_decode_value_c;
-#endif
}
#endif
#endif
diff --git a/vp8/decoder/arm/armv6/dboolhuff_v6.asm b/vp8/decoder/arm/armv6/dboolhuff_v6.asm
deleted file mode 100644
index 6515804..0000000
--- a/vp8/decoder/arm/armv6/dboolhuff_v6.asm
+++ /dev/null
@@ -1,163 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_decode_value_v6|
- EXPORT |vp8dx_start_decode_v6|
- EXPORT |vp8dx_stop_decode_v6|
- EXPORT |vp8dx_decode_bool_v6|
-
- ARM
- REQUIRE8
- PRESERVE8
-
- INCLUDE vpx_asm_offsets.asm
-
-br RN r0
-prob RN r1
-bits RN r1
- AREA |.text|, CODE, READONLY ; name this block of code
-
-; int z = 0;
-; int bit;
-; for ( bit=bits-1; bit>=0; bit-- )
-; {
-; z |= (vp8dx_decode_bool(br, 0x80)<<bit);
-; }
-; return z;
-
-;int vp8_decode_value_v6 ( BOOL_DECODER *br, int bits )
-|vp8_decode_value_v6| PROC
- stmdb sp!, {r4 - r6, lr}
- mov r4, br
- mov r5, bits
- mov r6, #0
-
- subs r5, r5, #1
- bmi decode_value_exit
-
-decode_value_loop
- mov prob, #0x80
- mov br, r4
- bl vp8dx_decode_bool_v6_internal ; needed for conversion to s file
- orr r6, r6, r0, lsl r5
- subs r5, r5, #1
- bpl decode_value_loop
-
-decode_value_exit
- mov r0, r6
- ldmia sp!, {r4 - r6, pc}
- ENDP ; |vp8_decode_value_v6|
-
-
-;void vp8dx_start_decode_v6 ( BOOL_DECODER *br, unsigned char *source )
-|vp8dx_start_decode_v6| PROC
- stmdb sp!, {r4 - r5, lr}
- mov r2, #0
- mov r3, #255
-
- str r2, [br, #bool_decoder_lowvalue]
- str r3, [br, #bool_decoder_range]
- str r1, [br, #bool_decoder_buffer]
-
- mov r3, #8
- mov r2, #4
- str r3, [br, #bool_decoder_count]
- str r2, [br, #bool_decoder_pos]
-
- ldrb r2, [r1, #3]
- ldrb r3, [r1, #2]
- ldrb r4, [r1, #1]
- ldrb r5, [r1]
-
- orr r1, r2, r3, lsl #8
- orr r1, r1, r4, lsl #16
- orr r1, r1, r5, lsl #24
-
- str r1, [br, #bool_decoder_value]
-
- ldmia sp!, {r4 - r5, pc}
- ENDP ; |vp8dx_start_decode_v6|
-
-
-;void vp8dx_stop_decode_v6 ( BOOL_DECODER *bc );
-|vp8dx_stop_decode_v6| PROC
- mov pc, lr
- ENDP ; |vp8dx_stop_decode_v6|
-
-
-; bigsplit RN r1
-; buffer_v RN r1
-; count_v RN r4
-; range_v RN r2
-; value_v RN r3
-; pos_v RN r5
-; split RN r6
-; bit RN lr
-;int vp8dx_decode_bool_v6 ( BOOL_DECODER *br, int probability )
-|vp8dx_decode_bool_v6| PROC
-vp8dx_decode_bool_v6_internal
- stmdb sp!, {r4 - r6, lr}
-
- ldr r2, [br, #bool_decoder_range]
- ldr r3, [br, #bool_decoder_value]
-
- mov r6, r2, lsl #8
- sub r6, r6, #256 ; split = 1 + (((range-1) * probability) >> 8)
- mov r12, #1
- smlawb r6, r6, prob, r12
-
- mov lr, #0
- subs r5, r3, r6, lsl #24
-
- ;cmp r3, r1
- movhs lr, #1
- movhs r3, r5
- subhs r2, r2, r6
- movlo r2, r6
-
- cmp r2, #0x80
- blt range_less_0x80
- ;strd r2, r3, [br, #bool_decoder_range]
- str r2, [br, #bool_decoder_range]
- str r3, [br, #bool_decoder_value]
- mov r0, lr
- ldmia sp!, {r4 - r6, pc}
-
-range_less_0x80
- ldr r5, [br, #bool_decoder_pos]
- ldr r1, [br, #bool_decoder_buffer]
- ldr r4, [br, #bool_decoder_count]
- add r1, r1, r5
-
- clz r12, r2
- sub r12, r12, #24
- subs r4, r4, r12
- ldrleb r6, [r1], #1
- mov r2, r2, lsl r12
- mov r3, r3, lsl r12
- addle r4, r4, #8
- rsble r12, r4, #8
- addle r5, r5, #1
- orrle r3, r3, r6, lsl r12
-
- ;strd r2, r3, [br, #bool_decoder_range]
- ;strd r4, r5, [br, #bool_decoder_count]
- str r2, [br, #bool_decoder_range]
- str r3, [br, #bool_decoder_value]
- str r4, [br, #bool_decoder_count]
- str r5, [br, #bool_decoder_pos]
-
- mov r0, lr
-
- ldmia sp!, {r4 - r6, pc}
- ENDP ; |vp8dx_decode_bool_v6|
-
- END
diff --git a/vp8/decoder/arm/dboolhuff_arm.h b/vp8/decoder/arm/dboolhuff_arm.h
deleted file mode 100644
index 985951c..0000000
--- a/vp8/decoder/arm/dboolhuff_arm.h
+++ /dev/null
@@ -1,43 +0,0 @@
-#ifndef DBOOLHUFF_ARM_H
-#define DBOOLHUFF_ARM_H
-
-/* JLK
- * There are currently no arm-optimized versions of
- * these functions. As they are implemented, they
- * can be uncommented below and added to
- * arm/dsystemdependent.c
- *
- * The existing asm code is likely so different as
- * to be useless. However, its been left (for now)
- * for reference.
- */
-#if 0
-#if HAVE_ARMV6
-#undef vp8_dbool_start
-#define vp8_dbool_start vp8dx_start_decode_v6
-
-#undef vp8_dbool_fill
-#define vp8_dbool_fill vp8_bool_decoder_fill_v6
-
-#undef vp8_dbool_debool
-#define vp8_dbool_debool vp8_decode_bool_v6
-
-#undef vp8_dbool_devalue
-#define vp8_dbool_devalue vp8_decode_value_v6
-#endif /* HAVE_ARMV6 */
-
-#if HAVE_ARMV7
-#undef vp8_dbool_start
-#define vp8_dbool_start vp8dx_start_decode_neon
-
-#undef vp8_dbool_fill
-#define vp8_dbool_fill vp8_bool_decoder_fill_neon
-
-#undef vp8_dbool_debool
-#define vp8_dbool_debool vp8_decode_bool_neon
-
-#undef vp8_dbool_devalue
-#define vp8_dbool_devalue vp8_decode_value_neon
-#endif /* HAVE_ARMV7 */
-#endif
-#endif /* DBOOLHUFF_ARM_H */
diff --git a/vp8/decoder/arm/neon/dboolhuff_neon.asm b/vp8/decoder/arm/neon/dboolhuff_neon.asm
deleted file mode 100644
index ff3ffda..0000000
--- a/vp8/decoder/arm/neon/dboolhuff_neon.asm
+++ /dev/null
@@ -1,160 +0,0 @@
-;
-; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
-;
-; Use of this source code is governed by a BSD-style license
-; that can be found in the LICENSE file in the root of the source
-; tree. An additional intellectual property rights grant can be found
-; in the file PATENTS. All contributing project authors may
-; be found in the AUTHORS file in the root of the source tree.
-;
-
-
- EXPORT |vp8_decode_value_neon|
- EXPORT |vp8dx_start_decode_neon|
- EXPORT |vp8dx_stop_decode_neon|
- EXPORT |vp8dx_decode_bool_neon|
-
- ARM
- REQUIRE8
- PRESERVE8
-
- INCLUDE vpx_asm_offsets.asm
-
- AREA |.text|, CODE, READONLY ; name this block of code
-
-; int z = 0;
-; int bit;
-; for ( bit=bits-1; bit>=0; bit-- )
-; {
-; z |= (vp8dx_decode_bool(br, 0x80)<<bit);
-; }
-; return z;
-
-;int vp8_decode_value_neon ( BOOL_DECODER *br, int bits )
-|vp8_decode_value_neon| PROC
- stmdb sp!, {r4 - r6, lr}
- mov r4, r0
- mov r5, r1
- mov r6, #0
-
- subs r5, r5, #1
- bmi decode_value_exit
-
-decode_value_loop
- mov r1, #0x80
- mov r0, r4
- bl vp8dx_decode_bool_neon_internal ; needed for conversion to s file
- orr r6, r6, r0, lsl r5
- subs r5, r5, #1
- bpl decode_value_loop
-
-decode_value_exit
- mov r0, r6
- ldmia sp!, {r4 - r6, pc}
- ENDP ; |vp8_decode_value_neon|
-
-
-;void vp8dx_start_decode_neon ( BOOL_DECODER *br, unsigned char *source )
-|vp8dx_start_decode_neon| PROC
- stmdb sp!, {r4 - r5, lr}
- mov r2, #0
- mov r3, #255
-
- str r2, [r0, #bool_decoder_lowvalue]
- str r3, [r0, #bool_decoder_range]
- str r1, [r0, #bool_decoder_buffer]
-
- mov r3, #8
- mov r2, #4
- str r3, [r0, #bool_decoder_count]
- str r2, [r0, #bool_decoder_pos]
-
- ldrb r2, [r1, #3]
- ldrb r3, [r1, #2]
- ldrb r4, [r1, #1]
- ldrb r5, [r1]
-
- orr r1, r2, r3, lsl #8
- orr r1, r1, r4, lsl #16
- orr r1, r1, r5, lsl #24
-
- str r1, [r0, #bool_decoder_value]
-
- ldmia sp!, {r4 - r5, pc}
- ENDP ; |vp8dx_start_decode_neon|
-
-
-;void vp8dx_stop_decode_neon ( BOOL_DECODER *bc );
-|vp8dx_stop_decode_neon| PROC
- mov pc, lr
- ENDP ; |vp8dx_stop_decode_neon|
-
-
-; bigsplit RN r1
-; buffer_v RN r1
-; count_v RN r4
-; range_v RN r2
-; value_v RN r3
-; pos_v RN r5
-; split RN r6
-; bit RN lr
-;int vp8dx_decode_bool_neon ( BOOL_DECODER *br, int probability )
-|vp8dx_decode_bool_neon| PROC
-vp8dx_decode_bool_neon_internal
-;LDRD and STRD doubleword data transfers must be eight-byte aligned. Use ALIGN 8
-;before memory allocation
- stmdb sp!, {r4 - r5, lr}
-
- ldr r2, [r0, #bool_decoder_range] ;load range (r2), value(r3)
- ldr r3, [r0, #bool_decoder_value]
- ;ldrd r2, r3, [r0, #bool_decoder_range] ;ldrd costs 2 cycles
- ;
-
- mov r4, r2, lsl #8
- sub r4, r4, #256
- mov r12, #1
-
- smlawb r4, r4, r1, r12 ;split = 1 + (((range-1) * probability) >> 8)
-
- mov lr, r0
- mov r0, #0 ;bit = 0
- ;
- subs r5, r3, r4, lsl #24
-
- subhs r2, r2, r4 ;range = br->range-split
- movlo r2, r4 ;range = split
- movhs r0, #1 ;bit = 1
- movhs r3, r5 ;value = value-bigsplit
-
- cmp r2, #0x80
- blt range_less_0x80
- strd r2, r3, [lr, #bool_decoder_range] ;store result
-
- ldmia sp!, {r4 - r5, pc}
-
-range_less_0x80
-
- ldrd r4, r5, [lr, #bool_decoder_count] ;load count, pos, buffer
- ldr r1, [lr, #bool_decoder_buffer]
-
- clz r12, r2
- add r1, r1, r5
-
- sub r12, r12, #24
- subs r4, r4, r12 ;count -= shift
- mov r2, r2, lsl r12 ;range <<= shift
- mov r3, r3, lsl r12 ;value <<= shift
- addle r4, r4, #8 ;count += 8
- ldrleb r12, [r1], #1 ;br->buffer[br->pos]
-
- rsble r1, r4, #8 ;-count
- addle r5, r5, #1 ;br->pos++
- orrle r3, r3, r12, lsl r1 ;value |= (br->buffer[br->pos]) << (-count)
-
- strd r2, r3, [lr, #bool_decoder_range] ;store result
- strd r4, r5, [lr, #bool_decoder_count]
-
- ldmia sp!, {r4 - r5, pc}
- ENDP ; |vp8dx_decode_bool_neon|
-
- END
diff --git a/vp8/decoder/dboolhuff.c b/vp8/decoder/dboolhuff.c
index 57cba16..8527d51 100644
--- a/vp8/decoder/dboolhuff.c
+++ b/vp8/decoder/dboolhuff.c
@@ -26,8 +26,9 @@
};
-int vp8dx_start_decode_c(BOOL_DECODER *br, const unsigned char *source,
- unsigned int source_sz)
+int vp8dx_start_decode(BOOL_DECODER *br,
+ const unsigned char *source,
+ unsigned int source_sz)
{
br->user_buffer_end = source+source_sz;
br->user_buffer = source;
@@ -39,13 +40,13 @@
return 1;
/* Populate the buffer */
- vp8dx_bool_decoder_fill_c(br);
+ vp8dx_bool_decoder_fill(br);
return 0;
}
-void vp8dx_bool_decoder_fill_c(BOOL_DECODER *br)
+void vp8dx_bool_decoder_fill(BOOL_DECODER *br)
{
const unsigned char *bufptr;
const unsigned char *bufend;
@@ -62,69 +63,3 @@
br->value = value;
br->count = count;
}
-
-#if 0
-/*
- * Until optimized versions of these functions are available, we
- * keep the implementation in the header to allow inlining.
- *
- * The RTCD-style invocations are still in place so this can
- * be switched by just uncommenting these functions here and
- * the DBOOLHUFF_INVOKE calls in the header.
- */
-int vp8dx_decode_bool_c(BOOL_DECODER *br, int probability)
-{
- unsigned int bit=0;
- VP8_BD_VALUE value;
- unsigned int split;
- VP8_BD_VALUE bigsplit;
- int count;
- unsigned int range;
-
- value = br->value;
- count = br->count;
- range = br->range;
-
- split = 1 + (((range-1) * probability) >> 8);
- bigsplit = (VP8_BD_VALUE)split << (VP8_BD_VALUE_SIZE - 8);
-
- range = split;
- if(value >= bigsplit)
- {
- range = br->range-split;
- value = value-bigsplit;
- bit = 1;
- }
-
- /*if(range>=0x80)
- {
- br->value = value;
- br->range = range;
- return bit;
- }*/
-
- {
- register unsigned int shift = vp8dx_bitreader_norm[range];
- range <<= shift;
- value <<= shift;
- count -= shift;
- }
- br->value = value;
- br->count = count;
- br->range = range;
- if (count < 0)
- vp8dx_bool_decoder_fill_c(br);
- return bit;
-}
-
-int vp8dx_decode_value_c(BOOL_DECODER *br, int bits)
-{
- int z = 0;
- int bit;
- for ( bit=bits-1; bit>=0; bit-- )
- {
- z |= (vp8dx_decode_bool(br, 0x80)<<bit);
- }
- return z;
-}
-#endif
diff --git a/vp8/decoder/dboolhuff.h b/vp8/decoder/dboolhuff.h
index d14f4dc..a83e3f0 100644
--- a/vp8/decoder/dboolhuff.h
+++ b/vp8/decoder/dboolhuff.h
@@ -25,10 +25,6 @@
Even relatively modest values like 100 would work fine.*/
# define VP8_LOTS_OF_BITS (0x40000000)
-
-
-struct vp8_dboolhuff_rtcd_vtable;
-
typedef struct
{
const unsigned char *user_buffer_end;
@@ -36,82 +32,15 @@
VP8_BD_VALUE value;
int count;
unsigned int range;
-#if CONFIG_RUNTIME_CPU_DETECT
- struct vp8_dboolhuff_rtcd_vtable *rtcd;
-#endif
} BOOL_DECODER;
-#define prototype_dbool_start(sym) int sym(BOOL_DECODER *br, \
- const unsigned char *source, unsigned int source_sz)
-#define prototype_dbool_fill(sym) void sym(BOOL_DECODER *br)
-#define prototype_dbool_debool(sym) int sym(BOOL_DECODER *br, int probability)
-#define prototype_dbool_devalue(sym) int sym(BOOL_DECODER *br, int bits)
-
-#if ARCH_ARM
-#include "arm/dboolhuff_arm.h"
-#endif
-
-#ifndef vp8_dbool_start
-#define vp8_dbool_start vp8dx_start_decode_c
-#endif
-
-#ifndef vp8_dbool_fill
-#define vp8_dbool_fill vp8dx_bool_decoder_fill_c
-#endif
-
-#ifndef vp8_dbool_debool
-#define vp8_dbool_debool vp8dx_decode_bool_c
-#endif
-
-#ifndef vp8_dbool_devalue
-#define vp8_dbool_devalue vp8dx_decode_value_c
-#endif
-
-extern prototype_dbool_start(vp8_dbool_start);
-extern prototype_dbool_fill(vp8_dbool_fill);
-extern prototype_dbool_debool(vp8_dbool_debool);
-extern prototype_dbool_devalue(vp8_dbool_devalue);
-
-typedef prototype_dbool_start((*vp8_dbool_start_fn_t));
-typedef prototype_dbool_fill((*vp8_dbool_fill_fn_t));
-typedef prototype_dbool_debool((*vp8_dbool_debool_fn_t));
-typedef prototype_dbool_devalue((*vp8_dbool_devalue_fn_t));
-
-typedef struct vp8_dboolhuff_rtcd_vtable {
- vp8_dbool_start_fn_t start;
- vp8_dbool_fill_fn_t fill;
- vp8_dbool_debool_fn_t debool;
- vp8_dbool_devalue_fn_t devalue;
-} vp8_dboolhuff_rtcd_vtable_t;
-
-/* There are no processor-specific versions of these
- * functions right now. Disable RTCD to avoid using
- * function pointers which gives a speed boost
- */
-/*#ifdef ENABLE_RUNTIME_CPU_DETECT
-#define DBOOLHUFF_INVOKE(ctx,fn) (ctx)->fn
-#define IF_RTCD(x) (x)
-#else*/
-#define DBOOLHUFF_INVOKE(ctx,fn) vp8_dbool_##fn
-#define IF_RTCD(x) NULL
-/*#endif*/
-
DECLARE_ALIGNED(16, extern const unsigned char, vp8dx_bitreader_norm[256]);
-/* wrapper functions to hide RTCD. static means inline means hopefully no
- * penalty
- */
-static int vp8dx_start_decode(BOOL_DECODER *br,
- struct vp8_dboolhuff_rtcd_vtable *rtcd,
- const unsigned char *source, unsigned int source_sz) {
-#if CONFIG_RUNTIME_CPU_DETECT
- br->rtcd = rtcd;
-#endif
- return DBOOLHUFF_INVOKE(rtcd, start)(br, source, source_sz);
-}
-static void vp8dx_bool_decoder_fill(BOOL_DECODER *br) {
- DBOOLHUFF_INVOKE(br->rtcd, fill)(br);
-}
+int vp8dx_start_decode(BOOL_DECODER *br,
+ const unsigned char *source,
+ unsigned int source_sz);
+
+void vp8dx_bool_decoder_fill(BOOL_DECODER *br);
/*The refill loop is used in several places, so define it in a macro to make
sure they're all consistent.
@@ -138,12 +67,6 @@
static int vp8dx_decode_bool(BOOL_DECODER *br, int probability) {
- /*
- * Until optimized versions of this function are available, we
- * keep the implementation in the header to allow inlining.
- *
- *return DBOOLHUFF_INVOKE(br->rtcd, debool)(br, probability);
- */
unsigned int bit = 0;
VP8_BD_VALUE value;
unsigned int split;
@@ -167,13 +90,6 @@
bit = 1;
}
- /*if(range>=0x80)
- {
- br->value = value;
- br->range = range;
- return bit
- }*/
-
{
register unsigned int shift = vp8dx_bitreader_norm[range];
range <<= shift;
@@ -190,12 +106,6 @@
static int vp8_decode_value(BOOL_DECODER *br, int bits)
{
- /*
- * Until optimized versions of this function are available, we
- * keep the implementation in the header to allow inlining.
- *
- *return DBOOLHUFF_INVOKE(br->rtcd, devalue)(br, bits);
- */
int z = 0;
int bit;
diff --git a/vp8/decoder/decodframe.c b/vp8/decoder/decodframe.c
index d3972b3..bf79b31 100644
--- a/vp8/decoder/decodframe.c
+++ b/vp8/decoder/decodframe.c
@@ -475,8 +475,7 @@
"Truncated packet or corrupt partition "
"%d length", i + 1);
- if (vp8dx_start_decode(bool_decoder, IF_RTCD(&pbi->dboolhuff),
- partition, partition_size))
+ if (vp8dx_start_decode(bool_decoder, partition, partition_size))
vpx_internal_error(&pc->error, VPX_CODEC_MEM_ERROR,
"Failed to allocate bool decoder %d", i + 1);
@@ -651,8 +650,7 @@
init_frame(pbi);
- if (vp8dx_start_decode(bc, IF_RTCD(&pbi->dboolhuff),
- data, data_end - data))
+ if (vp8dx_start_decode(bc, data, data_end - data))
vpx_internal_error(&pc->error, VPX_CODEC_MEM_ERROR,
"Failed to allocate bool decoder 0");
if (pc->frame_type == KEY_FRAME) {
diff --git a/vp8/decoder/generic/dsystemdependent.c b/vp8/decoder/generic/dsystemdependent.c
index 2e28472..709ec59 100644
--- a/vp8/decoder/generic/dsystemdependent.c
+++ b/vp8/decoder/generic/dsystemdependent.c
@@ -27,12 +27,6 @@
pbi->dequant.dc_idct_add_y_block = vp8_dequant_dc_idct_add_y_block_c;
pbi->dequant.idct_add_y_block = vp8_dequant_idct_add_y_block_c;
pbi->dequant.idct_add_uv_block = vp8_dequant_idct_add_uv_block_c;
- pbi->dboolhuff.start = vp8dx_start_decode_c;
- pbi->dboolhuff.fill = vp8dx_bool_decoder_fill_c;
-#if 0 /*For use with RTCD, when implemented*/
- pbi->dboolhuff.debool = vp8dx_decode_bool_c;
- pbi->dboolhuff.devalue = vp8dx_decode_value_c;
-#endif
#endif
#if ARCH_X86 || ARCH_X86_64
diff --git a/vp8/decoder/onyxd_int.h b/vp8/decoder/onyxd_int.h
index 7593edf..2f0fe70 100644
--- a/vp8/decoder/onyxd_int.h
+++ b/vp8/decoder/onyxd_int.h
@@ -125,7 +125,6 @@
#if CONFIG_RUNTIME_CPU_DETECT
vp8_dequant_rtcd_vtable_t dequant;
- struct vp8_dboolhuff_rtcd_vtable dboolhuff;
#endif
diff --git a/vp8/encoder/encodeframe.c b/vp8/encoder/encodeframe.c
index d535894..f06a4ca 100644
--- a/vp8/encoder/encodeframe.c
+++ b/vp8/encoder/encodeframe.c
@@ -399,13 +399,6 @@
// Clear Zbin mode boost for default case
cpi->zbin_mode_boost = 0;
- // vp8cx_init_quantizer() is first called in vp8_create_compressor(). A check is added here so that vp8cx_init_quantizer() is only called
- // when these values are not all zero.
- if (cpi->common.y1dc_delta_q | cpi->common.y2dc_delta_q | cpi->common.uvdc_delta_q | cpi->common.y2ac_delta_q | cpi->common.uvac_delta_q)
- {
- vp8cx_init_quantizer(cpi);
- }
-
// MB level quantizer setup
vp8cx_mb_init_quantizer(cpi, &cpi->mb);
}
diff --git a/vp8/encoder/onyx_if.c b/vp8/encoder/onyx_if.c
index 6a3f7f6..912d641 100644
--- a/vp8/encoder/onyx_if.c
+++ b/vp8/encoder/onyx_if.c
@@ -3123,11 +3123,14 @@
return 1;
}
+
static void set_quantizer(VP8_COMP *cpi, int Q)
{
VP8_COMMON *cm = &cpi->common;
MACROBLOCKD *mbd = &cpi->mb.e_mbd;
+ int update = 0;
+ update |= cm->base_qindex != Q;
cm->base_qindex = Q;
cm->y1dc_delta_q = 0;
@@ -3138,13 +3141,19 @@
if(Q<4)
{
+ update |= cm->y2dc_delta_q != 4-Q;
cm->y2dc_delta_q = 4-Q;
}
+
// Set Segment specific quatizers
mbd->segment_feature_data[MB_LVL_ALT_Q][0] = cpi->segment_feature_data[MB_LVL_ALT_Q][0];
mbd->segment_feature_data[MB_LVL_ALT_Q][1] = cpi->segment_feature_data[MB_LVL_ALT_Q][1];
mbd->segment_feature_data[MB_LVL_ALT_Q][2] = cpi->segment_feature_data[MB_LVL_ALT_Q][2];
mbd->segment_feature_data[MB_LVL_ALT_Q][3] = cpi->segment_feature_data[MB_LVL_ALT_Q][3];
+
+ if(update)
+ vp8cx_init_quantizer(cpi);
+
}
static void update_alt_ref_frame_and_stats(VP8_COMP *cpi)
@@ -3614,6 +3623,17 @@
// Test code for segmentation of gf/arf (0,0)
//segmentation_test_function((VP8_PTR) cpi);
+#if CONFIG_REALTIME_ONLY
+ if(cpi->oxcf.auto_key && cm->frame_type != KEY_FRAME)
+ {
+ if(cpi->force_next_frame_intra)
+ {
+ cm->frame_type = KEY_FRAME; /* delayed intra frame */
+ }
+ }
+ cpi->force_next_frame_intra = 0;
+#endif
+
// For an alt ref frame in 2 pass we skip the call to the second pass function that sets the target bandwidth
#if !(CONFIG_REALTIME_ONLY)
@@ -4124,6 +4144,14 @@
// (assuming that we didn't)!
if (cpi->pass != 2 && cpi->oxcf.auto_key && cm->frame_type != KEY_FRAME)
{
+
+#if CONFIG_REALTIME_ONLY
+ {
+ /* we don't do re-encoding in realtime mode
+ * if key frame is decided than we force it on next frame */
+ cpi->force_next_frame_intra = decide_key_frame(cpi);
+ }
+#else
if (decide_key_frame(cpi))
{
vp8_calc_auto_iframe_target_size(cpi);
@@ -4162,6 +4190,7 @@
resize_key_frame(cpi);
continue;
}
+#endif
}
vp8_clear_system_state();
diff --git a/vp8/encoder/onyx_int.h b/vp8/encoder/onyx_int.h
index 4d8f3ec..e8a452d 100644
--- a/vp8/encoder/onyx_int.h
+++ b/vp8/encoder/onyx_int.h
@@ -677,6 +677,9 @@
int *lf_ref_frame_sign_bias;
int *lf_ref_frame;
+#if CONFIG_REALTIME_ONLY
+ int force_next_frame_intra; /* force next frame to intra when kf_auto says so */
+#endif
} VP8_COMP;
void control_data_rate(VP8_COMP *cpi);
diff --git a/vp8/encoder/pickinter.c b/vp8/encoder/pickinter.c
index cfaf497..287ca61 100644
--- a/vp8/encoder/pickinter.c
+++ b/vp8/encoder/pickinter.c
@@ -24,7 +24,7 @@
#include "g_common.h"
#include "variance.h"
#include "mcomp.h"
-
+#include "rdopt.h"
#include "vpx_mem/vpx_mem.h"
#if CONFIG_RUNTIME_CPU_DETECT
@@ -412,7 +412,6 @@
}
-
int vp8_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int recon_uvoffset, int *returnrate, int *returndistortion, int *returnintra)
{
BLOCK *b = &x->block[0];
@@ -421,7 +420,7 @@
B_MODE_INFO best_bmodes[16];
MB_MODE_INFO best_mbmode;
PARTITION_INFO best_partition;
- MV best_ref_mv1;
+ MV best_ref_mv;
MV mode_mv[MB_MODE_COUNT];
MB_PREDICTION_MODE this_mode;
int num00;
@@ -439,9 +438,14 @@
int best_mode_index = 0;
int sse = INT_MAX;
+ MV mvp;
+ int near_sadidx[8] = {0, 1, 2, 3, 4, 5, 6, 7};
+ int saddone=0;
+ int sr=0; //search range got from mv_pred(). It uses step_param levels. (0-7)
+
MV nearest_mv[4];
MV near_mv[4];
- MV best_ref_mv[4];
+ MV frame_best_ref_mv[4];
int MDCounts[4][4];
unsigned char *y_buffer[4];
unsigned char *u_buffer[4];
@@ -461,7 +465,7 @@
YV12_BUFFER_CONFIG *lst_yv12 = &cpi->common.yv12_fb[cpi->common.lst_fb_idx];
vp8_find_near_mvs(&x->e_mbd, x->e_mbd.mode_info_context, &nearest_mv[LAST_FRAME], &near_mv[LAST_FRAME],
- &best_ref_mv[LAST_FRAME], MDCounts[LAST_FRAME], LAST_FRAME, cpi->common.ref_frame_sign_bias);
+ &frame_best_ref_mv[LAST_FRAME], MDCounts[LAST_FRAME], LAST_FRAME, cpi->common.ref_frame_sign_bias);
y_buffer[LAST_FRAME] = lst_yv12->y_buffer + recon_yoffset;
u_buffer[LAST_FRAME] = lst_yv12->u_buffer + recon_uvoffset;
@@ -475,7 +479,7 @@
YV12_BUFFER_CONFIG *gld_yv12 = &cpi->common.yv12_fb[cpi->common.gld_fb_idx];
vp8_find_near_mvs(&x->e_mbd, x->e_mbd.mode_info_context, &nearest_mv[GOLDEN_FRAME], &near_mv[GOLDEN_FRAME],
- &best_ref_mv[GOLDEN_FRAME], MDCounts[GOLDEN_FRAME], GOLDEN_FRAME, cpi->common.ref_frame_sign_bias);
+ &frame_best_ref_mv[GOLDEN_FRAME], MDCounts[GOLDEN_FRAME], GOLDEN_FRAME, cpi->common.ref_frame_sign_bias);
y_buffer[GOLDEN_FRAME] = gld_yv12->y_buffer + recon_yoffset;
u_buffer[GOLDEN_FRAME] = gld_yv12->u_buffer + recon_uvoffset;
@@ -489,7 +493,7 @@
YV12_BUFFER_CONFIG *alt_yv12 = &cpi->common.yv12_fb[cpi->common.alt_fb_idx];
vp8_find_near_mvs(&x->e_mbd, x->e_mbd.mode_info_context, &nearest_mv[ALTREF_FRAME], &near_mv[ALTREF_FRAME],
- &best_ref_mv[ALTREF_FRAME], MDCounts[ALTREF_FRAME], ALTREF_FRAME, cpi->common.ref_frame_sign_bias);
+ &frame_best_ref_mv[ALTREF_FRAME], MDCounts[ALTREF_FRAME], ALTREF_FRAME, cpi->common.ref_frame_sign_bias);
y_buffer[ALTREF_FRAME] = alt_yv12->y_buffer + recon_yoffset;
u_buffer[ALTREF_FRAME] = alt_yv12->u_buffer + recon_uvoffset;
@@ -529,10 +533,6 @@
+ vp8_cost_one(cpi->prob_gf_coded);
}
-
-
- best_rd = INT_MAX;
-
x->e_mbd.mode_info_context->mbmi.ref_frame = INTRA_FRAME;
// if we encode a new mv this is important
@@ -595,7 +595,7 @@
x->e_mbd.pre.v_buffer = v_buffer[x->e_mbd.mode_info_context->mbmi.ref_frame];
mode_mv[NEARESTMV] = nearest_mv[x->e_mbd.mode_info_context->mbmi.ref_frame];
mode_mv[NEARMV] = near_mv[x->e_mbd.mode_info_context->mbmi.ref_frame];
- best_ref_mv1 = best_ref_mv[x->e_mbd.mode_info_context->mbmi.ref_frame];
+ best_ref_mv = frame_best_ref_mv[x->e_mbd.mode_info_context->mbmi.ref_frame];
memcpy(mdcounts, MDCounts[x->e_mbd.mode_info_context->mbmi.ref_frame], sizeof(mdcounts));
}
@@ -608,6 +608,28 @@
continue;
}
+ if(x->e_mbd.mode_info_context->mbmi.mode == NEWMV)
+ {
+ if(!saddone)
+ {
+ vp8_cal_sad(cpi,xd,x, recon_yoffset ,&near_sadidx[0] );
+ saddone = 1;
+ }
+
+ vp8_mv_pred(cpi, &x->e_mbd, x->e_mbd.mode_info_context, &mvp,
+ x->e_mbd.mode_info_context->mbmi.ref_frame, cpi->common.ref_frame_sign_bias, &sr, &near_sadidx[0]);
+
+ /* adjust mvp to make sure it is within MV range */
+ if(mvp.row > best_ref_mv.row + MAX_FULL_PEL_VAL)
+ mvp.row = best_ref_mv.row + MAX_FULL_PEL_VAL;
+ else if(mvp.row < best_ref_mv.row - MAX_FULL_PEL_VAL)
+ mvp.row = best_ref_mv.row - MAX_FULL_PEL_VAL;
+ if(mvp.col > best_ref_mv.col + MAX_FULL_PEL_VAL)
+ mvp.col = best_ref_mv.col + MAX_FULL_PEL_VAL;
+ else if(mvp.col < best_ref_mv.col - MAX_FULL_PEL_VAL)
+ mvp.col = best_ref_mv.col - MAX_FULL_PEL_VAL;
+ }
+
switch (this_mode)
{
case B_PRED:
@@ -663,61 +685,46 @@
int n = 0;
int sadpb = x->sadperbit16;
+ int col_min = (best_ref_mv.col - MAX_FULL_PEL_VAL) >>3;
+ int col_max = (best_ref_mv.col + MAX_FULL_PEL_VAL) >>3;
+ int row_min = (best_ref_mv.row - MAX_FULL_PEL_VAL) >>3;
+ int row_max = (best_ref_mv.row + MAX_FULL_PEL_VAL) >>3;
+
+ int tmp_col_min = x->mv_col_min;
+ int tmp_col_max = x->mv_col_max;
+ int tmp_row_min = x->mv_row_min;
+ int tmp_row_max = x->mv_row_max;
+
+ // Get intersection of UMV window and valid MV window to reduce # of checks in diamond search.
+ if (x->mv_col_min < col_min )
+ x->mv_col_min = col_min;
+ if (x->mv_col_max > col_max )
+ x->mv_col_max = col_max;
+ if (x->mv_row_min < row_min )
+ x->mv_row_min = row_min;
+ if (x->mv_row_max > row_max )
+ x->mv_row_max = row_max;
+
// Further step/diamond searches as necessary
- if (cpi->Speed < 8)
{
- step_param = cpi->sf.first_step + ((cpi->Speed > 5) ? 1 : 0);
- further_steps = (cpi->sf.max_step_search_steps - 1) - step_param;
+ int speed_adjust = (cpi->Speed > 5) ? ((cpi->Speed >= 8)? 3 : 2) : 1;
+ step_param = cpi->sf.first_step + speed_adjust;
+ sr += speed_adjust;
+ //adjust search range according to sr from mv prediction
+ if(sr > step_param)
+ step_param = sr;
+ further_steps = (cpi->Speed >= 8)? 0: (cpi->sf.max_step_search_steps - 1 - step_param);
}
- else
- {
- step_param = cpi->sf.first_step + 2;
- further_steps = 0;
- }
-
-#if 0
-
- // Initial step Search
- bestsme = vp8_diamond_search_sad(x, b, d, &best_ref_mv1, &d->bmi.mv.as_mv, step_param, x->errorperbit, &num00, &cpi->fn_ptr, cpi->mb.mvsadcost, cpi->mb.mvcost, &best_ref_mv1);
- mode_mv[NEWMV].row = d->bmi.mv.as_mv.row;
- mode_mv[NEWMV].col = d->bmi.mv.as_mv.col;
-
- // Further step searches
- while (n < further_steps)
- {
- n++;
-
- if (num00)
- num00--;
- else
- {
- thissme = vp8_diamond_search_sad(x, b, d, &best_ref_mv1, &d->bmi.mv.as_mv, step_param + n, x->errorperbit, &num00, &cpi->fn_ptr, cpi->mb.mvsadcost, x->mvcost, &best_ref_mv1);
-
- if (thissme < bestsme)
- {
- bestsme = thissme;
- mode_mv[NEWMV].row = d->bmi.mv.as_mv.row;
- mode_mv[NEWMV].col = d->bmi.mv.as_mv.col;
- }
- else
- {
- d->bmi.mv.as_mv.row = mode_mv[NEWMV].row;
- d->bmi.mv.as_mv.col = mode_mv[NEWMV].col;
- }
- }
- }
-
-#else
if (cpi->sf.search_method == HEX)
{
- bestsme = vp8_hex_search(x, b, d, &best_ref_mv1, &d->bmi.mv.as_mv, step_param, sadpb/*x->errorperbit*/, &num00, &cpi->fn_ptr[BLOCK_16X16], x->mvsadcost, x->mvcost);
+ bestsme = vp8_hex_search(x, b, d, &best_ref_mv, &d->bmi.mv.as_mv, step_param, sadpb/*x->errorperbit*/, &num00, &cpi->fn_ptr[BLOCK_16X16], x->mvsadcost, x->mvcost);
mode_mv[NEWMV].row = d->bmi.mv.as_mv.row;
mode_mv[NEWMV].col = d->bmi.mv.as_mv.col;
}
else
{
- bestsme = cpi->diamond_search_sad(x, b, d, &best_ref_mv1, &d->bmi.mv.as_mv, step_param, sadpb / 2/*x->errorperbit*/, &num00, &cpi->fn_ptr[BLOCK_16X16], x->mvsadcost, x->mvcost, &best_ref_mv1); //sadpb < 9
+ bestsme = cpi->diamond_search_sad(x, b, d, &mvp, &d->bmi.mv.as_mv, step_param, sadpb / 2/*x->errorperbit*/, &num00, &cpi->fn_ptr[BLOCK_16X16], x->mvsadcost, x->mvcost, &best_ref_mv); //sadpb < 9
mode_mv[NEWMV].row = d->bmi.mv.as_mv.row;
mode_mv[NEWMV].col = d->bmi.mv.as_mv.col;
@@ -736,7 +743,7 @@
num00--;
else
{
- thissme = cpi->diamond_search_sad(x, b, d, &best_ref_mv1, &d->bmi.mv.as_mv, step_param + n, sadpb / 4/*x->errorperbit*/, &num00, &cpi->fn_ptr[BLOCK_16X16], x->mvsadcost, x->mvcost, &best_ref_mv1); //sadpb = 9
+ thissme = cpi->diamond_search_sad(x, b, d, &mvp, &d->bmi.mv.as_mv, step_param + n, sadpb / 4/*x->errorperbit*/, &num00, &cpi->fn_ptr[BLOCK_16X16], x->mvsadcost, x->mvcost, &best_ref_mv); //sadpb = 9
if (thissme < bestsme)
{
@@ -753,19 +760,21 @@
}
}
-#endif
+ x->mv_col_min = tmp_col_min;
+ x->mv_col_max = tmp_col_max;
+ x->mv_row_min = tmp_row_min;
+ x->mv_row_max = tmp_row_max;
+
+ if (bestsme < INT_MAX)
+ cpi->find_fractional_mv_step(x, b, d, &d->bmi.mv.as_mv, &best_ref_mv, x->errorperbit, &cpi->fn_ptr[BLOCK_16X16], cpi->mb.mvcost);
+
+ mode_mv[NEWMV].row = d->bmi.mv.as_mv.row;
+ mode_mv[NEWMV].col = d->bmi.mv.as_mv.col;
+
+ // mv cost;
+ rate2 += vp8_mv_bit_cost(&mode_mv[NEWMV], &best_ref_mv, cpi->mb.mvcost, 128);
}
- if (bestsme < INT_MAX)
- cpi->find_fractional_mv_step(x, b, d, &d->bmi.mv.as_mv, &best_ref_mv1, x->errorperbit, &cpi->fn_ptr[BLOCK_16X16], cpi->mb.mvcost);
-
- mode_mv[NEWMV].row = d->bmi.mv.as_mv.row;
- mode_mv[NEWMV].col = d->bmi.mv.as_mv.col;
-
- // mv cost;
- rate2 += vp8_mv_bit_cost(&mode_mv[NEWMV], &best_ref_mv1, cpi->mb.mvcost, 128);
-
-
case NEARESTMV:
case NEARMV:
diff --git a/vp8/encoder/rdopt.c b/vp8/encoder/rdopt.c
index 248d9c4..ea21cf4 100644
--- a/vp8/encoder/rdopt.c
+++ b/vp8/encoder/rdopt.c
@@ -1419,6 +1419,7 @@
return bsi.segment_rd;
}
+#endif
static void swap(int *x,int *y)
{
@@ -1651,6 +1652,62 @@
vp8_clamp_mv(mvp, xd);
}
+void vp8_cal_sad(VP8_COMP *cpi, MACROBLOCKD *xd, MACROBLOCK *x, int recon_yoffset, int near_sadidx[])
+{
+
+ int near_sad[8] = {0}; // 0-cf above, 1-cf left, 2-cf aboveleft, 3-lf current, 4-lf above, 5-lf left, 6-lf right, 7-lf below
+
+ //calculate sad for current frame 3 nearby MBs.
+ if( xd->mb_to_top_edge==0 && xd->mb_to_left_edge ==0)
+ {
+ near_sad[0] = near_sad[1] = near_sad[2] = INT_MAX;
+ }else if(xd->mb_to_top_edge==0)
+ { //only has left MB for sad calculation.
+ near_sad[0] = near_sad[2] = INT_MAX;
+ near_sad[1] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - 16,xd->dst.y_stride, 0x7fffffff);
+ }else if(xd->mb_to_left_edge ==0)
+ { //only has left MB for sad calculation.
+ near_sad[1] = near_sad[2] = INT_MAX;
+ near_sad[0] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - xd->dst.y_stride *16,xd->dst.y_stride, 0x7fffffff);
+ }else
+ {
+ near_sad[0] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - xd->dst.y_stride *16,xd->dst.y_stride, 0x7fffffff);
+ near_sad[1] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - 16,xd->dst.y_stride, 0x7fffffff);
+ near_sad[2] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - xd->dst.y_stride *16 -16,xd->dst.y_stride, 0x7fffffff);
+ }
+
+ if(cpi->common.last_frame_type != KEY_FRAME)
+ {
+ //calculate sad for last frame 5 nearby MBs.
+ unsigned char *pre_y_buffer = cpi->common.yv12_fb[cpi->common.lst_fb_idx].y_buffer + recon_yoffset;
+ int pre_y_stride = cpi->common.yv12_fb[cpi->common.lst_fb_idx].y_stride;
+
+ if(xd->mb_to_top_edge==0) near_sad[4] = INT_MAX;
+ if(xd->mb_to_left_edge ==0) near_sad[5] = INT_MAX;
+ if(xd->mb_to_right_edge ==0) near_sad[6] = INT_MAX;
+ if(xd->mb_to_bottom_edge==0) near_sad[7] = INT_MAX;
+
+ if(near_sad[4] != INT_MAX)
+ near_sad[4] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer - pre_y_stride *16, pre_y_stride, 0x7fffffff);
+ if(near_sad[5] != INT_MAX)
+ near_sad[5] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer - 16, pre_y_stride, 0x7fffffff);
+ near_sad[3] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer, pre_y_stride, 0x7fffffff);
+ if(near_sad[6] != INT_MAX)
+ near_sad[6] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer + 16, pre_y_stride, 0x7fffffff);
+ if(near_sad[7] != INT_MAX)
+ near_sad[7] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer + pre_y_stride *16, pre_y_stride, 0x7fffffff);
+ }
+
+ if(cpi->common.last_frame_type != KEY_FRAME)
+ {
+ quicksortsad(near_sad, near_sadidx, 0, 7);
+ }else
+ {
+ quicksortsad(near_sad, near_sadidx, 0, 2);
+ }
+}
+
+#if !(CONFIG_REALTIME_ONLY)
int vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int recon_uvoffset, int *returnrate, int *returndistortion, int *returnintra)
{
BLOCK *b = &x->block[0];
@@ -1688,7 +1745,6 @@
int force_no_skip = 0;
MV mvp;
- int near_sad[8] = {0}; // 0-cf above, 1-cf left, 2-cf aboveleft, 3-lf current, 4-lf above, 5-lf left, 6-lf right, 7-lf below
int near_sadidx[8] = {0, 1, 2, 3, 4, 5, 6, 7};
int saddone=0;
int sr=0; //search range got from mv_pred(). It uses step_param levels. (0-7)
@@ -1835,60 +1891,11 @@
lf_or_gf = frame_lf_or_gf[x->e_mbd.mode_info_context->mbmi.ref_frame];
}
-
if(x->e_mbd.mode_info_context->mbmi.mode == NEWMV)
{
if(!saddone)
{
- //calculate sad for current frame 3 nearby MBs.
- if( xd->mb_to_top_edge==0 && xd->mb_to_left_edge ==0)
- {
- near_sad[0] = near_sad[1] = near_sad[2] = INT_MAX;
- }else if(xd->mb_to_top_edge==0)
- { //only has left MB for sad calculation.
- near_sad[0] = near_sad[2] = INT_MAX;
- near_sad[1] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - 16,xd->dst.y_stride, 0x7fffffff);
- }else if(xd->mb_to_left_edge ==0)
- { //only has left MB for sad calculation.
- near_sad[1] = near_sad[2] = INT_MAX;
- near_sad[0] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - xd->dst.y_stride *16,xd->dst.y_stride, 0x7fffffff);
- }else
- {
- near_sad[0] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - xd->dst.y_stride *16,xd->dst.y_stride, 0x7fffffff);
- near_sad[1] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - 16,xd->dst.y_stride, 0x7fffffff);
- near_sad[2] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - xd->dst.y_stride *16 -16,xd->dst.y_stride, 0x7fffffff);
- }
-
- if(cpi->common.last_frame_type != KEY_FRAME)
- {
- //calculate sad for last frame 5 nearby MBs.
- unsigned char *pre_y_buffer = cpi->common.yv12_fb[cpi->common.lst_fb_idx].y_buffer + recon_yoffset;
- int pre_y_stride = cpi->common.yv12_fb[cpi->common.lst_fb_idx].y_stride;
-
- if(xd->mb_to_top_edge==0) near_sad[4] = INT_MAX;
- if(xd->mb_to_left_edge ==0) near_sad[5] = INT_MAX;
- if(xd->mb_to_right_edge ==0) near_sad[6] = INT_MAX;
- if(xd->mb_to_bottom_edge==0) near_sad[7] = INT_MAX;
-
- if(near_sad[4] != INT_MAX)
- near_sad[4] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer - pre_y_stride *16, pre_y_stride, 0x7fffffff);
- if(near_sad[5] != INT_MAX)
- near_sad[5] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer - 16, pre_y_stride, 0x7fffffff);
- near_sad[3] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer, pre_y_stride, 0x7fffffff);
- if(near_sad[6] != INT_MAX)
- near_sad[6] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer + 16, pre_y_stride, 0x7fffffff);
- if(near_sad[7] != INT_MAX)
- near_sad[7] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer + pre_y_stride *16, pre_y_stride, 0x7fffffff);
- }
-
- if(cpi->common.last_frame_type != KEY_FRAME)
- {
- quicksortsad(near_sad, near_sadidx, 0, 7);
- }else
- {
- quicksortsad(near_sad, near_sadidx, 0, 2);
- }
-
+ vp8_cal_sad(cpi,xd,x, recon_yoffset ,&near_sadidx[0] );
saddone = 1;
}
diff --git a/vp8/encoder/rdopt.h b/vp8/encoder/rdopt.h
index fb74dd4..d874409 100644
--- a/vp8/encoder/rdopt.h
+++ b/vp8/encoder/rdopt.h
@@ -17,5 +17,17 @@
int vp8_rd_pick_intra_mbuv_mode(VP8_COMP *cpi, MACROBLOCK *x, int *rate, int *rate_to, int *distortion);
extern int vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int recon_uvoffset, int *returnrate, int *returndistortion, int *returnintra);
+extern void vp8_mv_pred
+(
+ VP8_COMP *cpi,
+ MACROBLOCKD *xd,
+ const MODE_INFO *here,
+ MV *mvp,
+ int refframe,
+ int *ref_frame_sign_bias,
+ int *sr,
+ int near_sadidx[]
+);
+void vp8_cal_sad(VP8_COMP *cpi, MACROBLOCKD *xd, MACROBLOCK *x, int recon_yoffset, int near_sadidx[]);
#endif
