Merge "Removed bmi copy to/from BLOCKD"
diff --git a/vp8/common/x86/idctllm_sse2.asm b/vp8/common/x86/idctllm_sse2.asm
index 34a7e18..83d3765 100644
--- a/vp8/common/x86/idctllm_sse2.asm
+++ b/vp8/common/x86/idctllm_sse2.asm
@@ -11,7 +11,7 @@
%include "vpx_ports/x86_abi_support.asm"
-;void idct_dequant_0_2x_sse2
+;void vp8_idct_dequant_0_2x_sse2
; (
; short *qcoeff - 0
; short *dequant - 1
@@ -21,8 +21,8 @@
; int blk_stride - 5
; )
-global sym(idct_dequant_0_2x_sse2)
-sym(idct_dequant_0_2x_sse2):
+global sym(vp8_idct_dequant_0_2x_sse2)
+sym(vp8_idct_dequant_0_2x_sse2):
push rbp
mov rbp, rsp
SHADOW_ARGS_TO_STACK 6
@@ -97,8 +97,8 @@
pop rbp
ret
-global sym(idct_dequant_full_2x_sse2)
-sym(idct_dequant_full_2x_sse2):
+global sym(vp8_idct_dequant_full_2x_sse2)
+sym(vp8_idct_dequant_full_2x_sse2):
push rbp
mov rbp, rsp
SHADOW_ARGS_TO_STACK 7
@@ -353,7 +353,7 @@
pop rbp
ret
-;void idct_dequant_dc_0_2x_sse2
+;void vp8_idct_dequant_dc_0_2x_sse2
; (
; short *qcoeff - 0
; short *dequant - 1
@@ -362,8 +362,8 @@
; int dst_stride - 4
; short *dc - 5
; )
-global sym(idct_dequant_dc_0_2x_sse2)
-sym(idct_dequant_dc_0_2x_sse2):
+global sym(vp8_idct_dequant_dc_0_2x_sse2)
+sym(vp8_idct_dequant_dc_0_2x_sse2):
push rbp
mov rbp, rsp
SHADOW_ARGS_TO_STACK 7
@@ -438,8 +438,8 @@
pop rbp
ret
-global sym(idct_dequant_dc_full_2x_sse2)
-sym(idct_dequant_dc_full_2x_sse2):
+global sym(vp8_idct_dequant_dc_full_2x_sse2)
+sym(vp8_idct_dequant_dc_full_2x_sse2):
push rbp
mov rbp, rsp
SHADOW_ARGS_TO_STACK 7
diff --git a/vp8/decoder/x86/idct_blk_sse2.c b/vp8/decoder/x86/idct_blk_sse2.c
index 4c88db4..3a48068 100644
--- a/vp8/decoder/x86/idct_blk_sse2.c
+++ b/vp8/decoder/x86/idct_blk_sse2.c
@@ -12,17 +12,17 @@
#include "vp8/common/idct.h"
#include "vp8/decoder/dequantize.h"
-void idct_dequant_dc_0_2x_sse2
+void vp8_idct_dequant_dc_0_2x_sse2
(short *q, short *dq, unsigned char *pre,
unsigned char *dst, int dst_stride, short *dc);
-void idct_dequant_dc_full_2x_sse2
+void vp8_idct_dequant_dc_full_2x_sse2
(short *q, short *dq, unsigned char *pre,
unsigned char *dst, int dst_stride, short *dc);
-void idct_dequant_0_2x_sse2
+void vp8_idct_dequant_0_2x_sse2
(short *q, short *dq ,unsigned char *pre,
unsigned char *dst, int dst_stride, int blk_stride);
-void idct_dequant_full_2x_sse2
+void vp8_idct_dequant_full_2x_sse2
(short *q, short *dq ,unsigned char *pre,
unsigned char *dst, int dst_stride, int blk_stride);
@@ -35,14 +35,14 @@
for (i = 0; i < 4; i++)
{
if (((short *)(eobs))[0] & 0xfefe)
- idct_dequant_dc_full_2x_sse2 (q, dq, pre, dst, stride, dc);
+ vp8_idct_dequant_dc_full_2x_sse2 (q, dq, pre, dst, stride, dc);
else
- idct_dequant_dc_0_2x_sse2 (q, dq, pre, dst, stride, dc);
+ vp8_idct_dequant_dc_0_2x_sse2 (q, dq, pre, dst, stride, dc);
if (((short *)(eobs))[1] & 0xfefe)
- idct_dequant_dc_full_2x_sse2 (q+32, dq, pre+8, dst+8, stride, dc+2);
+ vp8_idct_dequant_dc_full_2x_sse2 (q+32, dq, pre+8, dst+8, stride, dc+2);
else
- idct_dequant_dc_0_2x_sse2 (q+32, dq, pre+8, dst+8, stride, dc+2);
+ vp8_idct_dequant_dc_0_2x_sse2 (q+32, dq, pre+8, dst+8, stride, dc+2);
q += 64;
dc += 4;
@@ -61,14 +61,14 @@
for (i = 0; i < 4; i++)
{
if (((short *)(eobs))[0] & 0xfefe)
- idct_dequant_full_2x_sse2 (q, dq, pre, dst, stride, 16);
+ vp8_idct_dequant_full_2x_sse2 (q, dq, pre, dst, stride, 16);
else
- idct_dequant_0_2x_sse2 (q, dq, pre, dst, stride, 16);
+ vp8_idct_dequant_0_2x_sse2 (q, dq, pre, dst, stride, 16);
if (((short *)(eobs))[1] & 0xfefe)
- idct_dequant_full_2x_sse2 (q+32, dq, pre+8, dst+8, stride, 16);
+ vp8_idct_dequant_full_2x_sse2 (q+32, dq, pre+8, dst+8, stride, 16);
else
- idct_dequant_0_2x_sse2 (q+32, dq, pre+8, dst+8, stride, 16);
+ vp8_idct_dequant_0_2x_sse2 (q+32, dq, pre+8, dst+8, stride, 16);
q += 64;
pre += 64;
@@ -82,33 +82,33 @@
unsigned char *dstu, unsigned char *dstv, int stride, char *eobs)
{
if (((short *)(eobs))[0] & 0xfefe)
- idct_dequant_full_2x_sse2 (q, dq, pre, dstu, stride, 8);
+ vp8_idct_dequant_full_2x_sse2 (q, dq, pre, dstu, stride, 8);
else
- idct_dequant_0_2x_sse2 (q, dq, pre, dstu, stride, 8);
+ vp8_idct_dequant_0_2x_sse2 (q, dq, pre, dstu, stride, 8);
q += 32;
pre += 32;
dstu += stride*4;
if (((short *)(eobs))[1] & 0xfefe)
- idct_dequant_full_2x_sse2 (q, dq, pre, dstu, stride, 8);
+ vp8_idct_dequant_full_2x_sse2 (q, dq, pre, dstu, stride, 8);
else
- idct_dequant_0_2x_sse2 (q, dq, pre, dstu, stride, 8);
+ vp8_idct_dequant_0_2x_sse2 (q, dq, pre, dstu, stride, 8);
q += 32;
pre += 32;
if (((short *)(eobs))[2] & 0xfefe)
- idct_dequant_full_2x_sse2 (q, dq, pre, dstv, stride, 8);
+ vp8_idct_dequant_full_2x_sse2 (q, dq, pre, dstv, stride, 8);
else
- idct_dequant_0_2x_sse2 (q, dq, pre, dstv, stride, 8);
+ vp8_idct_dequant_0_2x_sse2 (q, dq, pre, dstv, stride, 8);
q += 32;
pre += 32;
dstv += stride*4;
if (((short *)(eobs))[3] & 0xfefe)
- idct_dequant_full_2x_sse2 (q, dq, pre, dstv, stride, 8);
+ vp8_idct_dequant_full_2x_sse2 (q, dq, pre, dstv, stride, 8);
else
- idct_dequant_0_2x_sse2 (q, dq, pre, dstv, stride, 8);
+ vp8_idct_dequant_0_2x_sse2 (q, dq, pre, dstv, stride, 8);
}
diff --git a/vp8/encoder/encodeframe.c b/vp8/encoder/encodeframe.c
index 3f14124..19b52a3 100644
--- a/vp8/encoder/encodeframe.c
+++ b/vp8/encoder/encodeframe.c
@@ -1255,7 +1255,11 @@
cpi->zbin_mode_boost = MV_ZBIN_BOOST;
}
}
- vp8_update_zbin_extra(cpi, x);
+
+ /* The fast quantizer doesn't use zbin_extra, only do so with
+ * the regular quantizer. */
+ if (cpi->sf.improved_quant)
+ vp8_update_zbin_extra(cpi, x);
}
cpi->count_mb_ref_frame_usage[xd->mode_info_context->mbmi.ref_frame] ++;
diff --git a/vp8/encoder/firstpass.c b/vp8/encoder/firstpass.c
index 9cdc1e5..8559142 100644
--- a/vp8/encoder/firstpass.c
+++ b/vp8/encoder/firstpass.c
@@ -1473,7 +1473,6 @@
int i;
double boost_score = 0.0;
- double fwd_boost_score = 0.0;
double mv_ratio_accumulator = 0.0;
double decay_accumulator = 1.0;
double this_frame_mv_in_out = 0.0;
diff --git a/vp8/encoder/onyx_if.c b/vp8/encoder/onyx_if.c
index ff9a641..35e187e 100644
--- a/vp8/encoder/onyx_if.c
+++ b/vp8/encoder/onyx_if.c
@@ -3237,16 +3237,17 @@
// 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->compressor_speed == 2)
{
- if(cpi->force_next_frame_intra)
+ if(cpi->oxcf.auto_key && cm->frame_type != KEY_FRAME)
{
- cm->frame_type = KEY_FRAME; /* delayed intra frame */
+ if(cpi->force_next_frame_intra)
+ {
+ cm->frame_type = KEY_FRAME; /* delayed intra frame */
+ }
}
+ cpi->force_next_frame_intra = 0;
}
- 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)
@@ -3775,15 +3776,15 @@
// (assuming that we didn't)!
if (cpi->pass != 2 && cpi->oxcf.auto_key && cm->frame_type != KEY_FRAME)
{
+ int key_frame_decision = decide_key_frame(cpi);
-#if CONFIG_REALTIME_ONLY
+ if (cpi->compressor_speed == 2)
{
/* 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);
+ cpi->force_next_frame_intra = key_frame_decision;
}
-#else
- if (decide_key_frame(cpi))
+ else if (key_frame_decision)
{
// Reset all our sizing numbers and recode
cm->frame_type = KEY_FRAME;
@@ -3820,7 +3821,6 @@
continue;
}
-#endif
}
vp8_clear_system_state();
diff --git a/vp8/encoder/onyx_int.h b/vp8/encoder/onyx_int.h
index f75f6cb..aead2fb 100644
--- a/vp8/encoder/onyx_int.h
+++ b/vp8/encoder/onyx_int.h
@@ -609,9 +609,8 @@
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
+
int droppable;
} VP8_COMP;
