Merge "ARMv6 optimized Intra4x4 prediction"
diff --git a/build/make/configure.sh b/build/make/configure.sh
index 1bb50af..1279f78 100755
--- a/build/make/configure.sh
+++ b/build/make/configure.sh
@@ -901,8 +901,12 @@
[ "${AS##*/}" = nasm ] && add_asflags -Ox
AS_SFX=.asm
case ${tgt_os} in
- win*)
- add_asflags -f win${bits}
+ win32)
+ add_asflags -f win32
+ enabled debug && add_asflags -g cv8
+ ;;
+ win64)
+ add_asflags -f x64
enabled debug && add_asflags -g cv8
;;
linux*|solaris*)
diff --git a/vp8/common/recon.c b/vp8/common/recon.c
deleted file mode 100644
index eebf6dc..0000000
--- a/vp8/common/recon.c
+++ /dev/null
@@ -1,181 +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.
- */
-
-
-#include "vpx_config.h"
-#include "recon.h"
-#include "blockd.h"
-
-void vp8_recon_b_c
-(
- unsigned char *pred_ptr,
- short *diff_ptr,
- unsigned char *dst_ptr,
- int stride
-)
-{
- int r, c;
-
- for (r = 0; r < 4; r++)
- {
- for (c = 0; c < 4; c++)
- {
- int a = diff_ptr[c] + pred_ptr[c] ;
-
- if (a < 0)
- a = 0;
-
- if (a > 255)
- a = 255;
-
- dst_ptr[c] = (unsigned char) a ;
- }
-
- dst_ptr += stride;
- diff_ptr += 16;
- pred_ptr += 16;
- }
-}
-
-void vp8_recon4b_c
-(
- unsigned char *pred_ptr,
- short *diff_ptr,
- unsigned char *dst_ptr,
- int stride
-)
-{
- int r, c;
-
- for (r = 0; r < 4; r++)
- {
- for (c = 0; c < 16; c++)
- {
- int a = diff_ptr[c] + pred_ptr[c] ;
-
- if (a < 0)
- a = 0;
-
- if (a > 255)
- a = 255;
-
- dst_ptr[c] = (unsigned char) a ;
- }
-
- dst_ptr += stride;
- diff_ptr += 16;
- pred_ptr += 16;
- }
-}
-
-void vp8_recon2b_c
-(
- unsigned char *pred_ptr,
- short *diff_ptr,
- unsigned char *dst_ptr,
- int stride
-)
-{
- int r, c;
-
- for (r = 0; r < 4; r++)
- {
- for (c = 0; c < 8; c++)
- {
- int a = diff_ptr[c] + pred_ptr[c] ;
-
- if (a < 0)
- a = 0;
-
- if (a > 255)
- a = 255;
-
- dst_ptr[c] = (unsigned char) a ;
- }
-
- dst_ptr += stride;
- diff_ptr += 8;
- pred_ptr += 8;
- }
-}
-
-void vp8_recon_mby_c(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
-{
-#if ARCH_ARM
- BLOCKD *b = &x->block[0];
- RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
-
- /*b = &x->block[4];*/
- b += 4;
- RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
-
- /*b = &x->block[8];*/
- b += 4;
- RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
-
- /*b = &x->block[12];*/
- b += 4;
- RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
-#else
- int i;
-
- for (i = 0; i < 16; i += 4)
- {
- BLOCKD *b = &x->block[i];
-
- RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
- }
-#endif
-}
-
-void vp8_recon_mb_c(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
-{
-#if ARCH_ARM
- BLOCKD *b = &x->block[0];
-
- RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
- b += 4;
- RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
- b += 4;
- RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
- b += 4;
- RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
- b += 4;
-
- /*b = &x->block[16];*/
-
- RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
- b++;
- b++;
- RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
- b++;
- b++;
- RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
- b++;
- b++;
- RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
-#else
- int i;
-
- for (i = 0; i < 16; i += 4)
- {
- BLOCKD *b = &x->block[i];
-
- RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
- }
-
- for (i = 16; i < 24; i += 2)
- {
- BLOCKD *b = &x->block[i];
-
- RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
- }
-#endif
-}
diff --git a/vp8/common/threading.h b/vp8/common/threading.h
index 5927cb1..da6347d 100644
--- a/vp8/common/threading.h
+++ b/vp8/common/threading.h
@@ -17,7 +17,6 @@
/* Thread management macros */
#ifdef _WIN32
/* Win32 */
-#define _WIN32_WINNT 0x500 /* WINBASE.H - Enable signal_object_and_wait */
#include <process.h>
#include <windows.h>
#define THREAD_FUNCTION DWORD WINAPI
diff --git a/vp8/encoder/bitstream.c b/vp8/encoder/bitstream.c
index 45086cb..748b607 100644
--- a/vp8/encoder/bitstream.c
+++ b/vp8/encoder/bitstream.c
@@ -1214,20 +1214,20 @@
int k = 0;
unsigned int prev_coef_count_sum[MAX_ENTROPY_TOKENS] = {0};
int prev_coef_savings[MAX_ENTROPY_TOKENS] = {0};
+ const unsigned int (*probs)[MAX_ENTROPY_TOKENS];
/* Calculate new probabilities given the constraint that
* they must be equal over the prev coef contexts
*/
+
+ probs = (const unsigned int (*)[MAX_ENTROPY_TOKENS])
+ cpi->coef_counts[i][j];
+
+ /* Reset to default probabilities at key frames */
if (cpi->common.frame_type == KEY_FRAME)
- {
- /* Reset to default probabilities at key frames */
- sum_probs_over_prev_coef_context(default_coef_counts[i][j],
- prev_coef_count_sum);
- }
- else
- {
- sum_probs_over_prev_coef_context(cpi->coef_counts[i][j],
- prev_coef_count_sum);
- }
+ probs = default_coef_counts[i][j];
+
+ sum_probs_over_prev_coef_context(probs, prev_coef_count_sum);
+
do
{
/* at every context */
diff --git a/vp8/encoder/encodeframe.c b/vp8/encoder/encodeframe.c
index 7f2b46d..e996368 100644
--- a/vp8/encoder/encodeframe.c
+++ b/vp8/encoder/encodeframe.c
@@ -465,7 +465,7 @@
else
xd->mode_info_context->mbmi.segment_id = 0;
- vp8cx_mb_init_quantizer(cpi, x);
+ vp8cx_mb_init_quantizer(cpi, x, 1);
}
else
xd->mode_info_context->mbmi.segment_id = 0; // Set to Segment 0 by default
@@ -1255,7 +1255,7 @@
xd->mode_info_context->mbmi.segment_id = 0;
/* segment_id changed, so update */
- vp8cx_mb_init_quantizer(cpi, x);
+ vp8cx_mb_init_quantizer(cpi, x, 1);
}
}
}
diff --git a/vp8/encoder/ethreading.c b/vp8/encoder/ethreading.c
index 748942b..557080d 100644
--- a/vp8/encoder/ethreading.c
+++ b/vp8/encoder/ethreading.c
@@ -20,7 +20,7 @@
int recon_uvoffset);
extern int vp8cx_encode_intra_macro_block(VP8_COMP *cpi, MACROBLOCK *x,
TOKENEXTRA **t);
-extern void vp8cx_mb_init_quantizer(VP8_COMP *cpi, MACROBLOCK *x);
+extern void vp8cx_mb_init_quantizer(VP8_COMP *cpi, MACROBLOCK *x, int ok_to_skip);
extern void vp8_build_block_offsets(MACROBLOCK *x);
extern void vp8_setup_block_ptrs(MACROBLOCK *x);
@@ -163,7 +163,7 @@
else
xd->mode_info_context->mbmi.segment_id = 0;
- vp8cx_mb_init_quantizer(cpi, x);
+ vp8cx_mb_init_quantizer(cpi, x, 1);
}
else
xd->mode_info_context->mbmi.segment_id = 0; // Set to Segment 0 by default
diff --git a/vp8/encoder/onyx_if.c b/vp8/encoder/onyx_if.c
index 5a1cf1f..1d00e67 100644
--- a/vp8/encoder/onyx_if.c
+++ b/vp8/encoder/onyx_if.c
@@ -2605,6 +2605,7 @@
recon += recon_stride;
}
+ vp8_clear_system_state();
return total_sse;
}
@@ -3659,7 +3660,11 @@
else if (cpi->decimation_count > 0)
{
cpi->decimation_count --;
+
cpi->bits_off_target += cpi->av_per_frame_bandwidth;
+ if (cpi->bits_off_target > cpi->oxcf.maximum_buffer_size)
+ cpi->bits_off_target = cpi->oxcf.maximum_buffer_size;
+
cm->current_video_frame++;
cpi->frames_since_key++;
@@ -3676,9 +3681,11 @@
// Propagate bits saved by dropping the frame to higher layers
for (i=cpi->current_layer+1; i<cpi->oxcf.number_of_layers; i++)
{
- cpi->layer_context[i].bits_off_target
- += cpi->av_per_frame_bandwidth;
- cpi->layer_context[i].buffer_level = cpi->bits_off_target;
+ LAYER_CONTEXT *lc = &cpi->layer_context[i];
+ lc->bits_off_target += cpi->av_per_frame_bandwidth;
+ if (lc->bits_off_target > lc->maximum_buffer_size)
+ lc->bits_off_target = lc->maximum_buffer_size;
+ lc->buffer_level = lc->bits_off_target;
}
}
@@ -4525,6 +4532,10 @@
else
cpi->bits_off_target += cpi->av_per_frame_bandwidth - cpi->projected_frame_size;
+ // Clip the buffer level to the maximum specified buffer size
+ if (cpi->bits_off_target > cpi->oxcf.maximum_buffer_size)
+ cpi->bits_off_target = cpi->oxcf.maximum_buffer_size;
+
// Rolling monitors of whether we are over or underspending used to help regulate min and Max Q in two pass.
cpi->rolling_target_bits = ((cpi->rolling_target_bits * 3) + cpi->this_frame_target + 2) / 4;
cpi->rolling_actual_bits = ((cpi->rolling_actual_bits * 3) + cpi->projected_frame_size + 2) / 4;
@@ -4552,6 +4563,10 @@
lc->bits_off_target += bits_off_for_this_layer;
+ // Clip buffer level to maximum buffer size for the layer
+ if (lc->bits_off_target > lc->maximum_buffer_size)
+ lc->bits_off_target = lc->maximum_buffer_size;
+
lc->total_actual_bits += cpi->projected_frame_size;
lc->total_target_vs_actual += bits_off_for_this_layer;
lc->buffer_level = lc->bits_off_target;
diff --git a/vp8/encoder/quantize.c b/vp8/encoder/quantize.c
index 9c759fd..e57a264 100644
--- a/vp8/encoder/quantize.c
+++ b/vp8/encoder/quantize.c
@@ -583,7 +583,7 @@
cpi->zbin_mode_boost + \
x->act_zbin_adj ) ) >> 7)
-void vp8cx_mb_init_quantizer(VP8_COMP *cpi, MACROBLOCK *x)
+void vp8cx_mb_init_quantizer(VP8_COMP *cpi, MACROBLOCK *x, int ok_to_skip)
{
int i;
int QIndex;
@@ -607,7 +607,10 @@
else
QIndex = cpi->common.base_qindex;
- if (QIndex != x->q_index)
+ /* This initialization should be called at least once. Use ok_to_skip to
+ * decide if it is ok to skip.
+ */
+ if (!ok_to_skip || QIndex != x->q_index)
{
// Y
zbin_extra = ZBIN_EXTRA_Y;
@@ -716,7 +719,7 @@
cpi->zbin_mode_boost = 0;
// MB level quantizer setup
- vp8cx_mb_init_quantizer(cpi, &cpi->mb);
+ vp8cx_mb_init_quantizer(cpi, &cpi->mb, 0);
}
diff --git a/vp8/encoder/quantize.h b/vp8/encoder/quantize.h
index f1f0156..07b8813 100644
--- a/vp8/encoder/quantize.h
+++ b/vp8/encoder/quantize.h
@@ -86,7 +86,7 @@
extern void vp8_set_quantizer(struct VP8_COMP *cpi, int Q);
extern void vp8cx_frame_init_quantizer(struct VP8_COMP *cpi);
extern void vp8_update_zbin_extra(struct VP8_COMP *cpi, MACROBLOCK *x);
-extern void vp8cx_mb_init_quantizer(struct VP8_COMP *cpi, MACROBLOCK *x);
+extern void vp8cx_mb_init_quantizer(struct VP8_COMP *cpi, MACROBLOCK *x, int ok_to_skip);
extern void vp8cx_init_quantizer(struct VP8_COMP *cpi);
#endif
diff --git a/vp8/encoder/ratectrl.c b/vp8/encoder/ratectrl.c
index 1ac9050..dc6feb9 100644
--- a/vp8/encoder/ratectrl.c
+++ b/vp8/encoder/ratectrl.c
@@ -990,6 +990,8 @@
{
// Update the buffer level variable.
cpi->bits_off_target += cpi->av_per_frame_bandwidth;
+ if (cpi->bits_off_target > cpi->oxcf.maximum_buffer_size)
+ cpi->bits_off_target = cpi->oxcf.maximum_buffer_size;
cpi->buffer_level = cpi->bits_off_target;
}
else
diff --git a/vp8/encoder/rdopt.c b/vp8/encoder/rdopt.c
index f966b06..e8abf84 100644
--- a/vp8/encoder/rdopt.c
+++ b/vp8/encoder/rdopt.c
@@ -43,7 +43,6 @@
#endif
-extern void vp8cx_mb_init_quantizer(VP8_COMP *cpi, MACROBLOCK *x);
extern void vp8_update_zbin_extra(VP8_COMP *cpi, MACROBLOCK *x);
#define MAXF(a,b) (((a) > (b)) ? (a) : (b))
