Merge "Removed unused diff buffer"
diff --git a/args.c b/args.c
index 7b2cc3a..37ba778 100644
--- a/args.c
+++ b/args.c
@@ -57,7 +57,7 @@
}
else if (def->long_name)
{
- int name_len = strlen(def->long_name);
+ const size_t name_len = strlen(def->long_name);
if (strlen(arg.argv[0]) >= name_len + 2
&& arg.argv[0][1] == '-'
diff --git a/build/make/ads2gas_apple.pl b/build/make/ads2gas_apple.pl
index 78f4a97..81280bf 100755
--- a/build/make/ads2gas_apple.pl
+++ b/build/make/ads2gas_apple.pl
@@ -30,6 +30,8 @@
my @incoming_array;
+my @imported_functions;
+
# Perl trim function to remove whitespace from the start and end of the string
sub trim($)
{
@@ -132,7 +134,18 @@
# Make function visible to linker, and make additional symbol with
# prepended underscore
s/EXPORT\s+\|([\$\w]*)\|/.globl _$1\n\t.globl $1/;
- s/IMPORT\s+\|([\$\w]*)\|/.globl $1/;
+
+ # Prepend imported functions with _
+ if (s/IMPORT\s+\|([\$\w]*)\|/.globl $1/)
+ {
+ $function = trim($1);
+ push(@imported_functions, $function);
+ }
+
+ foreach $function (@imported_functions)
+ {
+ s/$function/_$function/;
+ }
# No vertical bars required; make additional symbol with prepended
# underscore
@@ -157,8 +170,8 @@
s/\sPRESERVE8/@ PRESERVE8/g;
# Strip PROC and ENDPROC
- s/PROC/@/g;
- s/ENDP/@/g;
+ s/\bPROC\b/@/g;
+ s/\bENDP\b/@/g;
# EQU directive
s/(.*)EQU(.*)/.set $1, $2/;
diff --git a/build/make/configure.sh b/build/make/configure.sh
index 1279f78..0426f92 100755
--- a/build/make/configure.sh
+++ b/build/make/configure.sh
@@ -561,6 +561,10 @@
tgt_isa=x86_64
tgt_os=darwin10
;;
+ *darwin11*)
+ tgt_isa=x86_64
+ tgt_os=darwin11
+ ;;
*mingw32*|*cygwin*)
[ -z "$tgt_isa" ] && tgt_isa=x86
tgt_os=win32
@@ -617,6 +621,9 @@
if [ -d "/Developer/SDKs/MacOSX10.6.sdk" ]; then
osx_sdk_dir="/Developer/SDKs/MacOSX10.6.sdk"
fi
+ if [ -d "/Developer/SDKs/MacOSX10.7.sdk" ]; then
+ osx_sdk_dir="/Developer/SDKs/MacOSX10.7.sdk"
+ fi
case ${toolchain} in
*-darwin8-*)
@@ -637,6 +644,12 @@
add_ldflags "-isysroot ${osx_sdk_dir}"
add_ldflags "-mmacosx-version-min=10.6"
;;
+ *-darwin11-*)
+ add_cflags "-isysroot ${osx_sdk_dir}"
+ add_cflags "-mmacosx-version-min=10.7"
+ add_ldflags "-isysroot ${osx_sdk_dir}"
+ add_ldflags "-mmacosx-version-min=10.7"
+ ;;
esac
# Handle Solaris variants. Solaris 10 needs -lposix4
@@ -732,7 +745,7 @@
TOOLCHAIN_PATH=${SDK_PATH}/usr/bin
CC=${TOOLCHAIN_PATH}/gcc
AR=${TOOLCHAIN_PATH}/ar
- LD=${TOOLCHAIN_PATH}/arm-apple-darwin10-gcc-4.2.1
+ LD=${TOOLCHAIN_PATH}/arm-apple-darwin10-llvm-gcc-4.2
AS=${TOOLCHAIN_PATH}/as
STRIP=${TOOLCHAIN_PATH}/strip
NM=${TOOLCHAIN_PATH}/nm
@@ -746,13 +759,13 @@
add_cflags -arch ${tgt_isa}
add_ldflags -arch_only ${tgt_isa}
- add_cflags "-isysroot ${SDK_PATH}/SDKs/iPhoneOS4.3.sdk"
+ add_cflags "-isysroot ${SDK_PATH}/SDKs/iPhoneOS5.0.sdk"
# This should be overridable
- alt_libc=${SDK_PATH}/SDKs/iPhoneOS4.3.sdk
+ alt_libc=${SDK_PATH}/SDKs/iPhoneOS5.0.sdk
# Add the paths for the alternate libc
- for d in usr/include usr/include/gcc/darwin/4.2/ usr/lib/gcc/arm-apple-darwin10/4.2.1/include/; do
+ for d in usr/include; do
try_dir="${alt_libc}/${d}"
[ -d "${try_dir}" ] && add_cflags -I"${try_dir}"
done
diff --git a/configure b/configure
index 363687a..6f20c6b 100755
--- a/configure
+++ b/configure
@@ -119,6 +119,7 @@
all_platforms="${all_platforms} x86-win32-vs9"
all_platforms="${all_platforms} x86_64-darwin9-gcc"
all_platforms="${all_platforms} x86_64-darwin10-gcc"
+all_platforms="${all_platforms} x86_64-darwin11-gcc"
all_platforms="${all_platforms} x86_64-linux-gcc"
all_platforms="${all_platforms} x86_64-linux-icc"
all_platforms="${all_platforms} x86_64-solaris-gcc"
diff --git a/examples.mk b/examples.mk
index 1f7dcc1..f6c9045 100644
--- a/examples.mk
+++ b/examples.mk
@@ -102,6 +102,7 @@
+= third_party/libyuv/include/libyuv/basic_types.h \
third_party/libyuv/include/libyuv/cpu_id.h \
third_party/libyuv/include/libyuv/scale.h \
+ third_party/libyuv/source/row.h \
third_party/libyuv/source/scale.c \
third_party/libyuv/source/cpu_id.c
vp8_multi_resolution_encoder.GUID = 04f8738e-63c8-423b-90fa-7c2703a374de
diff --git a/third_party/libyuv/README.webm b/third_party/libyuv/README.webm
index 32766be..d3495ca 100644
--- a/third_party/libyuv/README.webm
+++ b/third_party/libyuv/README.webm
@@ -1,6 +1,6 @@
Name: libyuv
URL: http://code.google.com/p/libyuv/
-Version: 90
+Version: 102
License: BSD
License File: LICENSE
diff --git a/third_party/libyuv/include/libyuv/basic_types.h b/third_party/libyuv/include/libyuv/basic_types.h
index 87f8bd2..30504ce 100644
--- a/third_party/libyuv/include/libyuv/basic_types.h
+++ b/third_party/libyuv/include/libyuv/basic_types.h
@@ -13,22 +13,13 @@
#include <stddef.h> // for NULL, size_t
-#ifndef WIN32
+#if !(defined(_MSC_VER) && (_MSC_VER < 1600))
#include <stdint.h> // for uintptr_t
#endif
#ifndef INT_TYPES_DEFINED
#define INT_TYPES_DEFINED
#ifdef COMPILER_MSVC
-typedef __int64 int64;
-#else
-typedef long long int64;
-#endif /* COMPILER_MSVC */
-typedef int int32;
-typedef short int16;
-typedef char int8;
-
-#ifdef COMPILER_MSVC
typedef unsigned __int64 uint64;
typedef __int64 int64;
#ifndef INT64_C
@@ -38,9 +29,20 @@
#define UINT64_C(x) x ## UI64
#endif
#define INT64_F "I64"
-#else
+#else // COMPILER_MSVC
+#ifdef __LP64__
+typedef unsigned long uint64;
+typedef long int64;
+#ifndef INT64_C
+#define INT64_C(x) x ## L
+#endif
+#ifndef UINT64_C
+#define UINT64_C(x) x ## UL
+#endif
+#define INT64_F "l"
+#else // __LP64__
typedef unsigned long long uint64;
-//typedef long long int64;
+typedef long long int64;
#ifndef INT64_C
#define INT64_C(x) x ## LL
#endif
@@ -48,10 +50,14 @@
#define UINT64_C(x) x ## ULL
#endif
#define INT64_F "ll"
-#endif /* COMPILER_MSVC */
+#endif // __LP64__
+#endif // COMPILER_MSVC
typedef unsigned int uint32;
+typedef int int32;
typedef unsigned short uint16;
+typedef short int16;
typedef unsigned char uint8;
+typedef char int8;
#endif // INT_TYPES_DEFINED
// Detect compiler is for x86 or x64.
@@ -60,7 +66,6 @@
#define CPU_X86 1
#endif
-#define IS_ALIGNED(p, a) (0==((uintptr_t)(p) & ((a)-1)))
#define ALIGNP(p, t) \
((uint8*)((((uintptr_t)(p) + \
((t)-1)) & ~((t)-1))))
diff --git a/third_party/libyuv/include/libyuv/cpu_id.h b/third_party/libyuv/include/libyuv/cpu_id.h
index 8ebafe9..4a53b5b 100644
--- a/third_party/libyuv/include/libyuv/cpu_id.h
+++ b/third_party/libyuv/include/libyuv/cpu_id.h
@@ -11,21 +11,39 @@
#ifndef INCLUDE_LIBYUV_CPU_ID_H_
#define INCLUDE_LIBYUV_CPU_ID_H_
-//namespace libyuv {
+#ifdef __cplusplus
+namespace libyuv {
+extern "C" {
+#endif
// These flags are only valid on x86 processors
static const int kCpuHasSSE2 = 1;
static const int kCpuHasSSSE3 = 2;
-// SIMD support on ARM processors
+// These flags are only valid on ARM processors
static const int kCpuHasNEON = 4;
+// Internal flag to indicate cpuid is initialized.
+static const int kCpuInitialized = 8;
+
// Detect CPU has SSE2 etc.
-int TestCpuFlag(int flag);
+// test_flag parameter should be one of kCpuHas constants above
+// returns non-zero if instruction set is detected
+static __inline int TestCpuFlag(int test_flag) {
+ extern int cpu_info_;
+ extern int InitCpuFlags();
+ return (cpu_info_ ? cpu_info_ : InitCpuFlags()) & test_flag;
+}
// For testing, allow CPU flags to be disabled.
-void MaskCpuFlagsForTest(int enable_flags);
+// ie MaskCpuFlags(~kCpuHasSSSE3) to disable SSSE3.
+// -1 to enable all cpu specific optimizations.
+// 0 to disable all cpu specific optimizations.
+void MaskCpuFlags(int enable_flags);
-//} // namespace libyuv
+#ifdef __cplusplus
+} // extern "C"
+} // namespace libyuv
+#endif
#endif // INCLUDE_LIBYUV_CPU_ID_H_
diff --git a/third_party/libyuv/include/libyuv/scale.h b/third_party/libyuv/include/libyuv/scale.h
index 5b2d364..21fe360 100644
--- a/third_party/libyuv/include/libyuv/scale.h
+++ b/third_party/libyuv/include/libyuv/scale.h
@@ -13,7 +13,10 @@
#include "third_party/libyuv/include/libyuv/basic_types.h"
-//namespace libyuv {
+#ifdef __cplusplus
+namespace libyuv {
+extern "C" {
+#endif
// Supported filtering
typedef enum {
@@ -42,16 +45,8 @@
int dst_width, int dst_height,
FilterMode filtering);
-// Legacy API
-// If dst_height_offset is non-zero, the image is offset by that many pixels
-// and stretched to (dst_height - dst_height_offset * 2) pixels high,
-// instead of dst_height.
-int Scale_1(const uint8* src, int src_width, int src_height,
- uint8* dst, int dst_width, int dst_height, int dst_height_offset,
- int interpolate);
-
-// Same, but specified src terms of each plane location and stride.
-int Scale_2(const uint8* src_y, const uint8* src_u, const uint8* src_v,
+// Legacy API. Deprecated
+int Scale(const uint8* src_y, const uint8* src_u, const uint8* src_v,
int src_stride_y, int src_stride_u, int src_stride_v,
int src_width, int src_height,
uint8* dst_y, uint8* dst_u, uint8* dst_v,
@@ -59,9 +54,17 @@
int dst_width, int dst_height,
int interpolate);
+// Legacy API. Deprecated
+int ScaleOffset(const uint8* src, int src_width, int src_height,
+ uint8* dst, int dst_width, int dst_height, int dst_yoffset,
+ int interpolate);
+
// For testing, allow disabling of optimizations.
void SetUseReferenceImpl(int use);
-//} // namespace libyuv
+#ifdef __cplusplus
+} // extern "C"
+} // namespace libyuv
+#endif
#endif // INCLUDE_LIBYUV_SCALE_H_
diff --git a/third_party/libyuv/source/cpu_id.c b/third_party/libyuv/source/cpu_id.c
index e3b66f2..fccf3dd 100644
--- a/third_party/libyuv/source/cpu_id.c
+++ b/third_party/libyuv/source/cpu_id.c
@@ -9,66 +9,73 @@
*/
#include "third_party/libyuv/include/libyuv/cpu_id.h"
-#include "third_party/libyuv/include/libyuv/basic_types.h" // for CPU_X86
#ifdef _MSC_VER
#include <intrin.h>
#endif
+#ifdef __ANDROID__
+#include <cpu-features.h>
+#endif
+
+#include "third_party/libyuv/include/libyuv/basic_types.h" // for CPU_X86
// TODO(fbarchard): Use cpuid.h when gcc 4.4 is used on OSX and Linux.
#if (defined(__pic__) || defined(__APPLE__)) && defined(__i386__)
static inline void __cpuid(int cpu_info[4], int info_type) {
- __asm__ volatile (
- "mov %%ebx, %%edi\n"
- "cpuid\n"
- "xchg %%edi, %%ebx\n"
+ asm volatile (
+ "mov %%ebx, %%edi \n"
+ "cpuid \n"
+ "xchg %%edi, %%ebx \n"
: "=a"(cpu_info[0]), "=D"(cpu_info[1]), "=c"(cpu_info[2]), "=d"(cpu_info[3])
: "a"(info_type)
);
}
#elif defined(__i386__) || defined(__x86_64__)
static inline void __cpuid(int cpu_info[4], int info_type) {
- __asm__ volatile (
- "cpuid\n"
+ asm volatile (
+ "cpuid \n"
: "=a"(cpu_info[0]), "=b"(cpu_info[1]), "=c"(cpu_info[2]), "=d"(cpu_info[3])
: "a"(info_type)
);
}
#endif
-//namespace libyuv {
+#ifdef __cplusplus
+namespace libyuv {
+extern "C" {
+#endif
// CPU detect function for SIMD instruction sets.
-static int cpu_info_initialized_ = 0;
-static int cpu_info_ = 0;
+int cpu_info_ = 0;
-// Global lock for cpu initialization.
-static void InitCpuFlags() {
+int InitCpuFlags() {
#ifdef CPU_X86
int cpu_info[4];
__cpuid(cpu_info, 1);
- cpu_info_ = (cpu_info[2] & 0x00000200 ? kCpuHasSSSE3 : 0) |
- (cpu_info[3] & 0x04000000 ? kCpuHasSSE2 : 0);
+ cpu_info_ = (cpu_info[3] & 0x04000000 ? kCpuHasSSE2 : 0) |
+ (cpu_info[2] & 0x00000200 ? kCpuHasSSSE3 : 0) |
+ kCpuInitialized;
+#elif defined(__ANDROID__) && defined(__ARM_NEON__)
+ uint64_t features = android_getCpuFeatures();
+ cpu_info_ = ((features & ANDROID_CPU_ARM_FEATURE_NEON) ? kCpuHasNEON : 0) |
+ kCpuInitialized;
#elif defined(__ARM_NEON__)
// gcc -mfpu=neon defines __ARM_NEON__
- // if code is specifically built for Neon-only, enable the flag.
- cpu_info_ |= kCpuHasNEON;
+ // Enable Neon if you want support for Neon and Arm, and use MaskCpuFlags
+ // to disable Neon on devices that do not have it.
+ cpu_info_ = kCpuHasNEON | kCpuInitialized;
#else
- cpu_info_ = 0;
+ cpu_info_ = kCpuInitialized;
#endif
- cpu_info_initialized_ = 1;
+ return cpu_info_;
}
-void MaskCpuFlagsForTest(int enable_flags) {
+void MaskCpuFlags(int enable_flags) {
InitCpuFlags();
- cpu_info_ &= enable_flags;
+ cpu_info_ = (cpu_info_ & enable_flags) | kCpuInitialized;
}
-int TestCpuFlag(int flag) {
- if (!cpu_info_initialized_) {
- InitCpuFlags();
- }
- return cpu_info_ & flag ? 1 : 0;
-}
-
-//} // namespace libyuv
+#ifdef __cplusplus
+} // extern "C"
+} // namespace libyuv
+#endif
diff --git a/third_party/libyuv/source/row.h b/third_party/libyuv/source/row.h
index 0486fe2..eabe180 100644
--- a/third_party/libyuv/source/row.h
+++ b/third_party/libyuv/source/row.h
@@ -14,7 +14,7 @@
#include "third_party/libyuv/include/libyuv/basic_types.h"
#define kMaxStride (2048 * 4)
-//#define IS_ALIGNED(p, a) (!((uintptr_t)(p) & ((a) - 1)))
+#define IS_ALIGNED(p, a) (!((uintptr_t)(p) & ((a) - 1)))
#if defined(COVERAGE_ENABLED) || defined(TARGET_IPHONE_SIMULATOR)
#define YUV_DISABLE_ASM
@@ -72,7 +72,10 @@
#define HAS_REVERSE_ROW_NEON
#endif
-//extern "C" {
+#ifdef __cplusplus
+namespace libyuv {
+extern "C" {
+#endif
#ifdef HAS_ARGBTOYROW_SSSE3
void ARGBToYRow_SSSE3(const uint8* src_argb, uint8* dst_y, int pix);
@@ -253,6 +256,9 @@
#endif
-//} // extern "C"
+#ifdef __cplusplus
+} // extern "C"
+} // namespace libyuv
+#endif
#endif // LIBYUV_SOURCE_ROW_H_
diff --git a/third_party/libyuv/source/scale.c b/third_party/libyuv/source/scale.c
index 02ffdac..930a7ae 100644
--- a/third_party/libyuv/source/scale.c
+++ b/third_party/libyuv/source/scale.c
@@ -15,6 +15,17 @@
#include "third_party/libyuv/include/libyuv/cpu_id.h"
#include "third_party/libyuv/source/row.h"
+
+#ifdef __cplusplus
+namespace libyuv {
+extern "C" {
+#endif
+
+/*
+ * Note: Defining YUV_DISABLE_ASM allows to use c version.
+ */
+//#define YUV_DISABLE_ASM
+
#if defined(_MSC_VER)
#define ALIGN16(var) __declspec(align(16)) var
#else
@@ -26,8 +37,6 @@
// Note: Some SSE2 reference manuals
// cpuvol1.pdf agner_instruction_tables.pdf 253666.pdf 253667.pdf
-//namespace libyuv {
-
// Set the following flag to true to revert to only
// using the reference implementation ScalePlaneBox(), and
// NOT the optimized versions. Useful for debugging and
@@ -40,9 +49,7 @@
use_reference_impl_ = use;
}
-// TODO: The preprocessor definitions for Win64 are not right in build system.
-// Disable optimized code for now.
-#define YUV_DISABLE_ASM
+// ScaleRowDown2Int also used by planar functions
/**
* NEON downscalers with interpolation.
@@ -511,83 +518,116 @@
!defined(YUV_DISABLE_ASM)
#if defined(_MSC_VER)
#define TALIGN16(t, var) __declspec(align(16)) t _ ## var
-#elif defined(OSX) && defined(__i386__)
+#elif (defined(__APPLE__) || defined(__MINGW32__) || defined(__CYGWIN__)) && defined(__i386__)
#define TALIGN16(t, var) t var __attribute__((aligned(16)))
#else
#define TALIGN16(t, var) t _ ## var __attribute__((aligned(16)))
#endif
+#if (defined(__APPLE__) || defined(__MINGW32__) || defined(__CYGWIN__)) && \
+ defined(__i386__)
+#define DECLARE_FUNCTION(name) \
+ ".text \n" \
+ ".globl _" #name " \n" \
+"_" #name ": \n"
+#else
+#define DECLARE_FUNCTION(name) \
+ ".text \n" \
+ ".global " #name " \n" \
+#name ": \n"
+#endif
+
+
// Offsets for source bytes 0 to 9
+//extern "C"
TALIGN16(const uint8, shuf0[16]) =
{ 0, 1, 3, 4, 5, 7, 8, 9, 128, 128, 128, 128, 128, 128, 128, 128 };
// Offsets for source bytes 11 to 20 with 8 subtracted = 3 to 12.
+//extern "C"
TALIGN16(const uint8, shuf1[16]) =
{ 3, 4, 5, 7, 8, 9, 11, 12, 128, 128, 128, 128, 128, 128, 128, 128 };
// Offsets for source bytes 21 to 31 with 16 subtracted = 5 to 31.
+//extern "C"
TALIGN16(const uint8, shuf2[16]) =
{ 5, 7, 8, 9, 11, 12, 13, 15, 128, 128, 128, 128, 128, 128, 128, 128 };
// Offsets for source bytes 0 to 10
+//extern "C"
TALIGN16(const uint8, shuf01[16]) =
{ 0, 1, 1, 2, 2, 3, 4, 5, 5, 6, 6, 7, 8, 9, 9, 10 };
// Offsets for source bytes 10 to 21 with 8 subtracted = 3 to 13.
+//extern "C"
TALIGN16(const uint8, shuf11[16]) =
{ 2, 3, 4, 5, 5, 6, 6, 7, 8, 9, 9, 10, 10, 11, 12, 13 };
// Offsets for source bytes 21 to 31 with 16 subtracted = 5 to 31.
+//extern "C"
TALIGN16(const uint8, shuf21[16]) =
{ 5, 6, 6, 7, 8, 9, 9, 10, 10, 11, 12, 13, 13, 14, 14, 15 };
// Coefficients for source bytes 0 to 10
+//extern "C"
TALIGN16(const uint8, madd01[16]) =
{ 3, 1, 2, 2, 1, 3, 3, 1, 2, 2, 1, 3, 3, 1, 2, 2 };
// Coefficients for source bytes 10 to 21
+//extern "C"
TALIGN16(const uint8, madd11[16]) =
{ 1, 3, 3, 1, 2, 2, 1, 3, 3, 1, 2, 2, 1, 3, 3, 1 };
// Coefficients for source bytes 21 to 31
+//extern "C"
TALIGN16(const uint8, madd21[16]) =
{ 2, 2, 1, 3, 3, 1, 2, 2, 1, 3, 3, 1, 2, 2, 1, 3 };
// Coefficients for source bytes 21 to 31
+//extern "C"
TALIGN16(const int16, round34[8]) =
{ 2, 2, 2, 2, 2, 2, 2, 2 };
+//extern "C"
TALIGN16(const uint8, shuf38a[16]) =
{ 0, 3, 6, 8, 11, 14, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 };
+//extern "C"
TALIGN16(const uint8, shuf38b[16]) =
{ 128, 128, 128, 128, 128, 128, 0, 3, 6, 8, 11, 14, 128, 128, 128, 128 };
// Arrange words 0,3,6 into 0,1,2
+//extern "C"
TALIGN16(const uint8, shufac0[16]) =
{ 0, 1, 6, 7, 12, 13, 128, 128, 128, 128, 128, 128, 128, 128, 128, 128 };
// Arrange words 0,3,6 into 3,4,5
+//extern "C"
TALIGN16(const uint8, shufac3[16]) =
{ 128, 128, 128, 128, 128, 128, 0, 1, 6, 7, 12, 13, 128, 128, 128, 128 };
// Scaling values for boxes of 3x3 and 2x3
+//extern "C"
TALIGN16(const uint16, scaleac3[8]) =
{ 65536 / 9, 65536 / 9, 65536 / 6, 65536 / 9, 65536 / 9, 65536 / 6, 0, 0 };
// Arrange first value for pixels 0,1,2,3,4,5
+//extern "C"
TALIGN16(const uint8, shufab0[16]) =
{ 0, 128, 3, 128, 6, 128, 8, 128, 11, 128, 14, 128, 128, 128, 128, 128 };
// Arrange second value for pixels 0,1,2,3,4,5
+//extern "C"
TALIGN16(const uint8, shufab1[16]) =
{ 1, 128, 4, 128, 7, 128, 9, 128, 12, 128, 15, 128, 128, 128, 128, 128 };
// Arrange third value for pixels 0,1,2,3,4,5
+//extern "C"
TALIGN16(const uint8, shufab2[16]) =
{ 2, 128, 5, 128, 128, 128, 10, 128, 13, 128, 128, 128, 128, 128, 128, 128 };
// Scaling values for boxes of 3x2 and 2x2
+//extern "C"
TALIGN16(const uint16, scaleab2[8]) =
{ 65536 / 3, 65536 / 3, 65536 / 2, 65536 / 3, 65536 / 3, 65536 / 2, 0, 0 };
#endif
@@ -1620,14 +1660,7 @@
void ScaleRowDown8Int_SSE2(const uint8* src_ptr, int src_stride,
uint8* dst_ptr, int dst_width);
asm(
- ".text \n"
-#if defined(OSX)
- ".globl _ScaleRowDown8Int_SSE2 \n"
-"_ScaleRowDown8Int_SSE2: \n"
-#else
- ".global ScaleRowDown8Int_SSE2 \n"
-"ScaleRowDown8Int_SSE2: \n"
-#endif
+ DECLARE_FUNCTION(ScaleRowDown8Int_SSE2)
"pusha \n"
"mov 0x24(%esp),%esi \n"
"mov 0x28(%esp),%ebx \n"
@@ -1691,14 +1724,7 @@
void ScaleRowDown34_SSSE3(const uint8* src_ptr, int src_stride,
uint8* dst_ptr, int dst_width);
asm(
- ".text \n"
-#if defined(OSX)
- ".globl _ScaleRowDown34_SSSE3 \n"
-"_ScaleRowDown34_SSSE3: \n"
-#else
- ".global ScaleRowDown34_SSSE3 \n"
-"ScaleRowDown34_SSSE3: \n"
-#endif
+ DECLARE_FUNCTION(ScaleRowDown34_SSSE3)
"pusha \n"
"mov 0x24(%esp),%esi \n"
"mov 0x2c(%esp),%edi \n"
@@ -1729,14 +1755,7 @@
void ScaleRowDown34_1_Int_SSSE3(const uint8* src_ptr, int src_stride,
uint8* dst_ptr, int dst_width);
asm(
- ".text \n"
-#if defined(OSX)
- ".globl _ScaleRowDown34_1_Int_SSSE3 \n"
-"_ScaleRowDown34_1_Int_SSSE3: \n"
-#else
- ".global ScaleRowDown34_1_Int_SSSE3 \n"
-"ScaleRowDown34_1_Int_SSSE3: \n"
-#endif
+ DECLARE_FUNCTION(ScaleRowDown34_1_Int_SSSE3)
"pusha \n"
"mov 0x24(%esp),%esi \n"
"mov 0x28(%esp),%ebp \n"
@@ -1790,14 +1809,7 @@
void ScaleRowDown34_0_Int_SSSE3(const uint8* src_ptr, int src_stride,
uint8* dst_ptr, int dst_width);
asm(
- ".text \n"
-#if defined(OSX)
- ".globl _ScaleRowDown34_0_Int_SSSE3 \n"
-"_ScaleRowDown34_0_Int_SSSE3: \n"
-#else
- ".global ScaleRowDown34_0_Int_SSSE3 \n"
-"ScaleRowDown34_0_Int_SSSE3: \n"
-#endif
+ DECLARE_FUNCTION(ScaleRowDown34_0_Int_SSSE3)
"pusha \n"
"mov 0x24(%esp),%esi \n"
"mov 0x28(%esp),%ebp \n"
@@ -1854,14 +1866,7 @@
void ScaleRowDown38_SSSE3(const uint8* src_ptr, int src_stride,
uint8* dst_ptr, int dst_width);
asm(
- ".text \n"
-#if defined(OSX)
- ".globl _ScaleRowDown38_SSSE3 \n"
-"_ScaleRowDown38_SSSE3: \n"
-#else
- ".global ScaleRowDown38_SSSE3 \n"
-"ScaleRowDown38_SSSE3: \n"
-#endif
+ DECLARE_FUNCTION(ScaleRowDown38_SSSE3)
"pusha \n"
"mov 0x24(%esp),%esi \n"
"mov 0x28(%esp),%edx \n"
@@ -1890,14 +1895,7 @@
void ScaleRowDown38_3_Int_SSSE3(const uint8* src_ptr, int src_stride,
uint8* dst_ptr, int dst_width);
asm(
- ".text \n"
-#if defined(OSX)
- ".globl _ScaleRowDown38_3_Int_SSSE3 \n"
-"_ScaleRowDown38_3_Int_SSSE3: \n"
-#else
- ".global ScaleRowDown38_3_Int_SSSE3 \n"
-"ScaleRowDown38_3_Int_SSSE3: \n"
-#endif
+ DECLARE_FUNCTION(ScaleRowDown38_3_Int_SSSE3)
"pusha \n"
"mov 0x24(%esp),%esi \n"
"mov 0x28(%esp),%edx \n"
@@ -1954,14 +1952,7 @@
void ScaleRowDown38_2_Int_SSSE3(const uint8* src_ptr, int src_stride,
uint8* dst_ptr, int dst_width);
asm(
- ".text \n"
-#if defined(OSX)
- ".globl _ScaleRowDown38_2_Int_SSSE3 \n"
-"_ScaleRowDown38_2_Int_SSSE3: \n"
-#else
- ".global ScaleRowDown38_2_Int_SSSE3 \n"
-"ScaleRowDown38_2_Int_SSSE3: \n"
-#endif
+ DECLARE_FUNCTION(ScaleRowDown38_2_Int_SSSE3)
"pusha \n"
"mov 0x24(%esp),%esi \n"
"mov 0x28(%esp),%edx \n"
@@ -2001,14 +1992,7 @@
uint16* dst_ptr, int src_width,
int src_height);
asm(
- ".text \n"
-#if defined(OSX)
- ".globl _ScaleAddRows_SSE2 \n"
-"_ScaleAddRows_SSE2: \n"
-#else
- ".global ScaleAddRows_SSE2 \n"
-"ScaleAddRows_SSE2: \n"
-#endif
+ DECLARE_FUNCTION(ScaleAddRows_SSE2)
"pusha \n"
"mov 0x24(%esp),%esi \n"
"mov 0x28(%esp),%edx \n"
@@ -2052,14 +2036,7 @@
const uint8* src_ptr, int src_stride,
int dst_width, int source_y_fraction);
asm(
- ".text \n"
-#if defined(OSX)
- ".globl _ScaleFilterRows_SSE2 \n"
-"_ScaleFilterRows_SSE2: \n"
-#else
- ".global ScaleFilterRows_SSE2 \n"
-"ScaleFilterRows_SSE2: \n"
-#endif
+ DECLARE_FUNCTION(ScaleFilterRows_SSE2)
"push %esi \n"
"push %edi \n"
"mov 0xc(%esp),%edi \n"
@@ -2147,14 +2124,7 @@
const uint8* src_ptr, int src_stride,
int dst_width, int source_y_fraction);
asm(
- ".text \n"
-#if defined(OSX)
- ".globl _ScaleFilterRows_SSSE3 \n"
-"_ScaleFilterRows_SSSE3: \n"
-#else
- ".global ScaleFilterRows_SSSE3 \n"
-"ScaleFilterRows_SSSE3: \n"
-#endif
+ DECLARE_FUNCTION(ScaleFilterRows_SSSE3)
"push %esi \n"
"push %edi \n"
"mov 0xc(%esp),%edi \n"
@@ -2318,7 +2288,7 @@
static void ScaleRowDown34_1_Int_SSSE3(const uint8* src_ptr, int src_stride,
uint8* dst_ptr, int dst_width) {
- asm volatile(
+ asm volatile (
"movdqa (%4),%%xmm2 \n" // _shuf01
"movdqa (%5),%%xmm3 \n" // _shuf11
"movdqa (%6),%%xmm4 \n" // _shuf21
@@ -2436,7 +2406,7 @@
#define HAS_SCALEROWDOWN38_SSSE3
static void ScaleRowDown38_SSSE3(const uint8* src_ptr, int src_stride,
uint8* dst_ptr, int dst_width) {
- asm volatile(
+ asm volatile (
"movdqa (%3),%%xmm4 \n"
"movdqa (%4),%%xmm5 \n"
"1:"
@@ -2560,7 +2530,7 @@
static void ScaleAddRows_SSE2(const uint8* src_ptr, int src_stride,
uint16* dst_ptr, int src_width,
int src_height) {
- asm volatile(
+ asm volatile (
"pxor %%xmm5,%%xmm5 \n"
"1:"
"movdqa (%0),%%xmm2 \n"
@@ -2602,7 +2572,7 @@
const uint8* src_ptr, int src_stride,
int dst_width, int source_y_fraction) {
if (source_y_fraction == 0) {
- asm volatile(
+ asm volatile (
"1:"
"movdqa (%1),%%xmm0 \n"
"lea 0x10(%1),%1 \n"
@@ -2620,7 +2590,7 @@
);
return;
} else if (source_y_fraction == 128) {
- asm volatile(
+ asm volatile (
"1:"
"movdqa (%1),%%xmm0 \n"
"movdqa (%1,%3,1),%%xmm2 \n"
@@ -2640,7 +2610,7 @@
);
return;
} else {
- asm volatile(
+ asm volatile (
"mov %3,%%eax \n"
"movd %%eax,%%xmm6 \n"
"punpcklwd %%xmm6,%%xmm6 \n"
@@ -2693,7 +2663,7 @@
const uint8* src_ptr, int src_stride,
int dst_width, int source_y_fraction) {
if (source_y_fraction == 0) {
- asm volatile(
+ asm volatile (
"1:"
"movdqa (%1),%%xmm0 \n"
"lea 0x10(%1),%1 \n"
@@ -2711,7 +2681,7 @@
);
return;
} else if (source_y_fraction == 128) {
- asm volatile(
+ asm volatile (
"1:"
"movdqa (%1),%%xmm0 \n"
"movdqa (%1,%3,1),%%xmm2 \n"
@@ -2731,7 +2701,7 @@
);
return;
} else {
- asm volatile(
+ asm volatile (
"mov %3,%%eax \n"
"shr %%eax \n"
"mov %%al,%%ah \n"
@@ -3095,10 +3065,7 @@
ScaleRowDown2 = filtering ? ScaleRowDown2Int_NEON : ScaleRowDown2_NEON;
} else
#endif
-/* TODO: Force to call C version all the time in ordert to get matching results
- * in multi-resolution encoder example.
- */
-#if 0 //defined(HAS_SCALEROWDOWN2_SSE2)
+#if defined(HAS_SCALEROWDOWN2_SSE2)
if (TestCpuFlag(kCpuHasSSE2) &&
IS_ALIGNED(dst_width, 16) &&
IS_ALIGNED(src_ptr, 16) && IS_ALIGNED(src_stride, 16) &&
@@ -3267,33 +3234,33 @@
}
}
{
- int src_row = 0;
+ int src_row = 0;
int y;
for (y = 0; y < dst_height; ++y) {
- switch (src_row) {
- case 0:
- ScaleRowDown34_0(src_ptr, src_stride, dst_ptr, dst_width);
- break;
+ switch (src_row) {
+ case 0:
+ ScaleRowDown34_0(src_ptr, src_stride, dst_ptr, dst_width);
+ break;
- case 1:
- ScaleRowDown34_1(src_ptr, src_stride, dst_ptr, dst_width);
- break;
+ case 1:
+ ScaleRowDown34_1(src_ptr, src_stride, dst_ptr, dst_width);
+ break;
- case 2:
- ScaleRowDown34_0(src_ptr + src_stride, -src_stride,
- dst_ptr, dst_width);
- break;
- }
- ++src_row;
+ case 2:
+ ScaleRowDown34_0(src_ptr + src_stride, -src_stride,
+ dst_ptr, dst_width);
+ break;
+ }
+ ++src_row;
+ src_ptr += src_stride;
+ dst_ptr += dst_stride;
+ if (src_row >= 3) {
src_ptr += src_stride;
- dst_ptr += dst_stride;
- if (src_row >= 3) {
- src_ptr += src_stride;
- src_row = 0;
- }
+ src_row = 0;
}
}
}
+}
/**
* Scale plane, 3/8
@@ -3350,27 +3317,27 @@
}
}
{
- int src_row = 0;
+ int src_row = 0;
int y;
for (y = 0; y < dst_height; ++y) {
- switch (src_row) {
- case 0:
- case 1:
- ScaleRowDown38_3(src_ptr, src_stride, dst_ptr, dst_width);
- src_ptr += src_stride * 3;
- ++src_row;
- break;
+ switch (src_row) {
+ case 0:
+ case 1:
+ ScaleRowDown38_3(src_ptr, src_stride, dst_ptr, dst_width);
+ src_ptr += src_stride * 3;
+ ++src_row;
+ break;
- case 2:
- ScaleRowDown38_2(src_ptr, src_stride, dst_ptr, dst_width);
- src_ptr += src_stride * 2;
- src_row = 0;
- break;
- }
- dst_ptr += dst_stride;
+ case 2:
+ ScaleRowDown38_2(src_ptr, src_stride, dst_ptr, dst_width);
+ src_ptr += src_stride * 2;
+ src_row = 0;
+ break;
}
+ dst_ptr += dst_stride;
}
}
+}
__inline static uint32 SumBox(int iboxwidth, int iboxheight,
int src_stride, const uint8* src_ptr) {
@@ -3421,15 +3388,15 @@
scaletbl[0] = 65536 / (minboxwidth * boxheight);
scaletbl[1] = 65536 / ((minboxwidth + 1) * boxheight);
{
- int *scaleptr = scaletbl - minboxwidth;
- int x = 0;
+ int *scaleptr = scaletbl - minboxwidth;
+ int x = 0;
int i;
for (i = 0; i < dst_width; ++i) {
- int ix = x >> 16;
+ int ix = x >> 16;
int boxwidth;
- x += dx;
+ x += dx;
boxwidth = (x >> 16) - ix;
- *dst_ptr++ = SumPixels(boxwidth, src_ptr + ix) * scaleptr[boxwidth] >> 16;
+ *dst_ptr++ = SumPixels(boxwidth, src_ptr + ix) * scaleptr[boxwidth] >> 16;
}
}
}
@@ -3509,20 +3476,20 @@
}
{
- int y = 0;
+ int y = 0;
int j;
for (j = 0; j < dst_height; ++j) {
- int iy = y >> 16;
- const uint8* const src = src_ptr + iy * src_stride;
+ int iy = y >> 16;
+ const uint8* const src = src_ptr + iy * src_stride;
int boxheight;
- y += dy;
- if (y > (src_height << 16)) {
- y = (src_height << 16);
- }
+ y += dy;
+ if (y > (src_height << 16)) {
+ y = (src_height << 16);
+ }
boxheight = (y >> 16) - iy;
- ScaleAddRows(src, src_stride, row, src_width, boxheight);
- ScaleAddCols(dst_width, boxheight, dx, row, dst_ptr);
- dst_ptr += dst_stride;
+ ScaleAddRows(src, src_stride, row, src_width, boxheight);
+ ScaleAddCols(dst_width, boxheight, dx, row, dst_ptr);
+ dst_ptr += dst_stride;
}
}
}
@@ -3614,24 +3581,24 @@
ScaleFilterCols = ScaleFilterCols_C;
{
- int y = 0;
- int maxy = ((src_height - 1) << 16) - 1; // max is filter of last 2 rows.
+ int y = 0;
+ int maxy = ((src_height - 1) << 16) - 1; // max is filter of last 2 rows.
int j;
for (j = 0; j < dst_height; ++j) {
- int iy = y >> 16;
- int fy = (y >> 8) & 255;
- const uint8* const src = src_ptr + iy * src_stride;
- ScaleFilterRows(row, src, src_stride, src_width, fy);
- ScaleFilterCols(dst_ptr, row, dst_width, dx);
- dst_ptr += dst_stride;
- y += dy;
- if (y > maxy) {
- y = maxy;
- }
+ int iy = y >> 16;
+ int fy = (y >> 8) & 255;
+ const uint8* const src = src_ptr + iy * src_stride;
+ ScaleFilterRows(row, src, src_stride, src_width, fy);
+ ScaleFilterCols(dst_ptr, row, dst_width, dx);
+ dst_ptr += dst_stride;
+ y += dy;
+ if (y > maxy) {
+ y = maxy;
}
}
}
}
+}
/**
* Scale plane to/from any dimensions, without interpolation.
@@ -3818,36 +3785,32 @@
src_stride_v = -src_stride_v;
}
{
- int halfsrc_width = (src_width + 1) >> 1;
- int halfsrc_height = (src_height + 1) >> 1;
- int halfdst_width = (dst_width + 1) >> 1;
- int halfoheight = (dst_height + 1) >> 1;
+ int src_halfwidth = (src_width + 1) >> 1;
+ int src_halfheight = (src_height + 1) >> 1;
+ int dst_halfwidth = (dst_width + 1) >> 1;
+ int dst_halfheight = (dst_height + 1) >> 1;
ScalePlane(src_y, src_stride_y, src_width, src_height,
dst_y, dst_stride_y, dst_width, dst_height,
filtering, use_reference_impl_);
- ScalePlane(src_u, src_stride_u, halfsrc_width, halfsrc_height,
- dst_u, dst_stride_u, halfdst_width, halfoheight,
- filtering, use_reference_impl_);
- ScalePlane(src_v, src_stride_v, halfsrc_width, halfsrc_height,
- dst_v, dst_stride_v, halfdst_width, halfoheight,
- filtering, use_reference_impl_);
+ ScalePlane(src_u, src_stride_u, src_halfwidth, src_halfheight,
+ dst_u, dst_stride_u, dst_halfwidth, dst_halfheight,
+ filtering, use_reference_impl_);
+ ScalePlane(src_v, src_stride_v, src_halfwidth, src_halfheight,
+ dst_v, dst_stride_v, dst_halfwidth, dst_halfheight,
+ filtering, use_reference_impl_);
}
return 0;
}
-int Scale_2(const uint8* src_y, const uint8* src_u, const uint8* src_v,
+// Deprecated api
+int Scale(const uint8* src_y, const uint8* src_u, const uint8* src_v,
int src_stride_y, int src_stride_u, int src_stride_v,
int src_width, int src_height,
uint8* dst_y, uint8* dst_u, uint8* dst_v,
int dst_stride_y, int dst_stride_u, int dst_stride_v,
int dst_width, int dst_height,
int interpolate) {
- int halfsrc_width;
- int halfsrc_height;
- int halfdst_width;
- int halfoheight;
- FilterMode filtering;
if (!src_y || !src_u || !src_v || src_width <= 0 || src_height == 0 ||
!dst_y || !dst_u || !dst_v || dst_width <= 0 || dst_height <= 0) {
return -1;
@@ -3864,51 +3827,58 @@
src_stride_u = -src_stride_u;
src_stride_v = -src_stride_v;
}
- halfsrc_width = (src_width + 1) >> 1;
- halfsrc_height = (src_height + 1) >> 1;
- halfdst_width = (dst_width + 1) >> 1;
- halfoheight = (dst_height + 1) >> 1;
- filtering = interpolate ? kFilterBox : kFilterNone;
+ {
+ int src_halfwidth = (src_width + 1) >> 1;
+ int src_halfheight = (src_height + 1) >> 1;
+ int dst_halfwidth = (dst_width + 1) >> 1;
+ int dst_halfheight = (dst_height + 1) >> 1;
+ FilterMode filtering = interpolate ? kFilterBox : kFilterNone;
ScalePlane(src_y, src_stride_y, src_width, src_height,
dst_y, dst_stride_y, dst_width, dst_height,
filtering, use_reference_impl_);
- ScalePlane(src_u, src_stride_u, halfsrc_width, halfsrc_height,
- dst_u, dst_stride_u, halfdst_width, halfoheight,
+ ScalePlane(src_u, src_stride_u, src_halfwidth, src_halfheight,
+ dst_u, dst_stride_u, dst_halfwidth, dst_halfheight,
filtering, use_reference_impl_);
- ScalePlane(src_v, src_stride_v, halfsrc_width, halfsrc_height,
- dst_v, dst_stride_v, halfdst_width, halfoheight,
+ ScalePlane(src_v, src_stride_v, src_halfwidth, src_halfheight,
+ dst_v, dst_stride_v, dst_halfwidth, dst_halfheight,
filtering, use_reference_impl_);
+ }
return 0;
}
-int Scale_1(const uint8* src, int src_width, int src_height,
- uint8* dst, int dst_width, int dst_height, int ooffset,
+// Deprecated api
+int ScaleOffset(const uint8* src, int src_width, int src_height,
+ uint8* dst, int dst_width, int dst_height, int dst_yoffset,
int interpolate) {
if (!src || src_width <= 0 || src_height <= 0 ||
- !dst || dst_width <= 0 || dst_height <= 0 || ooffset < 0 ||
- ooffset >= dst_height) {
+ !dst || dst_width <= 0 || dst_height <= 0 || dst_yoffset < 0 ||
+ dst_yoffset >= dst_height) {
return -1;
}
- ooffset = ooffset & ~1; // chroma requires offset to multiple of 2.
+ dst_yoffset = dst_yoffset & ~1; // chroma requires offset to multiple of 2.
{
- int halfsrc_width = (src_width + 1) >> 1;
- int halfsrc_height = (src_height + 1) >> 1;
- int halfdst_width = (dst_width + 1) >> 1;
- int halfoheight = (dst_height + 1) >> 1;
- int aheight = dst_height - ooffset * 2; // actual output height
- const uint8* const iyptr = src;
- uint8* oyptr = dst + ooffset * dst_width;
- const uint8* const iuptr = src + src_width * src_height;
- uint8* ouptr = dst + dst_width * dst_height + (ooffset >> 1) * halfdst_width;
- const uint8* const ivptr = src + src_width * src_height +
- halfsrc_width * halfsrc_height;
- uint8* ovptr = dst + dst_width * dst_height + halfdst_width * halfoheight +
- (ooffset >> 1) * halfdst_width;
- return Scale_2(iyptr, iuptr, ivptr, src_width, halfsrc_width, halfsrc_width,
- src_width, src_height, oyptr, ouptr, ovptr, dst_width,
- halfdst_width, halfdst_width, dst_width, aheight, interpolate);
+ int src_halfwidth = (src_width + 1) >> 1;
+ int src_halfheight = (src_height + 1) >> 1;
+ int dst_halfwidth = (dst_width + 1) >> 1;
+ int dst_halfheight = (dst_height + 1) >> 1;
+ int aheight = dst_height - dst_yoffset * 2; // actual output height
+ const uint8* const src_y = src;
+ const uint8* const src_u = src + src_width * src_height;
+ const uint8* const src_v = src + src_width * src_height +
+ src_halfwidth * src_halfheight;
+ uint8* dst_y = dst + dst_yoffset * dst_width;
+ uint8* dst_u = dst + dst_width * dst_height +
+ (dst_yoffset >> 1) * dst_halfwidth;
+ uint8* dst_v = dst + dst_width * dst_height + dst_halfwidth * dst_halfheight +
+ (dst_yoffset >> 1) * dst_halfwidth;
+ return Scale(src_y, src_u, src_v, src_width, src_halfwidth, src_halfwidth,
+ src_width, src_height, dst_y, dst_u, dst_v, dst_width,
+ dst_halfwidth, dst_halfwidth, dst_width, aheight, interpolate);
}
}
-//} // namespace libyuv
+#ifdef __cplusplus
+} // extern "C"
+} // namespace libyuv
+#endif
diff --git a/vp8/common/arm/arm_systemdependent.c b/vp8/common/arm/arm_systemdependent.c
index cd55a63..89a2be8 100644
--- a/vp8/common/arm/arm_systemdependent.c
+++ b/vp8/common/arm/arm_systemdependent.c
@@ -11,7 +11,6 @@
#include "vpx_config.h"
#include "vpx_ports/arm.h"
-#include "vp8/common/g_common.h"
#include "vp8/common/pragmas.h"
#include "vp8/common/subpixel.h"
#include "vp8/common/loopfilter.h"
@@ -63,6 +62,12 @@
rtcd->recon.copy8x8 = vp8_copy_mem8x8_v6;
rtcd->recon.copy8x4 = vp8_copy_mem8x4_v6;
rtcd->recon.intra4x4_predict = vp8_intra4x4_predict_armv6;
+
+ rtcd->dequant.block = vp8_dequantize_b_v6;
+ rtcd->dequant.idct_add = vp8_dequant_idct_add_v6;
+ rtcd->dequant.idct_add_y_block = vp8_dequant_idct_add_y_block_v6;
+ rtcd->dequant.idct_add_uv_block = vp8_dequant_idct_add_uv_block_v6;
+
}
#endif
@@ -97,6 +102,12 @@
vp8_build_intra_predictors_mby_neon;
rtcd->recon.build_intra_predictors_mby_s =
vp8_build_intra_predictors_mby_s_neon;
+
+ rtcd->dequant.block = vp8_dequantize_b_neon;
+ rtcd->dequant.idct_add = vp8_dequant_idct_add_neon;
+ rtcd->dequant.idct_add_y_block = vp8_dequant_idct_add_y_block_neon;
+ rtcd->dequant.idct_add_uv_block = vp8_dequant_idct_add_uv_block_neon;
+
}
#endif
diff --git a/vp8/decoder/arm/armv6/dequant_idct_v6.asm b/vp8/common/arm/armv6/dequant_idct_v6.asm
similarity index 100%
rename from vp8/decoder/arm/armv6/dequant_idct_v6.asm
rename to vp8/common/arm/armv6/dequant_idct_v6.asm
diff --git a/vp8/decoder/arm/armv6/dequantize_v6.asm b/vp8/common/arm/armv6/dequantize_v6.asm
similarity index 100%
rename from vp8/decoder/arm/armv6/dequantize_v6.asm
rename to vp8/common/arm/armv6/dequantize_v6.asm
diff --git a/vp8/decoder/arm/armv6/idct_blk_v6.c b/vp8/common/arm/armv6/idct_blk_v6.c
similarity index 98%
rename from vp8/decoder/arm/armv6/idct_blk_v6.c
rename to vp8/common/arm/armv6/idct_blk_v6.c
index c1ef285..9108929 100644
--- a/vp8/decoder/arm/armv6/idct_blk_v6.c
+++ b/vp8/common/arm/armv6/idct_blk_v6.c
@@ -10,7 +10,7 @@
#include "vpx_config.h"
#include "vp8/common/idct.h"
-#include "vp8/decoder/dequantize.h"
+#include "vp8/common/dequantize.h"
void vp8_dequant_idct_add_y_block_v6(short *q, short *dq,
diff --git a/vp8/decoder/arm/dequantize_arm.c b/vp8/common/arm/dequantize_arm.c
similarity index 93%
rename from vp8/decoder/arm/dequantize_arm.c
rename to vp8/common/arm/dequantize_arm.c
index 2918e05..20a8ac4 100644
--- a/vp8/decoder/arm/dequantize_arm.c
+++ b/vp8/common/arm/dequantize_arm.c
@@ -10,9 +10,8 @@
#include "vpx_config.h"
-#include "vp8/decoder/dequantize.h"
+#include "vp8/common/dequantize.h"
#include "vp8/common/idct.h"
-#include "vpx_mem/vpx_mem.h"
#if HAVE_ARMV7
extern void vp8_dequantize_b_loop_neon(short *Q, short *DQC, short *DQ);
diff --git a/vp8/decoder/arm/dequantize_arm.h b/vp8/common/arm/dequantize_arm.h
similarity index 88%
rename from vp8/decoder/arm/dequantize_arm.h
rename to vp8/common/arm/dequantize_arm.h
index 1123e84..0b4d8fe 100644
--- a/vp8/decoder/arm/dequantize_arm.h
+++ b/vp8/common/arm/dequantize_arm.h
@@ -22,13 +22,13 @@
#undef vp8_dequant_block
#define vp8_dequant_block vp8_dequantize_b_v6
-#undef vp8_dequant_idct_add
+#undef vp8_dequant_idct_add
#define vp8_dequant_idct_add vp8_dequant_idct_add_v6
-#undef vp8_dequant_idct_add_y_block
+#undef vp8_dequant_idct_add_y_block
#define vp8_dequant_idct_add_y_block vp8_dequant_idct_add_y_block_v6
-#undef vp8_dequant_idct_add_uv_block
+#undef vp8_dequant_idct_add_uv_block
#define vp8_dequant_idct_add_uv_block vp8_dequant_idct_add_uv_block_v6
#endif
#endif
@@ -44,13 +44,13 @@
#undef vp8_dequant_block
#define vp8_dequant_block vp8_dequantize_b_neon
-#undef vp8_dequant_idct_add
+#undef vp8_dequant_idct_add
#define vp8_dequant_idct_add vp8_dequant_idct_add_neon
-#undef vp8_dequant_idct_add_y_block
+#undef vp8_dequant_idct_add_y_block
#define vp8_dequant_idct_add_y_block vp8_dequant_idct_add_y_block_neon
-#undef vp8_dequant_idct_add_uv_block
+#undef vp8_dequant_idct_add_uv_block
#define vp8_dequant_idct_add_uv_block vp8_dequant_idct_add_uv_block_neon
#endif
diff --git a/vp8/decoder/arm/neon/dequant_idct_neon.asm b/vp8/common/arm/neon/dequant_idct_neon.asm
similarity index 100%
rename from vp8/decoder/arm/neon/dequant_idct_neon.asm
rename to vp8/common/arm/neon/dequant_idct_neon.asm
diff --git a/vp8/decoder/arm/neon/dequantizeb_neon.asm b/vp8/common/arm/neon/dequantizeb_neon.asm
similarity index 100%
rename from vp8/decoder/arm/neon/dequantizeb_neon.asm
rename to vp8/common/arm/neon/dequantizeb_neon.asm
diff --git a/vp8/decoder/arm/neon/idct_blk_neon.c b/vp8/common/arm/neon/idct_blk_neon.c
similarity index 98%
rename from vp8/decoder/arm/neon/idct_blk_neon.c
rename to vp8/common/arm/neon/idct_blk_neon.c
index 185895f..cc55843 100644
--- a/vp8/decoder/arm/neon/idct_blk_neon.c
+++ b/vp8/common/arm/neon/idct_blk_neon.c
@@ -10,7 +10,7 @@
#include "vpx_config.h"
#include "vp8/common/idct.h"
-#include "vp8/decoder/dequantize.h"
+#include "vp8/common/dequantize.h"
/* place these declarations here because we don't want to maintain them
* outside of this scope
diff --git a/vp8/decoder/arm/neon/idct_dequant_0_2x_neon.asm b/vp8/common/arm/neon/idct_dequant_0_2x_neon.asm
similarity index 100%
rename from vp8/decoder/arm/neon/idct_dequant_0_2x_neon.asm
rename to vp8/common/arm/neon/idct_dequant_0_2x_neon.asm
diff --git a/vp8/decoder/arm/neon/idct_dequant_full_2x_neon.asm b/vp8/common/arm/neon/idct_dequant_full_2x_neon.asm
similarity index 100%
rename from vp8/decoder/arm/neon/idct_dequant_full_2x_neon.asm
rename to vp8/common/arm/neon/idct_dequant_full_2x_neon.asm
diff --git a/vp8/common/bigend.h b/vp8/common/bigend.h
deleted file mode 100644
index 6ac3f8b..0000000
--- a/vp8/common/bigend.h
+++ /dev/null
@@ -1,32 +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.
- */
-
-
-#ifndef _bigend_h
-#define _bigend_h
-
-#if defined(__cplusplus)
-extern "C" {
-#endif
-
-#define invert2(x) ( (((x)>>8)&0x00ff) | (((x)<<8)&0xff00) )
-#define invert4(x) ( ((invert2(x)&0x0000ffff)<<16) | (invert2((x>>16))&0x0000ffff) )
-
-#define high_byte(x) (unsigned char)x
-#define mid2Byte(x) (unsigned char)(x >> 8)
-#define mid1Byte(x) (unsigned char)(x >> 16)
-#define low_byte(x) (unsigned char)(x >> 24)
-
-#define SWAPENDS 1
-
-#if defined(__cplusplus)
-}
-#endif
-#endif
diff --git a/vp8/common/blockd.h b/vp8/common/blockd.h
index 586fa0a..1f3c5d1 100644
--- a/vp8/common/blockd.h
+++ b/vp8/common/blockd.h
@@ -21,9 +21,6 @@
#include "subpixel.h"
#include "vpx_ports/mem.h"
-#define TRUE 1
-#define FALSE 0
-
/*#define DCPRED 1*/
#define DCPREDSIMTHRESH 0
#define DCPREDCNTTHRESH 3
diff --git a/vp8/common/common.h b/vp8/common/common.h
index 9a93da9..2cc1c54 100644
--- a/vp8/common/common.h
+++ b/vp8/common/common.h
@@ -18,8 +18,6 @@
#include "vpx_mem/vpx_mem.h"
-#include "common_types.h"
-
/* Only need this for fixed-size arrays, for structs just assign. */
#define vp8_copy( Dest, Src) { \
diff --git a/vp8/common/common_types.h b/vp8/common/common_types.h
deleted file mode 100644
index 4e62486..0000000
--- a/vp8/common/common_types.h
+++ /dev/null
@@ -1,18 +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.
- */
-
-
-#ifndef __INC_COMMON_TYPES
-#define __INC_COMMON_TYPES
-
-#define TRUE 1
-#define FALSE 0
-
-#endif
diff --git a/vp8/decoder/dequantize.c b/vp8/common/dequantize.c
similarity index 100%
rename from vp8/decoder/dequantize.c
rename to vp8/common/dequantize.c
diff --git a/vp8/decoder/dequantize.h b/vp8/common/dequantize.h
similarity index 100%
rename from vp8/decoder/dequantize.h
rename to vp8/common/dequantize.h
diff --git a/vp8/common/dma_desc.h b/vp8/common/dma_desc.h
deleted file mode 100644
index b923da6..0000000
--- a/vp8/common/dma_desc.h
+++ /dev/null
@@ -1,125 +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.
- */
-
-
-#ifndef _dma_desc_h
-#define _dma_desc_h
-
-#if defined(__cplusplus)
-extern "C" {
-#endif
-
-
-#define NDSIZE_LG 0x00000900 // Next Descriptor Size
-#define NDSIZE_SM 0x00000800 // Next Descriptor Size
-#define NDSIZE_7 0x00000700 // Next Descriptor Size
-#define NDSIZE_6 0x00000600 // Next Descriptor Size
-#define NDSIZE_5 0x00000500 // Next Descriptor Size
-#define NDSIZE_4 0x00000400 // Next Descriptor Size
-#define NDSIZE_3 0x00000300 // Next Descriptor Size
-#define NDSIZE_2 0x00000200 // Next Descriptor Size
-#define NDSIZE_1 0x00000100 // Next Descriptor Size
-
-#define FLOW_STOP 0x0000
-#define FLOW_AUTO 0x1000
-#define FLOW_DESC_AR 0x4000
-#define FLOW_DESC_SM 0x6000
-#define FLOW_DESC_LG 0x7000
-
- typedef struct
- {
- unsigned int ndp;
- //unsigned short ndpl;
- //unsigned short ndph;
- unsigned int sa;
- //unsigned short sal;
- //unsigned short sah;
-
- unsigned short dmacfg;
- unsigned short xcnt;
- unsigned short xmod;
- unsigned short ycnt;
- unsigned short ymod;
-
- } LARGE_DESC;
-
- typedef struct
- {
- unsigned short ndpl;
- unsigned short sal;
- unsigned short sah;
- unsigned short dmacfg;
- unsigned short xcnt;
- unsigned short xmod;
- unsigned short ycnt;
- unsigned short ymod;
- } SMALL_DESC;
-
- typedef struct
- {
- unsigned short sal;
- unsigned short sah;
- unsigned short dmacfg;
- unsigned short xcnt;
- unsigned short xmod;
- unsigned short ycnt;
- unsigned short ymod;
- } ARRAY_DESC_7;
-
- typedef struct
- {
- unsigned short sal;
- unsigned short sah;
- unsigned short dmacfg;
- unsigned short xcnt;
- unsigned short xmod;
- unsigned short ycnt;
- } ARRAY_DESC_6;
-
- typedef struct
- {
- unsigned short sal;
- unsigned short sah;
- unsigned short dmacfg;
- unsigned short xcnt;
- unsigned short xmod;
- } ARRAY_DESC_5;
-
- typedef struct
- {
- unsigned short sal;
- unsigned short sah;
- unsigned short dmacfg;
- unsigned short xcnt;
- } ARRAY_DESC_4;
-
- typedef struct
- {
- unsigned short sal;
- unsigned short sah;
- unsigned short dmacfg;
- } ARRAY_DESC_3;
-
- typedef struct
- {
- unsigned short sal;
- unsigned short sah;
- } ARRAY_DESC_2;
-
- typedef struct
- {
- unsigned short sal;
- } ARRAY_DESC_1;
-
-#if defined(__cplusplus)
-}
-#endif
-
-#endif //_dma_desc_h
diff --git a/vp8/common/duck_io.h b/vp8/common/duck_io.h
deleted file mode 100644
index 43daa65..0000000
--- a/vp8/common/duck_io.h
+++ /dev/null
@@ -1,116 +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.
- */
-
-
-#ifndef _duck_io_h
-#define _duck_io_h
-
-#if defined(__cplusplus)
-extern "C" {
-#endif
-
-#if defined (_WIN32)
- typedef __int64 int64_t;
-#elif defined(__MWERKS__)
- typedef long long int64_t;
-#elif defined(__APPLE__) || defined(__POWERPC)
-#include <ppc/types.h>
-#else
- typedef long long int64_t;
-#endif
-
- typedef struct
- {
- int64_t offset; // offset to start from
- int blocking; // non-zero for blocking
- } re_open_t;
-
-
- typedef enum
- {
- SAL_ERR_MAX = -10,
- SAL_ERROR = -11, // Default error
- SAL_ERR_WSASTARTUP = -12,
- SAL_ERR_SOCKET_CREATE = -13,
- SAL_ERR_RESOLVING_HOSTNAME = -14,
- SAL_ERR_SERVER_CONNECTION = -15,
- SAL_ERR_SENDING_DATA = -16,
- SAL_ERR_RECEIVING_DATA = -17,
- SAL_ERR_404_FILE_NOT_FOUND = -18,
- SAL_ERR_PARSING_HTTP_HEADER = -19,
- SAL_ERR_PARSING_CONTENT_LEN = -20,
- SAL_ERR_CONNECTION_TIMEOUT = -21,
- SAL_ERR_FILE_OPEN_FAILED = -22,
- SAL_ERR_MIN = -23
- } SAL_ERR; /* EMH 1-15-03 */
-
-
- typedef struct sal_err_map_temp
- {
- SAL_ERR code;
- const char *decode;
-
- } sal_err_map_t;
-
-
- static char *sal_err_text(SAL_ERR e)
- {
- int t;
- const sal_err_map_t g_sal_err_map[] =
- {
- { SAL_ERR_WSASTARTUP, "Error with WSAStartup" },
- { SAL_ERR_SOCKET_CREATE, "Error creating socket" },
- { SAL_ERR_RESOLVING_HOSTNAME, "Error resolving hostname" },
- { SAL_ERR_SERVER_CONNECTION, "Error connecting to server" },
- { SAL_ERR_SENDING_DATA, "Error sending data" },
- { SAL_ERR_RECEIVING_DATA, "Error receiving data" },
- { SAL_ERR_404_FILE_NOT_FOUND, "Error file not found " },
- { SAL_ERR_PARSING_HTTP_HEADER, "Error parsing http header" },
- { SAL_ERR_PARSING_CONTENT_LEN, "Error parsing content length" },
- { SAL_ERR_CONNECTION_TIMEOUT, "Error Connection timed out" },
- { SAL_ERR_FILE_OPEN_FAILED, "Error opening file" }
- };
-
- for (t = 0; t < sizeof(g_sal_err_map) / sizeof(sal_err_map_t); t++)
- {
- if (e == g_sal_err_map[t].code)
- return (char *) g_sal_err_map[t].decode;
- }
-
- return 0;
- }
-
-
-
-
-
-
-
- int duck_open(const char *fname, unsigned long user_data);
-
- void duck_close(int ghndl);
-
- int duck_read(int ghndl, unsigned char *buf, int nbytes);
-
- int64_t duck_seek(int g_hndl, int64_t offs, int origin);
-
- int duck_read_finished(int han, int flag); /* FWG 7-9-99 */
-
- int duck_name(int handle, char name[], size_t max_len); /* EMH 9-23-03 */
-
- int duck_read_blocking(int handle, unsigned char *buffer, int bytes); /* EMH 9-23-03 */
-
- int64_t duck_available_data(int handle); /* EMH 10-23-03 */
-
-#if defined(__cplusplus)
-}
-#endif
-
-#endif
diff --git a/vp8/common/findnearmv.h b/vp8/common/findnearmv.h
index 01909b9..a3443d7 100644
--- a/vp8/common/findnearmv.h
+++ b/vp8/common/findnearmv.h
@@ -60,10 +60,10 @@
int mb_to_bottom_edge)
{
unsigned int need_to_clamp;
- need_to_clamp = (mv->as_mv.col < mb_to_left_edge) ? 1 : 0;
- need_to_clamp |= (mv->as_mv.col > mb_to_right_edge) ? 1 : 0;
- need_to_clamp |= (mv->as_mv.row < mb_to_top_edge) ? 1 : 0;
- need_to_clamp |= (mv->as_mv.row > mb_to_bottom_edge) ? 1 : 0;
+ need_to_clamp = (mv->as_mv.col < mb_to_left_edge);
+ need_to_clamp |= (mv->as_mv.col > mb_to_right_edge);
+ need_to_clamp |= (mv->as_mv.row < mb_to_top_edge);
+ need_to_clamp |= (mv->as_mv.row > mb_to_bottom_edge);
return need_to_clamp;
}
diff --git a/vp8/common/g_common.h b/vp8/common/g_common.h
deleted file mode 100644
index 5f52398..0000000
--- a/vp8/common/g_common.h
+++ /dev/null
@@ -1,21 +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.
- */
-
-
-extern void (*vp8_clear_system_state)(void);
-extern void (*vp8_plane_add_noise)(unsigned char *Start, unsigned int Width, unsigned int Height, int Pitch, int DPitch, int q);
-extern void (*de_interlace)
-(
- unsigned char *src_ptr,
- unsigned char *dst_ptr,
- int Width,
- int Height,
- int Stride
-);
diff --git a/vp8/common/generic/systemdependent.c b/vp8/common/generic/systemdependent.c
index 9641d8c..01d7620 100644
--- a/vp8/common/generic/systemdependent.c
+++ b/vp8/common/generic/systemdependent.c
@@ -10,7 +10,6 @@
#include "vpx_config.h"
-#include "vp8/common/g_common.h"
#include "vp8/common/subpixel.h"
#include "vp8/common/loopfilter.h"
#include "vp8/common/recon.h"
@@ -70,6 +69,14 @@
#if CONFIG_RUNTIME_CPU_DETECT
VP8_COMMON_RTCD *rtcd = &ctx->rtcd;
+
+ rtcd->dequant.block = vp8_dequantize_b_c;
+ rtcd->dequant.idct_add = vp8_dequant_idct_add_c;
+ rtcd->dequant.idct_add_y_block = vp8_dequant_idct_add_y_block_c;
+ rtcd->dequant.idct_add_uv_block =
+ vp8_dequant_idct_add_uv_block_c;
+
+
rtcd->idct.idct16 = vp8_short_idct4x4llm_c;
rtcd->idct.idct1_scalar_add = vp8_dc_only_idct_add_c;
rtcd->idct.iwalsh1 = vp8_short_inv_walsh4x4_1_c;
diff --git a/vp8/decoder/idct_blk.c b/vp8/common/idct_blk.c
similarity index 100%
rename from vp8/decoder/idct_blk.c
rename to vp8/common/idct_blk.c
diff --git a/vp8/common/invtrans.c b/vp8/common/invtrans.c
deleted file mode 100644
index 95e6980..0000000
--- a/vp8/common/invtrans.c
+++ /dev/null
@@ -1,56 +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 "invtrans.h"
-
-
-void vp8_inverse_transform_b(const vp8_idct_rtcd_vtable_t *rtcd, BLOCKD *b,
- int pitch)
-{
- if (*b->eob > 1)
- {
- IDCT_INVOKE(rtcd, idct16)(b->dqcoeff, b->predictor, pitch,
- *(b->base_dst) + b->dst, b->dst_stride);
- }
- else
- {
- IDCT_INVOKE(rtcd, idct1_scalar_add)(b->dqcoeff[0], b->predictor, pitch,
- *(b->base_dst) + b->dst, b->dst_stride);
- }
-
-}
-
-void vp8_inverse_transform_mby(const vp8_idct_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
-{
- int i;
-
- if(x->mode_info_context->mbmi.mode != SPLITMV)
- {
- /* do 2nd order transform on the dc block */
- IDCT_INVOKE(rtcd, iwalsh16)(x->block[24].dqcoeff, x->dqcoeff);
- }
-
- for (i = 0; i < 16; i++)
- {
- vp8_inverse_transform_b(rtcd, &x->block[i], 16);
- }
-
-}
-void vp8_inverse_transform_mbuv(const vp8_idct_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
-{
- int i;
-
- for (i = 16; i < 24; i++)
- {
- vp8_inverse_transform_b(rtcd, &x->block[i], 8);
- }
-
-}
diff --git a/vp8/common/invtrans.h b/vp8/common/invtrans.h
index d14573b..2bcbeec 100644
--- a/vp8/common/invtrans.h
+++ b/vp8/common/invtrans.h
@@ -15,9 +15,66 @@
#include "vpx_config.h"
#include "idct.h"
#include "blockd.h"
-extern void vp8_inverse_transform_b(const vp8_idct_rtcd_vtable_t *rtcd, BLOCKD *b, int pitch);
-extern void vp8_inverse_transform_mb(const vp8_idct_rtcd_vtable_t *rtcd, MACROBLOCKD *x);
-extern void vp8_inverse_transform_mby(const vp8_idct_rtcd_vtable_t *rtcd, MACROBLOCKD *x);
-extern void vp8_inverse_transform_mbuv(const vp8_idct_rtcd_vtable_t *rtcd, MACROBLOCKD *x);
+#include "onyxc_int.h"
+#if CONFIG_MULTITHREAD
+#include "vpx_mem/vpx_mem.h"
+#endif
+
+static void eob_adjust(char *eobs, short *diff)
+{
+ /* eob adjust.... the idct can only skip if both the dc and eob are zero */
+ int js;
+ for(js = 0; js < 16; js++)
+ {
+ if((eobs[js] == 0) && (diff[0] != 0))
+ eobs[js]++;
+ diff+=16;
+ }
+}
+
+static void vp8_inverse_transform_mby(MACROBLOCKD *xd,
+ const VP8_COMMON_RTCD *rtcd)
+{
+ short *DQC = xd->block[0].dequant;
+ /* save the dc dequant constant in case it is overridden */
+ short dc_dequant_temp = DQC[0];
+
+#if CONFIG_MULTITHREAD
+ DECLARE_ALIGNED(16, short, local_dequant[16]);
+#endif
+
+ if (xd->mode_info_context->mbmi.mode != SPLITMV)
+ {
+ /* do 2nd order transform on the dc block */
+ if (xd->eobs[24] > 1)
+ {
+ IDCT_INVOKE(&rtcd->idct, iwalsh16)
+ (&xd->block[24].dqcoeff[0], xd->qcoeff);
+ }
+ else
+ {
+ IDCT_INVOKE(&rtcd->idct, iwalsh1)
+ (&xd->block[24].dqcoeff[0], xd->qcoeff);
+ }
+ eob_adjust(xd->eobs, xd->qcoeff);
+
+#if CONFIG_MULTITHREAD
+ DQC = local_dequant;
+
+ vpx_memcpy(DQC, xd->block[0].dequant,
+ sizeof(local_dequant));
+#endif
+
+ /* override the dc dequant constant */
+ DQC[0] = 1;
+ }
+ DEQUANT_INVOKE (&rtcd->dequant, idct_add_y_block)
+ (xd->qcoeff, DQC,
+ xd->dst.y_buffer,
+ xd->dst.y_stride, xd->eobs);
+
+ /* restore the dc dequant constant */
+ DQC[0] = dc_dequant_temp;
+}
#endif
diff --git a/vp8/common/littlend.h b/vp8/common/littlend.h
deleted file mode 100644
index 99df116..0000000
--- a/vp8/common/littlend.h
+++ /dev/null
@@ -1,33 +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.
- */
-
-
-#ifndef _littlend_h
-#define _littlend_h
-
-#if defined(__cplusplus)
-extern "C" {
-#endif
-
-#define invert2(x) (x)
-#define invert4(x) (x)
-
-#define low_byte(x) (unsigned char)x
-#define mid1Byte(x) (unsigned char)(x >> 8)
-#define mid2Byte(x) (unsigned char)(x >> 16)
-#define high_byte(x) (unsigned char)(x >> 24)
-
-#define SWAPENDS 0
-
-#if defined(__cplusplus)
-}
-#endif
-
-#endif
diff --git a/vp8/common/onyx.h b/vp8/common/onyx.h
index 37fa5a0..d17a32b 100644
--- a/vp8/common/onyx.h
+++ b/vp8/common/onyx.h
@@ -22,9 +22,9 @@
#include "vpx/vp8cx.h"
#include "vpx/vpx_encoder.h"
#include "vpx_scale/yv12config.h"
-#include "type_aliases.h"
#include "ppflags.h"
- typedef int *VP8_PTR;
+
+ struct VP8_COMP;
/* Create/destroy static data structures. */
@@ -226,27 +226,27 @@
void vp8_initialize();
- VP8_PTR vp8_create_compressor(VP8_CONFIG *oxcf);
- void vp8_remove_compressor(VP8_PTR *comp);
+ struct VP8_COMP* vp8_create_compressor(VP8_CONFIG *oxcf);
+ void vp8_remove_compressor(struct VP8_COMP* *comp);
- void vp8_init_config(VP8_PTR onyx, VP8_CONFIG *oxcf);
- void vp8_change_config(VP8_PTR onyx, VP8_CONFIG *oxcf);
+ void vp8_init_config(struct VP8_COMP* onyx, VP8_CONFIG *oxcf);
+ void vp8_change_config(struct VP8_COMP* onyx, VP8_CONFIG *oxcf);
// receive a frames worth of data caller can assume that a copy of this frame is made
// and not just a copy of the pointer..
- int vp8_receive_raw_frame(VP8_PTR comp, unsigned int frame_flags, YV12_BUFFER_CONFIG *sd, int64_t time_stamp, int64_t end_time_stamp);
- int vp8_get_compressed_data(VP8_PTR comp, unsigned int *frame_flags, unsigned long *size, unsigned char *dest, unsigned char *dest_end, int64_t *time_stamp, int64_t *time_end, int flush);
- int vp8_get_preview_raw_frame(VP8_PTR comp, YV12_BUFFER_CONFIG *dest, vp8_ppflags_t *flags);
+ int vp8_receive_raw_frame(struct VP8_COMP* comp, unsigned int frame_flags, YV12_BUFFER_CONFIG *sd, int64_t time_stamp, int64_t end_time_stamp);
+ int vp8_get_compressed_data(struct VP8_COMP* comp, unsigned int *frame_flags, unsigned long *size, unsigned char *dest, unsigned char *dest_end, int64_t *time_stamp, int64_t *time_end, int flush);
+ int vp8_get_preview_raw_frame(struct VP8_COMP* comp, YV12_BUFFER_CONFIG *dest, vp8_ppflags_t *flags);
- int vp8_use_as_reference(VP8_PTR comp, int ref_frame_flags);
- int vp8_update_reference(VP8_PTR comp, int ref_frame_flags);
- int vp8_get_reference(VP8_PTR comp, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd);
- int vp8_set_reference(VP8_PTR comp, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd);
- int vp8_update_entropy(VP8_PTR comp, int update);
- int vp8_set_roimap(VP8_PTR comp, unsigned char *map, unsigned int rows, unsigned int cols, int delta_q[4], int delta_lf[4], unsigned int threshold[4]);
- int vp8_set_active_map(VP8_PTR comp, unsigned char *map, unsigned int rows, unsigned int cols);
- int vp8_set_internal_size(VP8_PTR comp, VPX_SCALING horiz_mode, VPX_SCALING vert_mode);
- int vp8_get_quantizer(VP8_PTR c);
+ int vp8_use_as_reference(struct VP8_COMP* comp, int ref_frame_flags);
+ int vp8_update_reference(struct VP8_COMP* comp, int ref_frame_flags);
+ int vp8_get_reference(struct VP8_COMP* comp, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd);
+ int vp8_set_reference(struct VP8_COMP* comp, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd);
+ int vp8_update_entropy(struct VP8_COMP* comp, int update);
+ int vp8_set_roimap(struct VP8_COMP* comp, unsigned char *map, unsigned int rows, unsigned int cols, int delta_q[4], int delta_lf[4], unsigned int threshold[4]);
+ int vp8_set_active_map(struct VP8_COMP* comp, unsigned char *map, unsigned int rows, unsigned int cols);
+ int vp8_set_internal_size(struct VP8_COMP* comp, VPX_SCALING horiz_mode, VPX_SCALING vert_mode);
+ int vp8_get_quantizer(struct VP8_COMP* c);
#ifdef __cplusplus
}
diff --git a/vp8/common/onyxc_int.h b/vp8/common/onyxc_int.h
index 936fa9f..f733ff7 100644
--- a/vp8/common/onyxc_int.h
+++ b/vp8/common/onyxc_int.h
@@ -22,6 +22,7 @@
#if CONFIG_POSTPROC
#include "postproc.h"
#endif
+#include "dequantize.h"
/*#ifdef PACKET_TESTING*/
#include "header.h"
@@ -73,6 +74,7 @@
typedef struct VP8_COMMON_RTCD
{
#if CONFIG_RUNTIME_CPU_DETECT
+ vp8_dequant_rtcd_vtable_t dequant;
vp8_idct_rtcd_vtable_t idct;
vp8_recon_rtcd_vtable_t recon;
vp8_subpix_rtcd_vtable_t subpix;
diff --git a/vp8/common/onyxd.h b/vp8/common/onyxd.h
index 43fa00b..35a8b6e 100644
--- a/vp8/common/onyxd.h
+++ b/vp8/common/onyxd.h
@@ -18,13 +18,13 @@
extern "C"
{
#endif
-#include "type_aliases.h"
#include "vpx_scale/yv12config.h"
#include "ppflags.h"
#include "vpx_ports/mem.h"
#include "vpx/vpx_codec.h"
- typedef void *VP8D_PTR;
+ struct VP8D_COMP;
+
typedef struct
{
int Width;
@@ -49,19 +49,19 @@
void vp8dx_initialize(void);
- void vp8dx_set_setting(VP8D_PTR comp, VP8D_SETTING oxst, int x);
+ void vp8dx_set_setting(struct VP8D_COMP* comp, VP8D_SETTING oxst, int x);
- int vp8dx_get_setting(VP8D_PTR comp, VP8D_SETTING oxst);
+ int vp8dx_get_setting(struct VP8D_COMP* comp, VP8D_SETTING oxst);
- int vp8dx_receive_compressed_data(VP8D_PTR comp, unsigned long size, const unsigned char *dest, int64_t time_stamp);
- int vp8dx_get_raw_frame(VP8D_PTR comp, YV12_BUFFER_CONFIG *sd, int64_t *time_stamp, int64_t *time_end_stamp, vp8_ppflags_t *flags);
+ int vp8dx_receive_compressed_data(struct VP8D_COMP* comp, unsigned long size, const unsigned char *dest, int64_t time_stamp);
+ int vp8dx_get_raw_frame(struct VP8D_COMP* comp, YV12_BUFFER_CONFIG *sd, int64_t *time_stamp, int64_t *time_end_stamp, vp8_ppflags_t *flags);
- vpx_codec_err_t vp8dx_get_reference(VP8D_PTR comp, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd);
- vpx_codec_err_t vp8dx_set_reference(VP8D_PTR comp, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd);
+ vpx_codec_err_t vp8dx_get_reference(struct VP8D_COMP* comp, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd);
+ vpx_codec_err_t vp8dx_set_reference(struct VP8D_COMP* comp, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd);
- VP8D_PTR vp8dx_create_decompressor(VP8D_CONFIG *oxcf);
+ struct VP8D_COMP* vp8dx_create_decompressor(VP8D_CONFIG *oxcf);
- void vp8dx_remove_decompressor(VP8D_PTR comp);
+ void vp8dx_remove_decompressor(struct VP8D_COMP* comp);
#ifdef __cplusplus
}
diff --git a/vp8/common/ppc/systemdependent.c b/vp8/common/ppc/systemdependent.c
index 1f5d790..7046a63 100644
--- a/vp8/common/ppc/systemdependent.c
+++ b/vp8/common/ppc/systemdependent.c
@@ -9,7 +9,6 @@
*/
-#include "g_common.h"
#include "subpixel.h"
#include "loopfilter.h"
#include "recon.h"
diff --git a/vp8/common/reconinter.c b/vp8/common/reconinter.c
index 24c09a3..6c7af41 100644
--- a/vp8/common/reconinter.c
+++ b/vp8/common/reconinter.c
@@ -334,11 +334,12 @@
/*encoder only*/
-void vp8_build_inter16x16_predictors_mby(MACROBLOCKD *x)
+void vp8_build_inter16x16_predictors_mby(MACROBLOCKD *x,
+ unsigned char *dst_y,
+ int dst_ystride)
{
unsigned char *ptr_base;
unsigned char *ptr;
- unsigned char *pred_ptr = x->predictor;
int mv_row = x->mode_info_context->mbmi.mv.as_mv.row;
int mv_col = x->mode_info_context->mbmi.mv.as_mv.col;
int pre_stride = x->block[0].pre_stride;
@@ -348,11 +349,13 @@
if ((mv_row | mv_col) & 7)
{
- x->subpixel_predict16x16(ptr, pre_stride, mv_col & 7, mv_row & 7, pred_ptr, 16);
+ x->subpixel_predict16x16(ptr, pre_stride, mv_col & 7, mv_row & 7,
+ dst_y, dst_ystride);
}
else
{
- RECON_INVOKE(&x->rtcd->recon, copy16x16)(ptr, pre_stride, pred_ptr, 16);
+ RECON_INVOKE(&x->rtcd->recon, copy16x16)(ptr, pre_stride, dst_y,
+ dst_ystride);
}
}
@@ -596,69 +599,3 @@
build_inter4x4_predictors_mb(xd);
}
}
-/* encoder only*/
-static void build_inter4x4_predictors_mb_e(MACROBLOCKD *x)
-{
- int i;
-
- if (x->mode_info_context->mbmi.partitioning < 3)
- {
- x->block[ 0].bmi = x->mode_info_context->bmi[ 0];
- x->block[ 2].bmi = x->mode_info_context->bmi[ 2];
- x->block[ 8].bmi = x->mode_info_context->bmi[ 8];
- x->block[10].bmi = x->mode_info_context->bmi[10];
-
- build_inter_predictors4b(x, &x->block[ 0], x->block[ 0].predictor, 16);
- build_inter_predictors4b(x, &x->block[ 2], x->block[ 2].predictor, 16);
- build_inter_predictors4b(x, &x->block[ 8], x->block[ 8].predictor, 16);
- build_inter_predictors4b(x, &x->block[10], x->block[10].predictor, 16);
- }
- else
- {
- for (i = 0; i < 16; i += 2)
- {
- BLOCKD *d0 = &x->block[i];
- BLOCKD *d1 = &x->block[i+1];
-
- x->block[i+0].bmi = x->mode_info_context->bmi[i+0];
- x->block[i+1].bmi = x->mode_info_context->bmi[i+1];
-
- if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
- build_inter_predictors2b(x, d0, d0->predictor, 16);
- else
- {
- build_inter_predictors_b(d0, d0->predictor, 16, x->subpixel_predict);
- build_inter_predictors_b(d1, d1->predictor, 16, x->subpixel_predict);
- }
-
- }
-
- }
-
- for (i = 16; i < 24; i += 2)
- {
- BLOCKD *d0 = &x->block[i];
- BLOCKD *d1 = &x->block[i+1];
-
- if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
- build_inter_predictors2b(x, d0, d0->predictor, 8);
- else
- {
- build_inter_predictors_b(d0, d0->predictor, 8, x->subpixel_predict);
- build_inter_predictors_b(d1, d1->predictor, 8, x->subpixel_predict);
- }
- }
-}
-void vp8_build_inter_predictors_mb_e(MACROBLOCKD *xd)
-{
- if (xd->mode_info_context->mbmi.mode != SPLITMV)
- {
- vp8_build_inter16x16_predictors_mb(xd, xd->predictor, &xd->predictor[256],
- &xd->predictor[320], 16, 8);
- }
- else
- {
- build_4x4uvmvs(xd);
- build_inter4x4_predictors_mb_e(xd);
- }
-}
diff --git a/vp8/common/reconinter.h b/vp8/common/reconinter.h
index 86f9d5a..f57ff73 100644
--- a/vp8/common/reconinter.h
+++ b/vp8/common/reconinter.h
@@ -21,11 +21,13 @@
int dst_uvstride);
-extern void vp8_build_inter16x16_predictors_mby(MACROBLOCKD *x);
-extern void vp8_build_inter_predictors_b(BLOCKD *d, int pitch, vp8_subpix_fn_t sppf);
+extern void vp8_build_inter16x16_predictors_mby(MACROBLOCKD *x,
+ unsigned char *dst_y,
+ int dst_ystride);
+extern void vp8_build_inter_predictors_b(BLOCKD *d, int pitch,
+ vp8_subpix_fn_t sppf);
extern void vp8_build_inter16x16_predictors_mbuv(MACROBLOCKD *x);
extern void vp8_build_inter4x4_predictors_mbuv(MACROBLOCKD *x);
-extern void vp8_build_inter_predictors_mb_e(MACROBLOCKD *xd);
#endif
diff --git a/vp8/common/type_aliases.h b/vp8/common/type_aliases.h
deleted file mode 100644
index 22b531a..0000000
--- a/vp8/common/type_aliases.h
+++ /dev/null
@@ -1,117 +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.
- */
-
-
-/****************************************************************************
-*
-* Module Title : type_aliases.h
-*
-* Description : Standard type aliases
-*
-****************************************************************************/
-#ifndef __INC_TYPE_ALIASES_H
-#define __INC_TYPE_ALIASES_H
-
-/****************************************************************************
-* Macros
-****************************************************************************/
-#define EXPORT
-#define IMPORT extern /* Used to declare imported data & routines */
-#define PRIVATE static /* Used to declare & define module-local data */
-#define LOCAL static /* Used to define all persistent routine-local data */
-#define STD_IN_PATH 0 /* Standard input path */
-#define STD_OUT_PATH 1 /* Standard output path */
-#define STD_ERR_PATH 2 /* Standard error path */
-#define STD_IN_FILE stdin /* Standard input file pointer */
-#define STD_OUT_FILE stdout /* Standard output file pointer */
-#define STD_ERR_FILE stderr /* Standard error file pointer */
-#define max_int 0x7FFFFFFF
-
-#define __export
-#define _export
-
-#define CCONV
-
-#ifndef NULL
-#ifdef __cplusplus
-#define NULL 0
-#else
-#define NULL ((void *)0)
-#endif
-#endif
-
-#ifndef FALSE
-#define FALSE 0
-#endif
-
-#ifndef TRUE
-#define TRUE 1
-#endif
-
-/****************************************************************************
-* Typedefs
-****************************************************************************/
-#ifndef TYPE_INT8
-#define TYPE_INT8
-typedef signed char INT8;
-#endif
-
-#ifndef TYPE_INT16
-/*#define TYPE_INT16*/
-typedef signed short INT16;
-#endif
-
-#ifndef TYPE_INT32
-/*#define TYPE_INT32*/
-typedef signed int INT32;
-#endif
-
-#ifndef TYPE_UINT8
-/*#define TYPE_UINT8*/
-typedef unsigned char UINT8;
-#endif
-
-#ifndef TYPE_UINT32
-/*#define TYPE_UINT32*/
-typedef unsigned int UINT32;
-#endif
-
-#ifndef TYPE_UINT16
-/*#define TYPE_UINT16*/
-typedef unsigned short UINT16;
-#endif
-
-#ifndef TYPE_BOOL
-/*#define TYPE_BOOL*/
-typedef int BOOL;
-#endif
-
-typedef unsigned char BOOLEAN;
-
-#ifdef _MSC_VER
-typedef __int64 INT64;
-#else
-
-#ifndef TYPE_INT64
-#ifdef _TMS320C6X
-/* for now we only have 40bits */
-typedef long INT64;
-#else
-typedef long long INT64;
-#endif
-#endif
-
-#endif
-
-/* Floating point */
-typedef double FLOAT64;
-typedef float FLOAT32;
-
-#endif
diff --git a/vp8/decoder/x86/dequantize_mmx.asm b/vp8/common/x86/dequantize_mmx.asm
similarity index 100%
rename from vp8/decoder/x86/dequantize_mmx.asm
rename to vp8/common/x86/dequantize_mmx.asm
diff --git a/vp8/decoder/x86/dequantize_x86.h b/vp8/common/x86/dequantize_x86.h
similarity index 100%
rename from vp8/decoder/x86/dequantize_x86.h
rename to vp8/common/x86/dequantize_x86.h
diff --git a/vp8/decoder/x86/idct_blk_mmx.c b/vp8/common/x86/idct_blk_mmx.c
similarity index 91%
rename from vp8/decoder/x86/idct_blk_mmx.c
rename to vp8/common/x86/idct_blk_mmx.c
index 29276e5..49cebd6 100644
--- a/vp8/decoder/x86/idct_blk_mmx.c
+++ b/vp8/common/x86/idct_blk_mmx.c
@@ -10,7 +10,17 @@
#include "vpx_config.h"
#include "vp8/common/idct.h"
-#include "vp8/decoder/dequantize.h"
+#include "vp8/common/dequantize.h"
+
+extern void vp8_dequantize_b_impl_mmx(short *sq, short *dq, short *q);
+
+void vp8_dequantize_b_mmx(BLOCKD *d)
+{
+ short *sq = (short *) d->qcoeff;
+ short *dq = (short *) d->dqcoeff;
+ short *q = (short *) d->dequant;
+ vp8_dequantize_b_impl_mmx(sq, dq, q);
+}
void vp8_dequant_idct_add_y_block_mmx
(short *q, short *dq,
diff --git a/vp8/decoder/x86/idct_blk_sse2.c b/vp8/common/x86/idct_blk_sse2.c
similarity index 98%
rename from vp8/decoder/x86/idct_blk_sse2.c
rename to vp8/common/x86/idct_blk_sse2.c
index 03c2878..44e440c 100644
--- a/vp8/decoder/x86/idct_blk_sse2.c
+++ b/vp8/common/x86/idct_blk_sse2.c
@@ -10,7 +10,7 @@
#include "vpx_config.h"
#include "vp8/common/idct.h"
-#include "vp8/decoder/dequantize.h"
+#include "vp8/common/dequantize.h"
void vp8_idct_dequant_0_2x_sse2
(short *q, short *dq ,
diff --git a/vp8/common/x86/loopfilter_sse2.asm b/vp8/common/x86/loopfilter_sse2.asm
index 86927d9..2ad010a 100644
--- a/vp8/common/x86/loopfilter_sse2.asm
+++ b/vp8/common/x86/loopfilter_sse2.asm
@@ -1385,52 +1385,54 @@
SHADOW_ARGS_TO_STACK 3
SAVE_XMM 7
GET_GOT rbx
- push rsi
- push rdi
; end prolog
- mov rsi, arg(0) ;src_ptr
+ mov rcx, arg(0) ;src_ptr
movsxd rax, dword ptr arg(1) ;src_pixel_step ; destination pitch?
- mov rdx, arg(2) ;blimit
- movdqa xmm3, XMMWORD PTR [rdx]
- mov rdi, rsi ; rdi points to row +1 for indirect addressing
- add rdi, rax
+ lea rdx, [rcx + rax]
neg rax
; calculate mask
- movdqa xmm1, [rsi+2*rax] ; p1
- movdqa xmm0, [rdi] ; q1
+ movdqa xmm0, [rdx] ; q1
+ mov rdx, arg(2) ;blimit
+ movdqa xmm1, [rcx+2*rax] ; p1
+
movdqa xmm2, xmm1
movdqa xmm7, xmm0
- movdqa xmm4, xmm0
+
psubusb xmm0, xmm1 ; q1-=p1
- psubusb xmm1, xmm4 ; p1-=q1
+ psubusb xmm1, xmm7 ; p1-=q1
por xmm1, xmm0 ; abs(p1-q1)
pand xmm1, [GLOBAL(tfe)] ; set lsb of each byte to zero
psrlw xmm1, 1 ; abs(p1-q1)/2
- movdqa xmm5, [rsi+rax] ; p0
- movdqa xmm4, [rsi] ; q0
+ movdqa xmm3, XMMWORD PTR [rdx]
+
+ movdqa xmm5, [rcx+rax] ; p0
+ movdqa xmm4, [rcx] ; q0
movdqa xmm0, xmm4 ; q0
movdqa xmm6, xmm5 ; p0
psubusb xmm5, xmm4 ; p0-=q0
psubusb xmm4, xmm6 ; q0-=p0
por xmm5, xmm4 ; abs(p0 - q0)
+
+ movdqa xmm4, [GLOBAL(t80)]
+
paddusb xmm5, xmm5 ; abs(p0-q0)*2
paddusb xmm5, xmm1 ; abs (p0 - q0) *2 + abs(p1-q1)/2
-
psubusb xmm5, xmm3 ; abs(p0 - q0) *2 + abs(p1-q1)/2 > blimit
pxor xmm3, xmm3
pcmpeqb xmm5, xmm3
+
; start work on filters
- pxor xmm2, [GLOBAL(t80)] ; p1 offset to convert to signed values
- pxor xmm7, [GLOBAL(t80)] ; q1 offset to convert to signed values
+ pxor xmm2, xmm4 ; p1 offset to convert to signed values
+ pxor xmm7, xmm4 ; q1 offset to convert to signed values
psubsb xmm2, xmm7 ; p1 - q1
- pxor xmm6, [GLOBAL(t80)] ; offset to convert to signed values
- pxor xmm0, [GLOBAL(t80)] ; offset to convert to signed values
+ pxor xmm6, xmm4 ; offset to convert to signed values
+ pxor xmm0, xmm4 ; offset to convert to signed values
movdqa xmm3, xmm0 ; q0
psubsb xmm0, xmm6 ; q0 - p0
paddsb xmm2, xmm0 ; p1 - q1 + 1 * (q0 - p0)
@@ -1438,42 +1440,36 @@
paddsb xmm2, xmm0 ; p1 - q1 + 3 * (q0 - p0)
pand xmm5, xmm2 ; mask filter values we don't care about
- ; do + 4 side
- paddsb xmm5, [GLOBAL(t4)] ; 3* (q0 - p0) + (p1 - q1) + 4
-
- movdqa xmm0, xmm5 ; get a copy of filters
- psllw xmm0, 8 ; shift left 8
- psraw xmm0, 3 ; arithmetic shift right 11
- psrlw xmm0, 8
- movdqa xmm1, xmm5 ; get a copy of filters
- psraw xmm1, 11 ; arithmetic shift right 11
- psllw xmm1, 8 ; shift left 8 to put it back
-
- por xmm0, xmm1 ; put the two together to get result
-
- psubsb xmm3, xmm0 ; q0-= q0 add
- pxor xmm3, [GLOBAL(t80)] ; unoffset
- movdqa [rsi], xmm3 ; write back
-
- ; now do +3 side
+ paddsb xmm5, [GLOBAL(t4)] ; 3* (q0 - p0) + (p1 - q1) + 4
+ movdqa xmm0, xmm5
psubsb xmm5, [GLOBAL(t1s)] ; +3 instead of +4
- movdqa xmm0, xmm5 ; get a copy of filters
- psllw xmm0, 8 ; shift left 8
- psraw xmm0, 3 ; arithmetic shift right 11
- psrlw xmm0, 8
- psraw xmm5, 11 ; arithmetic shift right 11
- psllw xmm5, 8 ; shift left 8 to put it back
- por xmm0, xmm5 ; put the two together to get result
+ movdqa xmm1, [GLOBAL(te0)]
+ movdqa xmm2, [GLOBAL(t1f)]
+ pxor xmm7, xmm7
+ pcmpgtb xmm7, xmm0 ;save sign
+ pand xmm7, xmm1 ;preserve the upper 3 bits
+ psrlw xmm0, 3
+ pand xmm0, xmm2 ;clear out upper 3 bits
+ por xmm0, xmm7 ;add sign
+ psubsb xmm3, xmm0 ; q0-= q0sz add
- paddsb xmm6, xmm0 ; p0+= p0 add
- pxor xmm6, [GLOBAL(t80)] ; unoffset
- movdqa [rsi+rax], xmm6 ; write back
+ pxor xmm7, xmm7
+ pcmpgtb xmm7, xmm5 ;save sign
+ pand xmm7, xmm1 ;preserve the upper 3 bits
+ psrlw xmm5, 3
+ pand xmm5, xmm2 ;clear out upper 3 bits
+ por xmm5, xmm7 ;add sign
+ paddsb xmm6, xmm5 ; p0+= p0 add
+
+ pxor xmm3, xmm4 ; unoffset
+ movdqa [rcx], xmm3 ; write back
+
+ pxor xmm6, xmm4 ; unoffset
+ movdqa [rcx+rax], xmm6 ; write back
; begin epilog
- pop rdi
- pop rsi
RESTORE_GOT
RESTORE_XMM
UNSHADOW_ARGS
@@ -1536,9 +1532,6 @@
punpckldq xmm0, xmm1 ; 71 61 51 41 31 21 11 01 70 60 50 40 30 20 10 00
punpckhdq xmm2, xmm1 ; 73 63 53 43 33 23 13 03 72 62 52 42 32 22 12 02
- movdqa t0, xmm0 ; save to t0
- movdqa t1, xmm2 ; save to t1
-
lea rsi, [rsi + rax*8]
lea rdi, [rsi + rax]
lea rdx, [rsi + rax*4]
@@ -1551,26 +1544,24 @@
punpckldq xmm4, xmm1 ; c3 c2 c1 c0 83 82 81 80
punpckldq xmm6, xmm3 ; d3 d2 d1 d0 93 92 91 90
- movd xmm0, [rsi + rax*2] ; a3 a2 a1 a0
+ movd xmm1, [rsi + rax*2] ; a3 a2 a1 a0
movd xmm5, [rdx + rax*2] ; e3 e2 e1 e0
- movd xmm2, [rdi + rax*2] ; b3 b2 b1 b0
+ movd xmm3, [rdi + rax*2] ; b3 b2 b1 b0
movd xmm7, [rcx + rax*2] ; f3 f2 f1 f0
- punpckldq xmm0, xmm5 ; e3 e2 e1 e0 a3 a2 a1 a0
- punpckldq xmm2, xmm7 ; f3 f2 f1 f0 b3 b2 b1 b0
+ punpckldq xmm1, xmm5 ; e3 e2 e1 e0 a3 a2 a1 a0
+ punpckldq xmm3, xmm7 ; f3 f2 f1 f0 b3 b2 b1 b0
punpcklbw xmm4, xmm6 ; d3 c3 d2 c2 d1 c1 d0 c0 93 83 92 82 91 81 90 80
- punpcklbw xmm0, xmm2 ; f3 e3 f2 e2 f1 e1 f0 e0 b3 a3 b2 a2 b1 a1 b0 a0
+ punpcklbw xmm1, xmm3 ; f3 e3 f2 e2 f1 e1 f0 e0 b3 a3 b2 a2 b1 a1 b0 a0
- movdqa xmm1, xmm4
- punpcklwd xmm4, xmm0 ; b3 a3 93 83 b2 a2 92 82 b1 a1 91 81 b0 a0 90 80
- punpckhwd xmm1, xmm0 ; f3 e3 d3 c3 f2 e2 d2 c2 f1 e1 d1 c1 f0 e0 d0 c0
+ movdqa xmm7, xmm4
+ punpcklwd xmm4, xmm1 ; b3 a3 93 83 b2 a2 92 82 b1 a1 91 81 b0 a0 90 80
+ punpckhwd xmm7, xmm1 ; f3 e3 d3 c3 f2 e2 d2 c2 f1 e1 d1 c1 f0 e0 d0 c0
movdqa xmm6, xmm4
- punpckldq xmm4, xmm1 ; f1 e1 d1 c1 b1 a1 91 81 f0 e0 d0 c0 b0 a0 90 80
- punpckhdq xmm6, xmm1 ; f3 e3 d3 c3 b3 a3 93 83 f2 e2 d2 c2 b2 a2 92 82
+ punpckldq xmm4, xmm7 ; f1 e1 d1 c1 b1 a1 91 81 f0 e0 d0 c0 b0 a0 90 80
+ punpckhdq xmm6, xmm7 ; f3 e3 d3 c3 b3 a3 93 83 f2 e2 d2 c2 b2 a2 92 82
- movdqa xmm0, t0 ; 71 61 51 41 31 21 11 01 70 60 50 40 30 20 10 00
- movdqa xmm2, t1 ; 73 63 53 43 33 23 13 03 72 62 52 42 32 22 12 02
movdqa xmm1, xmm0
movdqa xmm3, xmm2
@@ -1579,6 +1570,8 @@
punpcklqdq xmm2, xmm6 ; q0 f2 e2 d2 c2 b2 a2 92 82 72 62 52 42 32 22 12 02
punpckhqdq xmm3, xmm6 ; q1 f3 e3 d3 c3 b3 a3 93 83 73 63 53 43 33 23 13 03
+ mov rdx, arg(2) ;blimit
+
; calculate mask
movdqa xmm6, xmm0 ; p1
movdqa xmm7, xmm3 ; q1
@@ -1588,6 +1581,8 @@
pand xmm6, [GLOBAL(tfe)] ; set lsb of each byte to zero
psrlw xmm6, 1 ; abs(p1-q1)/2
+ movdqa xmm7, [rdx]
+
movdqa xmm5, xmm1 ; p0
movdqa xmm4, xmm2 ; q0
psubusb xmm5, xmm2 ; p0-=q0
@@ -1596,8 +1591,7 @@
paddusb xmm5, xmm5 ; abs(p0-q0)*2
paddusb xmm5, xmm6 ; abs (p0 - q0) *2 + abs(p1-q1)/2
- mov rdx, arg(2) ;blimit
- movdqa xmm7, XMMWORD PTR [rdx]
+ movdqa xmm4, [GLOBAL(t80)]
psubusb xmm5, xmm7 ; abs(p0 - q0) *2 + abs(p1-q1)/2 > blimit
pxor xmm7, xmm7
@@ -1607,59 +1601,48 @@
movdqa t0, xmm0
movdqa t1, xmm3
- pxor xmm0, [GLOBAL(t80)] ; p1 offset to convert to signed values
- pxor xmm3, [GLOBAL(t80)] ; q1 offset to convert to signed values
-
+ pxor xmm0, xmm4 ; p1 offset to convert to signed values
+ pxor xmm3, xmm4 ; q1 offset to convert to signed values
psubsb xmm0, xmm3 ; p1 - q1
+
movdqa xmm6, xmm1 ; p0
+; movdqa xmm7, xmm2 ; q0
- movdqa xmm7, xmm2 ; q0
- pxor xmm6, [GLOBAL(t80)] ; offset to convert to signed values
+ pxor xmm6, xmm4 ; offset to convert to signed values
+ pxor xmm2, xmm4 ; offset to convert to signed values
- pxor xmm7, [GLOBAL(t80)] ; offset to convert to signed values
- movdqa xmm3, xmm7 ; offseted ; q0
-
- psubsb xmm7, xmm6 ; q0 - p0
- paddsb xmm0, xmm7 ; p1 - q1 + 1 * (q0 - p0)
-
- paddsb xmm0, xmm7 ; p1 - q1 + 2 * (q0 - p0)
- paddsb xmm0, xmm7 ; p1 - q1 + 3 * (q0 - p0)
-
+ movdqa xmm3, xmm2 ; offseted ; q0
+ psubsb xmm2, xmm6 ; q0 - p0
+ paddsb xmm0, xmm2 ; p1 - q1 + 1 * (q0 - p0)
+ paddsb xmm0, xmm2 ; p1 - q1 + 2 * (q0 - p0)
+ paddsb xmm0, xmm2 ; p1 - q1 + 3 * (q0 - p0)
pand xmm5, xmm0 ; mask filter values we don't care about
-
paddsb xmm5, [GLOBAL(t4)] ; 3* (q0 - p0) + (p1 - q1) + 4
-
- movdqa xmm0, xmm5 ; get a copy of filters
- psllw xmm0, 8 ; shift left 8
-
- psraw xmm0, 3 ; arithmetic shift right 11
- psrlw xmm0, 8
-
- movdqa xmm7, xmm5 ; get a copy of filters
- psraw xmm7, 11 ; arithmetic shift right 11
-
- psllw xmm7, 8 ; shift left 8 to put it back
- por xmm0, xmm7 ; put the two together to get result
-
- psubsb xmm3, xmm0 ; q0-= q0sz add
- pxor xmm3, [GLOBAL(t80)] ; unoffset q0
-
- ; now do +3 side
+ movdqa xmm0, xmm5
psubsb xmm5, [GLOBAL(t1s)] ; +3 instead of +4
- movdqa xmm0, xmm5 ; get a copy of filters
- psllw xmm0, 8 ; shift left 8
- psraw xmm0, 3 ; arithmetic shift right 11
+ movdqa xmm1, [GLOBAL(te0)]
+ movdqa xmm2, [GLOBAL(t1f)]
- psrlw xmm0, 8
- psraw xmm5, 11 ; arithmetic shift right 11
+ pxor xmm7, xmm7
+ pcmpgtb xmm7, xmm0 ;save sign
+ pand xmm7, xmm1 ;preserve the upper 3 bits
+ psrlw xmm0, 3
+ pand xmm0, xmm2 ;clear out upper 3 bits
+ por xmm0, xmm7 ;add sign
+ psubsb xmm3, xmm0 ; q0-= q0sz add
- psllw xmm5, 8 ; shift left 8 to put it back
- por xmm0, xmm5 ; put the two together to get result
+ pxor xmm7, xmm7
+ pcmpgtb xmm7, xmm5 ;save sign
+ pand xmm7, xmm1 ;preserve the upper 3 bits
+ psrlw xmm5, 3
+ pand xmm5, xmm2 ;clear out upper 3 bits
+ por xmm5, xmm7 ;add sign
+ paddsb xmm6, xmm5 ; p0+= p0 add
- paddsb xmm6, xmm0 ; p0+= p0 add
- pxor xmm6, [GLOBAL(t80)] ; unoffset p0
+ pxor xmm3, xmm4 ; unoffset q0
+ pxor xmm6, xmm4 ; unoffset p0
movdqa xmm0, t0 ; p1
movdqa xmm4, t1 ; q1
@@ -1763,3 +1746,9 @@
align 16
s63:
times 8 dw 0x003f
+align 16
+te0:
+ times 16 db 0xe0
+align 16
+t1f:
+ times 16 db 0x1f
diff --git a/vp8/common/x86/recon_sse2.asm b/vp8/common/x86/recon_sse2.asm
index a82c1b4..4b68ef5 100644
--- a/vp8/common/x86/recon_sse2.asm
+++ b/vp8/common/x86/recon_sse2.asm
@@ -559,12 +559,492 @@
vp8_intra_pred_uv_ho mmx2
vp8_intra_pred_uv_ho ssse3
+;void vp8_intra_pred_y_dc_sse2(
+; unsigned char *dst,
+; int dst_stride
+; unsigned char *src,
+; int src_stride,
+; )
+global sym(vp8_intra_pred_y_dc_sse2)
+sym(vp8_intra_pred_y_dc_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 4
+ push rsi
+ push rdi
+ ; end prolog
+
+ ; from top
+ mov rsi, arg(2) ;src;
+ movsxd rax, dword ptr arg(3) ;src_stride;
+ sub rsi, rax
+ pxor xmm0, xmm0
+ movdqa xmm1, [rsi]
+ psadbw xmm1, xmm0
+ movq xmm2, xmm1
+ punpckhqdq xmm1, xmm1
+ paddw xmm1, xmm2
+
+ ; from left
+ dec rsi
+ lea rdi, [rax*3]
+ movzx ecx, byte [rsi+rax]
+ movzx edx, byte [rsi+rax*2]
+ add ecx, edx
+ movzx edx, byte [rsi+rdi]
+ add ecx, edx
+ lea rsi, [rsi+rax*4]
+ movzx edx, byte [rsi]
+ add ecx, edx
+ movzx edx, byte [rsi+rax]
+ add ecx, edx
+ movzx edx, byte [rsi+rax*2]
+ add ecx, edx
+ movzx edx, byte [rsi+rdi]
+ add ecx, edx
+ lea rsi, [rsi+rax*4]
+ movzx edx, byte [rsi]
+ add ecx, edx
+ movzx edx, byte [rsi+rax]
+ add ecx, edx
+ movzx edx, byte [rsi+rax*2]
+ add ecx, edx
+ movzx edx, byte [rsi+rdi]
+ add ecx, edx
+ lea rsi, [rsi+rax*4]
+ movzx edx, byte [rsi]
+ add ecx, edx
+ movzx edx, byte [rsi+rax]
+ add ecx, edx
+ movzx edx, byte [rsi+rax*2]
+ add ecx, edx
+ movzx edx, byte [rsi+rdi]
+ add ecx, edx
+ movzx edx, byte [rsi+rax*4]
+ add ecx, edx
+
+ ; add up
+ pextrw edx, xmm1, 0x0
+ lea edx, [edx+ecx+16]
+ sar edx, 5
+ movd xmm1, edx
+ ; FIXME use pshufb for ssse3 version
+ pshuflw xmm1, xmm1, 0x0
+ punpcklqdq xmm1, xmm1
+ packuswb xmm1, xmm1
+
+ ; write out
+ mov rsi, 2
+ mov rdi, arg(0) ;dst;
+ movsxd rcx, dword ptr arg(1) ;dst_stride
+ lea rax, [rcx*3]
+
+.label
+ movdqa [rdi ], xmm1
+ movdqa [rdi+rcx ], xmm1
+ movdqa [rdi+rcx*2], xmm1
+ movdqa [rdi+rax ], xmm1
+ lea rdi, [rdi+rcx*4]
+ movdqa [rdi ], xmm1
+ movdqa [rdi+rcx ], xmm1
+ movdqa [rdi+rcx*2], xmm1
+ movdqa [rdi+rax ], xmm1
+ lea rdi, [rdi+rcx*4]
+ dec rsi
+ jnz .label
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_intra_pred_y_dctop_sse2(
+; unsigned char *dst,
+; int dst_stride
+; unsigned char *src,
+; int src_stride,
+; )
+global sym(vp8_intra_pred_y_dctop_sse2)
+sym(vp8_intra_pred_y_dctop_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 4
+ push rsi
+ GET_GOT rbx
+ ; end prolog
+
+ ; from top
+ mov rcx, arg(2) ;src;
+ movsxd rax, dword ptr arg(3) ;src_stride;
+ sub rcx, rax
+ pxor xmm0, xmm0
+ movdqa xmm1, [rcx]
+ psadbw xmm1, xmm0
+ movdqa xmm2, xmm1
+ punpckhqdq xmm1, xmm1
+ paddw xmm1, xmm2
+
+ ; add up
+ paddw xmm1, [GLOBAL(dc_8)]
+ psraw xmm1, 4
+ ; FIXME use pshufb for ssse3 version
+ pshuflw xmm1, xmm1, 0x0
+ punpcklqdq xmm1, xmm1
+ packuswb xmm1, xmm1
+
+ ; write out
+ mov rsi, 2
+ mov rdx, arg(0) ;dst;
+ movsxd rcx, dword ptr arg(1) ;dst_stride
+ lea rax, [rcx*3]
+
+.label
+ movdqa [rdx ], xmm1
+ movdqa [rdx+rcx ], xmm1
+ movdqa [rdx+rcx*2], xmm1
+ movdqa [rdx+rax ], xmm1
+ lea rdx, [rdx+rcx*4]
+ movdqa [rdx ], xmm1
+ movdqa [rdx+rcx ], xmm1
+ movdqa [rdx+rcx*2], xmm1
+ movdqa [rdx+rax ], xmm1
+ lea rdx, [rdx+rcx*4]
+ dec rsi
+ jnz .label
+
+ ; begin epilog
+ RESTORE_GOT
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_intra_pred_y_dcleft_sse2(
+; unsigned char *dst,
+; int dst_stride
+; unsigned char *src,
+; int src_stride,
+; )
+global sym(vp8_intra_pred_y_dcleft_sse2)
+sym(vp8_intra_pred_y_dcleft_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 4
+ push rsi
+ push rdi
+ ; end prolog
+
+ ; from left
+ mov rsi, arg(2) ;src;
+ movsxd rax, dword ptr arg(3) ;src_stride;
+ dec rsi
+ lea rdi, [rax*3]
+ movzx ecx, byte [rsi]
+ movzx edx, byte [rsi+rax]
+ add ecx, edx
+ movzx edx, byte [rsi+rax*2]
+ add ecx, edx
+ movzx edx, byte [rsi+rdi]
+ add ecx, edx
+ lea rsi, [rsi+rax*4]
+ movzx edx, byte [rsi]
+ add ecx, edx
+ movzx edx, byte [rsi+rax]
+ add ecx, edx
+ movzx edx, byte [rsi+rax*2]
+ add ecx, edx
+ movzx edx, byte [rsi+rdi]
+ add ecx, edx
+ lea rsi, [rsi+rax*4]
+ movzx edx, byte [rsi]
+ add ecx, edx
+ movzx edx, byte [rsi+rax]
+ add ecx, edx
+ movzx edx, byte [rsi+rax*2]
+ add ecx, edx
+ movzx edx, byte [rsi+rdi]
+ add ecx, edx
+ lea rsi, [rsi+rax*4]
+ movzx edx, byte [rsi]
+ add ecx, edx
+ movzx edx, byte [rsi+rax]
+ add ecx, edx
+ movzx edx, byte [rsi+rax*2]
+ add ecx, edx
+ movzx edx, byte [rsi+rdi]
+ lea edx, [ecx+edx+8]
+
+ ; add up
+ shr edx, 4
+ movd xmm1, edx
+ ; FIXME use pshufb for ssse3 version
+ pshuflw xmm1, xmm1, 0x0
+ punpcklqdq xmm1, xmm1
+ packuswb xmm1, xmm1
+
+ ; write out
+ mov rsi, 2
+ mov rdi, arg(0) ;dst;
+ movsxd rcx, dword ptr arg(1) ;dst_stride
+ lea rax, [rcx*3]
+
+.label
+ movdqa [rdi ], xmm1
+ movdqa [rdi+rcx ], xmm1
+ movdqa [rdi+rcx*2], xmm1
+ movdqa [rdi+rax ], xmm1
+ lea rdi, [rdi+rcx*4]
+ movdqa [rdi ], xmm1
+ movdqa [rdi+rcx ], xmm1
+ movdqa [rdi+rcx*2], xmm1
+ movdqa [rdi+rax ], xmm1
+ lea rdi, [rdi+rcx*4]
+ dec rsi
+ jnz .label
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_intra_pred_y_dc128_sse2(
+; unsigned char *dst,
+; int dst_stride
+; unsigned char *src,
+; int src_stride,
+; )
+global sym(vp8_intra_pred_y_dc128_sse2)
+sym(vp8_intra_pred_y_dc128_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 4
+ push rsi
+ GET_GOT rbx
+ ; end prolog
+
+ ; write out
+ mov rsi, 2
+ movdqa xmm1, [GLOBAL(dc_128)]
+ mov rax, arg(0) ;dst;
+ movsxd rdx, dword ptr arg(1) ;dst_stride
+ lea rcx, [rdx*3]
+
+.label
+ movdqa [rax ], xmm1
+ movdqa [rax+rdx ], xmm1
+ movdqa [rax+rdx*2], xmm1
+ movdqa [rax+rcx ], xmm1
+ lea rax, [rax+rdx*4]
+ movdqa [rax ], xmm1
+ movdqa [rax+rdx ], xmm1
+ movdqa [rax+rdx*2], xmm1
+ movdqa [rax+rcx ], xmm1
+ lea rax, [rax+rdx*4]
+ dec rsi
+ jnz .label
+
+ ; begin epilog
+ RESTORE_GOT
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_intra_pred_y_tm_sse2(
+; unsigned char *dst,
+; int dst_stride
+; unsigned char *src,
+; int src_stride,
+; )
+%macro vp8_intra_pred_y_tm 1
+global sym(vp8_intra_pred_y_tm_%1)
+sym(vp8_intra_pred_y_tm_%1):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 4
+ push rsi
+ push rdi
+ GET_GOT rbx
+ ; end prolog
+
+ ; read top row
+ mov edx, 8
+ mov rsi, arg(2) ;src;
+ movsxd rax, dword ptr arg(3) ;src_stride;
+ sub rsi, rax
+ pxor xmm0, xmm0
+%ifidn %1, ssse3
+ movdqa xmm3, [GLOBAL(dc_1024)]
+%endif
+ movdqa xmm1, [rsi]
+ movdqa xmm2, xmm1
+ punpcklbw xmm1, xmm0
+ punpckhbw xmm2, xmm0
+
+ ; set up left ptrs ans subtract topleft
+ movd xmm4, [rsi-1]
+ lea rsi, [rsi+rax-1]
+%ifidn %1, sse2
+ punpcklbw xmm4, xmm0
+ pshuflw xmm4, xmm4, 0x0
+ punpcklqdq xmm4, xmm4
+%else
+ pshufb xmm4, xmm3
+%endif
+ psubw xmm1, xmm4
+ psubw xmm2, xmm4
+
+ ; set up dest ptrs
+ mov rdi, arg(0) ;dst;
+ movsxd rcx, dword ptr arg(1) ;dst_stride
+vp8_intra_pred_y_tm_%1_loop:
+ movd xmm4, [rsi]
+ movd xmm5, [rsi+rax]
+%ifidn %1, sse2
+ punpcklbw xmm4, xmm0
+ punpcklbw xmm5, xmm0
+ pshuflw xmm4, xmm4, 0x0
+ pshuflw xmm5, xmm5, 0x0
+ punpcklqdq xmm4, xmm4
+ punpcklqdq xmm5, xmm5
+%else
+ pshufb xmm4, xmm3
+ pshufb xmm5, xmm3
+%endif
+ movdqa xmm6, xmm4
+ movdqa xmm7, xmm5
+ paddw xmm4, xmm1
+ paddw xmm6, xmm2
+ paddw xmm5, xmm1
+ paddw xmm7, xmm2
+ packuswb xmm4, xmm6
+ packuswb xmm5, xmm7
+ movdqa [rdi ], xmm4
+ movdqa [rdi+rcx], xmm5
+ lea rsi, [rsi+rax*2]
+ lea rdi, [rdi+rcx*2]
+ dec edx
+ jnz vp8_intra_pred_y_tm_%1_loop
+
+ ; begin epilog
+ RESTORE_GOT
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+%endmacro
+
+vp8_intra_pred_y_tm sse2
+vp8_intra_pred_y_tm ssse3
+
+;void vp8_intra_pred_y_ve_sse2(
+; unsigned char *dst,
+; int dst_stride
+; unsigned char *src,
+; int src_stride,
+; )
+global sym(vp8_intra_pred_y_ve_sse2)
+sym(vp8_intra_pred_y_ve_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 4
+ push rsi
+ ; end prolog
+
+ ; read from top
+ mov rax, arg(2) ;src;
+ movsxd rdx, dword ptr arg(3) ;src_stride;
+ sub rax, rdx
+ movdqa xmm1, [rax]
+
+ ; write out
+ mov rsi, 2
+ mov rax, arg(0) ;dst;
+ movsxd rdx, dword ptr arg(1) ;dst_stride
+ lea rcx, [rdx*3]
+
+.label
+ movdqa [rax ], xmm1
+ movdqa [rax+rdx ], xmm1
+ movdqa [rax+rdx*2], xmm1
+ movdqa [rax+rcx ], xmm1
+ lea rax, [rax+rdx*4]
+ movdqa [rax ], xmm1
+ movdqa [rax+rdx ], xmm1
+ movdqa [rax+rdx*2], xmm1
+ movdqa [rax+rcx ], xmm1
+ lea rax, [rax+rdx*4]
+ dec rsi
+ jnz .label
+
+ ; begin epilog
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
+;void vp8_intra_pred_y_ho_sse2(
+; unsigned char *dst,
+; int dst_stride
+; unsigned char *src,
+; int src_stride,
+; )
+global sym(vp8_intra_pred_y_ho_sse2)
+sym(vp8_intra_pred_y_ho_sse2):
+ push rbp
+ mov rbp, rsp
+ SHADOW_ARGS_TO_STACK 4
+ push rsi
+ push rdi
+ ; end prolog
+
+ ; read from left and write out
+ mov edx, 8
+ mov rsi, arg(2) ;src;
+ movsxd rax, dword ptr arg(3) ;src_stride;
+ mov rdi, arg(0) ;dst;
+ movsxd rcx, dword ptr arg(1) ;dst_stride
+ dec rsi
+
+vp8_intra_pred_y_ho_sse2_loop:
+ movd xmm0, [rsi]
+ movd xmm1, [rsi+rax]
+ ; FIXME use pshufb for ssse3 version
+ punpcklbw xmm0, xmm0
+ punpcklbw xmm1, xmm1
+ pshuflw xmm0, xmm0, 0x0
+ pshuflw xmm1, xmm1, 0x0
+ punpcklqdq xmm0, xmm0
+ punpcklqdq xmm1, xmm1
+ movdqa [rdi ], xmm0
+ movdqa [rdi+rcx], xmm1
+ lea rsi, [rsi+rax*2]
+ lea rdi, [rdi+rcx*2]
+ dec edx
+ jnz vp8_intra_pred_y_ho_sse2_loop
+
+ ; begin epilog
+ pop rdi
+ pop rsi
+ UNSHADOW_ARGS
+ pop rbp
+ ret
+
SECTION_RODATA
+align 16
dc_128:
- times 8 db 128
+ times 16 db 128
dc_4:
times 4 dw 4
align 16
+dc_8:
+ times 8 dw 8
+align 16
dc_1024:
times 8 dw 0x400
align 16
diff --git a/vp8/common/x86/recon_wrapper_sse2.c b/vp8/common/x86/recon_wrapper_sse2.c
index fcc75a9..44221cd 100644
--- a/vp8/common/x86/recon_wrapper_sse2.c
+++ b/vp8/common/x86/recon_wrapper_sse2.c
@@ -94,3 +94,69 @@
vp8_intra_pred_uv_tm_ssse3,
vp8_intra_pred_uv_ho_ssse3);
}
+
+extern build_intra_predictors_mbuv_prototype(vp8_intra_pred_y_dc_sse2);
+extern build_intra_predictors_mbuv_prototype(vp8_intra_pred_y_dctop_sse2);
+extern build_intra_predictors_mbuv_prototype(vp8_intra_pred_y_dcleft_sse2);
+extern build_intra_predictors_mbuv_prototype(vp8_intra_pred_y_dc128_sse2);
+extern build_intra_predictors_mbuv_prototype(vp8_intra_pred_y_ho_sse2);
+extern build_intra_predictors_mbuv_prototype(vp8_intra_pred_y_ve_sse2);
+extern build_intra_predictors_mbuv_prototype(vp8_intra_pred_y_tm_sse2);
+extern build_intra_predictors_mbuv_prototype(vp8_intra_pred_y_tm_ssse3);
+
+static void vp8_build_intra_predictors_mby_x86(MACROBLOCKD *x,
+ unsigned char *dst_y,
+ int dst_stride,
+ build_intra_predictors_mbuv_fn_t tm_func)
+{
+ int mode = x->mode_info_context->mbmi.mode;
+ build_intra_predictors_mbuv_fn_t fn;
+ int src_stride = x->dst.y_stride;
+ switch (mode) {
+ case V_PRED: fn = vp8_intra_pred_y_ve_sse2; break;
+ case H_PRED: fn = vp8_intra_pred_y_ho_sse2; break;
+ case TM_PRED: fn = tm_func; break;
+ case DC_PRED:
+ if (x->up_available) {
+ if (x->left_available) {
+ fn = vp8_intra_pred_y_dc_sse2; break;
+ } else {
+ fn = vp8_intra_pred_y_dctop_sse2; break;
+ }
+ } else if (x->left_available) {
+ fn = vp8_intra_pred_y_dcleft_sse2; break;
+ } else {
+ fn = vp8_intra_pred_y_dc128_sse2; break;
+ }
+ break;
+ default: return;
+ }
+
+ fn(dst_y, dst_stride, x->dst.y_buffer, src_stride);
+ return;
+}
+
+void vp8_build_intra_predictors_mby_sse2(MACROBLOCKD *x)
+{
+ vp8_build_intra_predictors_mby_x86(x, x->predictor, 16,
+ vp8_intra_pred_y_tm_sse2);
+}
+
+void vp8_build_intra_predictors_mby_ssse3(MACROBLOCKD *x)
+{
+ vp8_build_intra_predictors_mby_x86(x, x->predictor, 16,
+ vp8_intra_pred_y_tm_ssse3);
+}
+
+void vp8_build_intra_predictors_mby_s_sse2(MACROBLOCKD *x)
+{
+ vp8_build_intra_predictors_mby_x86(x, x->dst.y_buffer, x->dst.y_stride,
+ vp8_intra_pred_y_tm_sse2);
+}
+
+void vp8_build_intra_predictors_mby_s_ssse3(MACROBLOCKD *x)
+{
+ vp8_build_intra_predictors_mby_x86(x, x->dst.y_buffer, x->dst.y_stride,
+ vp8_intra_pred_y_tm_ssse3);
+
+}
diff --git a/vp8/common/x86/recon_x86.h b/vp8/common/x86/recon_x86.h
index fbb3dcb..afacc60 100644
--- a/vp8/common/x86/recon_x86.h
+++ b/vp8/common/x86/recon_x86.h
@@ -42,6 +42,8 @@
extern prototype_copy_block(vp8_copy_mem16x16_sse2);
extern prototype_build_intra_predictors(vp8_build_intra_predictors_mbuv_sse2);
extern prototype_build_intra_predictors(vp8_build_intra_predictors_mbuv_s_sse2);
+extern prototype_build_intra_predictors(vp8_build_intra_predictors_mby_sse2);
+extern prototype_build_intra_predictors(vp8_build_intra_predictors_mby_s_sse2);
#if !CONFIG_RUNTIME_CPU_DETECT
#undef vp8_recon_copy16x16
@@ -53,12 +55,20 @@
#undef vp8_recon_build_intra_predictors_mbuv_s
#define vp8_recon_build_intra_predictors_mbuv_s vp8_build_intra_predictors_mbuv_s_sse2
+#undef vp8_recon_build_intra_predictors_mby
+#define vp8_recon_build_intra_predictors_mby vp8_build_intra_predictors_mby_sse2
+
+#undef vp8_recon_build_intra_predictors_mby_s
+#define vp8_recon_build_intra_predictors_mby_s vp8_build_intra_predictors_mby_s_sse2
+
#endif
#endif
#if HAVE_SSSE3
extern prototype_build_intra_predictors(vp8_build_intra_predictors_mbuv_ssse3);
extern prototype_build_intra_predictors(vp8_build_intra_predictors_mbuv_s_ssse3);
+extern prototype_build_intra_predictors(vp8_build_intra_predictors_mby_ssse3);
+extern prototype_build_intra_predictors(vp8_build_intra_predictors_mby_s_ssse3);
#if !CONFIG_RUNTIME_CPU_DETECT
#undef vp8_recon_build_intra_predictors_mbuv
@@ -67,6 +77,12 @@
#undef vp8_recon_build_intra_predictors_mbuv_s
#define vp8_recon_build_intra_predictors_mbuv_s vp8_build_intra_predictors_mbuv_s_ssse3
+#undef vp8_recon_build_intra_predictors_mby
+#define vp8_recon_build_intra_predictors_mby vp8_build_intra_predictors_mby_ssse3
+
+#undef vp8_recon_build_intra_predictors_mby_s
+#define vp8_recon_build_intra_predictors_mby_s vp8_build_intra_predictors_mby_s_ssse3
+
#endif
#endif
#endif
diff --git a/vp8/common/x86/x86_systemdependent.c b/vp8/common/x86/x86_systemdependent.c
index b24cbe4..e1e1b79 100644
--- a/vp8/common/x86/x86_systemdependent.c
+++ b/vp8/common/x86/x86_systemdependent.c
@@ -11,7 +11,6 @@
#include "vpx_config.h"
#include "vpx_ports/x86.h"
-#include "vp8/common/g_common.h"
#include "vp8/common/subpixel.h"
#include "vp8/common/loopfilter.h"
#include "vp8/common/recon.h"
@@ -37,6 +36,11 @@
if (flags & HAS_MMX)
{
+ rtcd->dequant.block = vp8_dequantize_b_mmx;
+ rtcd->dequant.idct_add = vp8_dequant_idct_add_mmx;
+ rtcd->dequant.idct_add_y_block = vp8_dequant_idct_add_y_block_mmx;
+ rtcd->dequant.idct_add_uv_block = vp8_dequant_idct_add_uv_block_mmx;
+
rtcd->idct.idct16 = vp8_short_idct4x4llm_mmx;
rtcd->idct.idct1_scalar_add = vp8_dc_only_idct_add_mmx;
rtcd->idct.iwalsh16 = vp8_short_inv_walsh4x4_mmx;
@@ -81,6 +85,13 @@
vp8_build_intra_predictors_mbuv_sse2;
rtcd->recon.build_intra_predictors_mbuv_s =
vp8_build_intra_predictors_mbuv_s_sse2;
+ rtcd->recon.build_intra_predictors_mby =
+ vp8_build_intra_predictors_mby_sse2;
+ rtcd->recon.build_intra_predictors_mby_s =
+ vp8_build_intra_predictors_mby_s_sse2;
+
+ rtcd->dequant.idct_add_y_block = vp8_dequant_idct_add_y_block_sse2;
+ rtcd->dequant.idct_add_uv_block = vp8_dequant_idct_add_uv_block_sse2;
rtcd->idct.iwalsh16 = vp8_short_inv_walsh4x4_sse2;
@@ -124,6 +135,10 @@
vp8_build_intra_predictors_mbuv_ssse3;
rtcd->recon.build_intra_predictors_mbuv_s =
vp8_build_intra_predictors_mbuv_s_ssse3;
+ rtcd->recon.build_intra_predictors_mby =
+ vp8_build_intra_predictors_mby_ssse3;
+ rtcd->recon.build_intra_predictors_mby_s =
+ vp8_build_intra_predictors_mby_s_ssse3;
}
#endif
diff --git a/vp8/decoder/arm/arm_dsystemdependent.c b/vp8/decoder/arm/arm_dsystemdependent.c
index f802c51..bf0a348 100644
--- a/vp8/decoder/arm/arm_dsystemdependent.c
+++ b/vp8/decoder/arm/arm_dsystemdependent.c
@@ -11,9 +11,6 @@
#include "vpx_config.h"
#include "vpx_ports/arm.h"
-#include "vp8/common/blockd.h"
-#include "vp8/common/pragmas.h"
-#include "vp8/decoder/dequantize.h"
#include "vp8/decoder/onyxd_int.h"
void vp8_arch_arm_decode_init(VP8D_COMP *pbi)
@@ -30,20 +27,12 @@
#if HAVE_ARMV6
if (flags & HAS_MEDIA)
{
- pbi->dequant.block = vp8_dequantize_b_v6;
- pbi->dequant.idct_add = vp8_dequant_idct_add_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;
}
#endif
#if HAVE_ARMV7
if (flags & HAS_NEON)
{
- pbi->dequant.block = vp8_dequantize_b_neon;
- pbi->dequant.idct_add = vp8_dequant_idct_add_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;
}
#endif
#endif
diff --git a/vp8/decoder/decodframe.c b/vp8/decoder/decodframe.c
index 31eafcf..11d0e38 100644
--- a/vp8/decoder/decodframe.c
+++ b/vp8/decoder/decodframe.c
@@ -15,7 +15,7 @@
#include "vp8/common/reconintra4x4.h"
#include "vp8/common/recon.h"
#include "vp8/common/reconinter.h"
-#include "dequantize.h"
+#include "vp8/common/dequantize.h"
#include "detokenize.h"
#include "vp8/common/invtrans.h"
#include "vp8/common/alloccommon.h"
@@ -32,7 +32,7 @@
#endif
#include "vpx_mem/vpx_mem.h"
#include "vp8/common/idct.h"
-#include "dequantize.h"
+
#include "vp8/common/threading.h"
#include "decoderthreading.h"
#include "dboolhuff.h"
@@ -109,32 +109,12 @@
#define RTCD_VTABLE(x) NULL
#endif
-/* skip_recon_mb() is Modified: Instead of writing the result to predictor buffer and then copying it
- * to dst buffer, we can write the result directly to dst buffer. This eliminates unnecessary copy.
- */
-static void skip_recon_mb(VP8D_COMP *pbi, MACROBLOCKD *xd)
-{
- if (xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME)
- {
- RECON_INVOKE(&pbi->common.rtcd.recon, build_intra_predictors_mbuv_s)(xd);
- RECON_INVOKE(&pbi->common.rtcd.recon,
- build_intra_predictors_mby_s)(xd);
- }
- else
- {
- vp8_build_inter16x16_predictors_mb(xd, xd->dst.y_buffer,
- xd->dst.u_buffer, xd->dst.v_buffer,
- xd->dst.y_stride, xd->dst.uv_stride);
- }
-}
-
static void decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd,
unsigned int mb_idx)
{
- int eobtotal = 0;
- int throw_residual = 0;
MB_PREDICTION_MODE mode;
int i;
+ int corruption_detected = 0;
if (xd->mode_info_context->mbmi.mb_skip_coeff)
{
@@ -142,28 +122,52 @@
}
else if (!vp8dx_bool_error(xd->current_bc))
{
+ int eobtotal;
eobtotal = vp8_decode_mb_tokens(pbi, xd);
+
+ /* Special case: Force the loopfilter to skip when eobtotal is zero */
+ xd->mode_info_context->mbmi.mb_skip_coeff = (eobtotal==0);
}
-
-
mode = xd->mode_info_context->mbmi.mode;
- if (eobtotal == 0 && mode != B_PRED && mode != SPLITMV &&
- !vp8dx_bool_error(xd->current_bc))
- {
- /* Special case: Force the loopfilter to skip when eobtotal and
- * mb_skip_coeff are zero.
- * */
- xd->mode_info_context->mbmi.mb_skip_coeff = 1;
-
- skip_recon_mb(pbi, xd);
- return;
- }
-
if (xd->segmentation_enabled)
mb_init_dequantizer(pbi, xd);
+
+#if CONFIG_ERROR_CONCEALMENT
+
+ if(pbi->ec_active)
+ {
+ int throw_residual;
+ /* When we have independent partitions we can apply residual even
+ * though other partitions within the frame are corrupt.
+ */
+ throw_residual = (!pbi->independent_partitions &&
+ pbi->frame_corrupt_residual);
+ throw_residual = (throw_residual || vp8dx_bool_error(xd->current_bc));
+
+ if ((mb_idx >= pbi->mvs_corrupt_from_mb || throw_residual))
+ {
+ /* MB with corrupt residuals or corrupt mode/motion vectors.
+ * Better to use the predictor as reconstruction.
+ */
+ pbi->frame_corrupt_residual = 1;
+ vpx_memset(xd->qcoeff, 0, sizeof(xd->qcoeff));
+ vp8_conceal_corrupt_mb(xd);
+
+
+ corruption_detected = 1;
+
+ /* force idct to be skipped for B_PRED and use the
+ * prediction only for reconstruction
+ * */
+ vpx_memset(xd->eobs, 0, 25);
+ }
+ }
+#endif
+
+
/* do prediction */
if (xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME)
{
@@ -173,121 +177,117 @@
{
RECON_INVOKE(&pbi->common.rtcd.recon,
build_intra_predictors_mby_s)(xd);
- } else {
+ }
+ else
+ {
+ /* clear out residual eob info */
+ if(xd->mode_info_context->mbmi.mb_skip_coeff)
+ vpx_memset(xd->eobs, 0, 25);
+
vp8_intra_prediction_down_copy(xd);
+
+ for (i = 0; i < 16; i++)
+ {
+ BLOCKD *b = &xd->block[i];
+ int b_mode = xd->mode_info_context->bmi[i].as_mode;
+
+ RECON_INVOKE(RTCD_VTABLE(recon), intra4x4_predict)
+ ( *(b->base_dst) + b->dst, b->dst_stride, b_mode,
+ *(b->base_dst) + b->dst, b->dst_stride );
+
+ if (xd->eobs[i])
+ {
+ if (xd->eobs[i] > 1)
+ {
+ DEQUANT_INVOKE(&pbi->common.rtcd.dequant, idct_add)
+ (b->qcoeff, b->dequant,
+ *(b->base_dst) + b->dst, b->dst_stride);
+ }
+ else
+ {
+ IDCT_INVOKE(RTCD_VTABLE(idct), idct1_scalar_add)
+ (b->qcoeff[0] * b->dequant[0],
+ *(b->base_dst) + b->dst, b->dst_stride,
+ *(b->base_dst) + b->dst, b->dst_stride);
+ ((int *)b->qcoeff)[0] = 0;
+ }
+ }
+ }
}
}
else
{
vp8_build_inter_predictors_mb(xd);
}
- /* When we have independent partitions we can apply residual even
- * though other partitions within the frame are corrupt.
- */
- throw_residual = (!pbi->independent_partitions &&
- pbi->frame_corrupt_residual);
- throw_residual = (throw_residual || vp8dx_bool_error(xd->current_bc));
+
#if CONFIG_ERROR_CONCEALMENT
- if (pbi->ec_active &&
- (mb_idx >= pbi->mvs_corrupt_from_mb || throw_residual))
+ if (corruption_detected)
{
- /* MB with corrupt residuals or corrupt mode/motion vectors.
- * Better to use the predictor as reconstruction.
- */
- pbi->frame_corrupt_residual = 1;
- vpx_memset(xd->qcoeff, 0, sizeof(xd->qcoeff));
- vp8_conceal_corrupt_mb(xd);
return;
}
#endif
- /* dequantization and idct */
- if (mode == B_PRED)
+ if(!xd->mode_info_context->mbmi.mb_skip_coeff)
{
- for (i = 0; i < 16; i++)
+ /* dequantization and idct */
+ if (mode != B_PRED)
{
- BLOCKD *b = &xd->block[i];
- int b_mode = xd->mode_info_context->bmi[i].as_mode;
+ short *DQC = xd->block[0].dequant;
- RECON_INVOKE(RTCD_VTABLE(recon), intra4x4_predict)
- ( *(b->base_dst) + b->dst, b->dst_stride, b_mode,
- *(b->base_dst) + b->dst, b->dst_stride );
+ /* save the dc dequant constant in case it is overridden */
+ short dc_dequant_temp = DQC[0];
- if (xd->eobs[i] )
+ if (mode != SPLITMV)
{
- if (xd->eobs[i] > 1)
+ BLOCKD *b = &xd->block[24];
+
+ /* do 2nd order transform on the dc block */
+ if (xd->eobs[24] > 1)
{
- DEQUANT_INVOKE(&pbi->dequant, idct_add)
- (b->qcoeff, b->dequant,
- *(b->base_dst) + b->dst, b->dst_stride);
+ DEQUANT_INVOKE(&pbi->common.rtcd.dequant, block)(b);
+
+ IDCT_INVOKE(RTCD_VTABLE(idct), iwalsh16)(&b->dqcoeff[0],
+ xd->qcoeff);
+ ((int *)b->qcoeff)[0] = 0;
+ ((int *)b->qcoeff)[1] = 0;
+ ((int *)b->qcoeff)[2] = 0;
+ ((int *)b->qcoeff)[3] = 0;
+ ((int *)b->qcoeff)[4] = 0;
+ ((int *)b->qcoeff)[5] = 0;
+ ((int *)b->qcoeff)[6] = 0;
+ ((int *)b->qcoeff)[7] = 0;
}
else
{
- IDCT_INVOKE(RTCD_VTABLE(idct), idct1_scalar_add)
- (b->qcoeff[0] * b->dequant[0],
- *(b->base_dst) + b->dst, b->dst_stride,
- *(b->base_dst) + b->dst, b->dst_stride);
+ b->dqcoeff[0] = b->qcoeff[0] * b->dequant[0];
+ IDCT_INVOKE(RTCD_VTABLE(idct), iwalsh1)(&b->dqcoeff[0],
+ xd->qcoeff);
((int *)b->qcoeff)[0] = 0;
}
- }
- }
- }
- else
- {
- short *DQC = xd->block[0].dequant;
- /* save the dc dequant constant in case it is overridden */
- short dc_dequant_temp = DQC[0];
-
- if (mode != SPLITMV)
- {
- BLOCKD *b = &xd->block[24];
-
- /* do 2nd order transform on the dc block */
- if (xd->eobs[24] > 1)
- {
- DEQUANT_INVOKE(&pbi->dequant, block)(b);
-
- IDCT_INVOKE(RTCD_VTABLE(idct), iwalsh16)(&b->dqcoeff[0],
- xd->qcoeff);
- ((int *)b->qcoeff)[0] = 0;
- ((int *)b->qcoeff)[1] = 0;
- ((int *)b->qcoeff)[2] = 0;
- ((int *)b->qcoeff)[3] = 0;
- ((int *)b->qcoeff)[4] = 0;
- ((int *)b->qcoeff)[5] = 0;
- ((int *)b->qcoeff)[6] = 0;
- ((int *)b->qcoeff)[7] = 0;
- }
- else
- {
- b->dqcoeff[0] = b->qcoeff[0] * b->dequant[0];
- IDCT_INVOKE(RTCD_VTABLE(idct), iwalsh1)(&b->dqcoeff[0],
- xd->qcoeff);
- ((int *)b->qcoeff)[0] = 0;
+ /* override the dc dequant constant in order to preserve the
+ * dc components
+ */
+ DQC[0] = 1;
}
- /* override the dc dequant constant */
- DQC[0] = 1;
+ DEQUANT_INVOKE (&pbi->common.rtcd.dequant, idct_add_y_block)
+ (xd->qcoeff, xd->block[0].dequant,
+ xd->dst.y_buffer,
+ xd->dst.y_stride, xd->eobs);
+
+ /* restore the dc dequant constant */
+ DQC[0] = dc_dequant_temp;
}
- DEQUANT_INVOKE (&pbi->dequant, idct_add_y_block)
- (xd->qcoeff, xd->block[0].dequant,
- xd->dst.y_buffer,
- xd->dst.y_stride, xd->eobs);
-
- /* restore the dc dequant constant */
- DQC[0] = dc_dequant_temp;
+ DEQUANT_INVOKE (&pbi->common.rtcd.dequant, idct_add_uv_block)
+ (xd->qcoeff+16*16, xd->block[16].dequant,
+ xd->dst.u_buffer, xd->dst.v_buffer,
+ xd->dst.uv_stride, xd->eobs+16);
}
-
- DEQUANT_INVOKE (&pbi->dequant, idct_add_uv_block)
- (xd->qcoeff+16*16, xd->block[16].dequant,
- xd->dst.u_buffer, xd->dst.v_buffer,
- xd->dst.uv_stride, xd->eobs+16);
}
-
static int get_delta_q(vp8_reader *bc, int prev, int *q_update)
{
int ret_val = 0;
@@ -484,7 +484,8 @@
const unsigned char* token_part_sizes)
{
vp8_reader *bool_decoder = &pbi->bc2;
- int fragment_idx, partition_idx;
+ unsigned int partition_idx;
+ int fragment_idx;
int num_token_partitions;
const unsigned char *first_fragment_end = pbi->fragments[0] +
pbi->fragment_sizes[0];
diff --git a/vp8/decoder/detokenize.c b/vp8/decoder/detokenize.c
index 1d45685..ba94c58 100644
--- a/vp8/decoder/detokenize.c
+++ b/vp8/decoder/detokenize.c
@@ -15,7 +15,7 @@
#include "vpx_ports/mem.h"
#include "detokenize.h"
-#define BOOL_DATA UINT8
+#define BOOL_DATA unsigned char
#define OCB_X PREV_COEF_CONTEXTS * ENTROPY_NODES
DECLARE_ALIGNED(16, static const unsigned char, coef_bands_x[16]) =
@@ -157,10 +157,10 @@
DECODE_AND_APPLYSIGN(val) \
Prob = coef_probs + (ENTROPY_NODES*2); \
if(c < 15){\
- qcoeff_ptr [ scan[c] ] = (INT16) v; \
+ qcoeff_ptr [ scan[c] ] = (int16_t) v; \
++c; \
goto DO_WHILE; }\
- qcoeff_ptr [ 15 ] = (INT16) v; \
+ qcoeff_ptr [ 15 ] = (int16_t) v; \
goto BLOCK_FINISHED;
@@ -172,7 +172,7 @@
{\
range = range-split;\
value = value-bigsplit;\
- val += ((UINT16)1<<bits_count);\
+ val += ((uint16_t)1<<bits_count);\
}\
else\
{\
@@ -340,12 +340,12 @@
if (c < 15)
{
- qcoeff_ptr [ scan[c] ] = (INT16) v;
+ qcoeff_ptr [ scan[c] ] = (int16_t) v;
++c;
goto DO_WHILE;
}
- qcoeff_ptr [ 15 ] = (INT16) v;
+ qcoeff_ptr [ 15 ] = (int16_t) v;
BLOCK_FINISHED:
eobs[i] = c;
eobtotal += c;
diff --git a/vp8/decoder/generic/dsystemdependent.c b/vp8/decoder/generic/dsystemdependent.c
index d9f9ba3..8a84e56 100644
--- a/vp8/decoder/generic/dsystemdependent.c
+++ b/vp8/decoder/generic/dsystemdependent.c
@@ -10,7 +10,7 @@
#include "vpx_config.h"
-#include "vp8/decoder/dequantize.h"
+#include "vp8/common/dequantize.h"
#include "vp8/decoder/onyxd_int.h"
extern void vp8_arch_x86_decode_init(VP8D_COMP *pbi);
@@ -20,11 +20,7 @@
{
/* Pure C: */
#if CONFIG_RUNTIME_CPU_DETECT
- pbi->mb.rtcd = &pbi->common.rtcd;
- pbi->dequant.block = vp8_dequantize_b_c;
- pbi->dequant.idct_add = vp8_dequant_idct_add_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->mb.rtcd = &pbi->common.rtcd;
#endif
#if ARCH_X86 || ARCH_X86_64
diff --git a/vp8/decoder/onyxd_if.c b/vp8/decoder/onyxd_if.c
index cf525f4..80648d3 100644
--- a/vp8/decoder/onyxd_if.c
+++ b/vp8/decoder/onyxd_if.c
@@ -20,7 +20,6 @@
#include "vpx_scale/yv12extend.h"
#include "vp8/common/loopfilter.h"
#include "vp8/common/swapyv12buffer.h"
-#include "vp8/common/g_common.h"
#include "vp8/common/threading.h"
#include "decoderthreading.h"
#include <stdio.h>
@@ -57,7 +56,7 @@
}
-VP8D_PTR vp8dx_create_decompressor(VP8D_CONFIG *oxcf)
+struct VP8D_COMP * vp8dx_create_decompressor(VP8D_CONFIG *oxcf)
{
VP8D_COMP *pbi = vpx_memalign(32, sizeof(VP8D_COMP));
@@ -117,14 +116,12 @@
*/
pbi->independent_partitions = 0;
- return (VP8D_PTR) pbi;
+ return pbi;
}
-void vp8dx_remove_decompressor(VP8D_PTR ptr)
+void vp8dx_remove_decompressor(VP8D_COMP *pbi)
{
- VP8D_COMP *pbi = (VP8D_COMP *) ptr;
-
if (!pbi)
return;
@@ -142,9 +139,8 @@
}
-vpx_codec_err_t vp8dx_get_reference(VP8D_PTR ptr, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd)
+vpx_codec_err_t vp8dx_get_reference(VP8D_COMP *pbi, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd)
{
- VP8D_COMP *pbi = (VP8D_COMP *) ptr;
VP8_COMMON *cm = &pbi->common;
int ref_fb_idx;
@@ -174,9 +170,8 @@
}
-vpx_codec_err_t vp8dx_set_reference(VP8D_PTR ptr, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd)
+vpx_codec_err_t vp8dx_set_reference(VP8D_COMP *pbi, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd)
{
- VP8D_COMP *pbi = (VP8D_COMP *) ptr;
VP8_COMMON *cm = &pbi->common;
int *ref_fb_ptr = NULL;
int free_fb;
@@ -301,19 +296,18 @@
return err;
}
-int vp8dx_receive_compressed_data(VP8D_PTR ptr, unsigned long size, const unsigned char *source, int64_t time_stamp)
+int vp8dx_receive_compressed_data(VP8D_COMP *pbi, unsigned long size, const unsigned char *source, int64_t time_stamp)
{
#if HAVE_ARMV7
int64_t dx_store_reg[8];
#endif
- VP8D_COMP *pbi = (VP8D_COMP *) ptr;
VP8_COMMON *cm = &pbi->common;
int retcode = 0;
/*if(pbi->ready_for_new_data == 0)
return -1;*/
- if (ptr == 0)
+ if (pbi == 0)
{
return -1;
}
@@ -575,10 +569,9 @@
pbi->common.error.setjmp = 0;
return retcode;
}
-int vp8dx_get_raw_frame(VP8D_PTR ptr, YV12_BUFFER_CONFIG *sd, int64_t *time_stamp, int64_t *time_end_stamp, vp8_ppflags_t *flags)
+int vp8dx_get_raw_frame(VP8D_COMP *pbi, YV12_BUFFER_CONFIG *sd, int64_t *time_stamp, int64_t *time_end_stamp, vp8_ppflags_t *flags)
{
int ret = -1;
- VP8D_COMP *pbi = (VP8D_COMP *) ptr;
if (pbi->ready_for_new_data == 1)
return ret;
diff --git a/vp8/decoder/onyxd_int.h b/vp8/decoder/onyxd_int.h
index 519a7f2..cb2593b 100644
--- a/vp8/decoder/onyxd_int.h
+++ b/vp8/decoder/onyxd_int.h
@@ -16,7 +16,8 @@
#include "treereader.h"
#include "vp8/common/onyxc_int.h"
#include "vp8/common/threading.h"
-#include "dequantize.h"
+
+
#if CONFIG_ERROR_CONCEALMENT
#include "ec_types.h"
#endif
@@ -43,7 +44,7 @@
} DATARATE;
-typedef struct VP8Decompressor
+typedef struct VP8D_COMP
{
DECLARE_ALIGNED(16, MACROBLOCKD, mb);
@@ -93,11 +94,6 @@
DATARATE dr[16];
-#if CONFIG_RUNTIME_CPU_DETECT
- vp8_dequant_rtcd_vtable_t dequant;
-#endif
-
-
vp8_prob prob_intra;
vp8_prob prob_last;
vp8_prob prob_gf;
diff --git a/vp8/decoder/threading.c b/vp8/decoder/threading.c
index 1967781..947b3a1 100644
--- a/vp8/decoder/threading.c
+++ b/vp8/decoder/threading.c
@@ -189,7 +189,7 @@
{
if (xd->eobs[i] > 1)
{
- DEQUANT_INVOKE(&pbi->dequant, idct_add)
+ DEQUANT_INVOKE(&pbi->common.rtcd.dequant, idct_add)
(b->qcoeff, b->dequant,
*(b->base_dst) + b->dst, b->dst_stride);
}
@@ -217,7 +217,7 @@
/* do 2nd order transform on the dc block */
if (xd->eobs[24] > 1)
{
- DEQUANT_INVOKE(&pbi->dequant, block)(b);
+ DEQUANT_INVOKE(&pbi->common.rtcd.dequant, block)(b);
IDCT_INVOKE(RTCD_VTABLE(idct), iwalsh16)(&b->dqcoeff[0],
xd->qcoeff);
@@ -248,13 +248,13 @@
DQC = local_dequant;
}
- DEQUANT_INVOKE (&pbi->dequant, idct_add_y_block)
+ DEQUANT_INVOKE (&pbi->common.rtcd.dequant, idct_add_y_block)
(xd->qcoeff, DQC,
xd->dst.y_buffer,
xd->dst.y_stride, xd->eobs);
}
- DEQUANT_INVOKE (&pbi->dequant, idct_add_uv_block)
+ DEQUANT_INVOKE (&pbi->common.rtcd.dequant, idct_add_uv_block)
(xd->qcoeff+16*16, xd->block[16].dequant,
xd->dst.u_buffer, xd->dst.v_buffer,
xd->dst.uv_stride, xd->eobs+16);
diff --git a/vp8/decoder/x86/x86_dsystemdependent.c b/vp8/decoder/x86/x86_dsystemdependent.c
index 91dba7e..27bf5dd 100644
--- a/vp8/decoder/x86/x86_dsystemdependent.c
+++ b/vp8/decoder/x86/x86_dsystemdependent.c
@@ -13,47 +13,7 @@
#include "vpx_ports/x86.h"
#include "vp8/decoder/onyxd_int.h"
-
-#if HAVE_MMX
-void vp8_dequantize_b_impl_mmx(short *sq, short *dq, short *q);
-
-void vp8_dequantize_b_mmx(BLOCKD *d)
-{
- short *sq = (short *) d->qcoeff;
- short *dq = (short *) d->dqcoeff;
- short *q = (short *) d->dequant;
- vp8_dequantize_b_impl_mmx(sq, dq, q);
-}
-#endif
-
void vp8_arch_x86_decode_init(VP8D_COMP *pbi)
{
-#if CONFIG_RUNTIME_CPU_DETECT
- int flags = x86_simd_caps();
- /* Note:
- *
- * This platform can be built without runtime CPU detection as well. If
- * you modify any of the function mappings present in this file, be sure
- * to also update them in static mapings (<arch>/filename_<arch>.h)
- */
- /* Override default functions with fastest ones for this CPU. */
-#if HAVE_MMX
- if (flags & HAS_MMX)
- {
- pbi->dequant.block = vp8_dequantize_b_mmx;
- pbi->dequant.idct_add = vp8_dequant_idct_add_mmx;
- pbi->dequant.idct_add_y_block = vp8_dequant_idct_add_y_block_mmx;
- pbi->dequant.idct_add_uv_block = vp8_dequant_idct_add_uv_block_mmx;
- }
-#endif
-#if HAVE_SSE2
- if (flags & HAS_SSE2)
- {
- pbi->dequant.idct_add_y_block = vp8_dequant_idct_add_y_block_sse2;
- pbi->dequant.idct_add_uv_block = vp8_dequant_idct_add_uv_block_sse2;
- }
-#endif
-
-#endif
}
diff --git a/vp8/encoder/arm/variance_arm.c b/vp8/encoder/arm/variance_arm.c
index e77be9f..7fc7473 100644
--- a/vp8/encoder/arm/variance_arm.c
+++ b/vp8/encoder/arm/variance_arm.c
@@ -11,9 +11,9 @@
#include "vpx_config.h"
#include "vp8/encoder/variance.h"
#include "vp8/common/filter.h"
-#include "vp8/common/arm/bilinearfilter_arm.h"
#if HAVE_ARMV6
+#include "vp8/common/arm/bilinearfilter_arm.h"
unsigned int vp8_sub_pixel_variance8x8_armv6
(
diff --git a/vp8/encoder/bitstream.c b/vp8/encoder/bitstream.c
index 5eea39c..669bfad 100644
--- a/vp8/encoder/bitstream.c
+++ b/vp8/encoder/bitstream.c
@@ -878,7 +878,28 @@
}
}
}
+void vp8_convert_rfct_to_prob(VP8_COMP *const cpi)
+{
+ const int *const rfct = cpi->count_mb_ref_frame_usage;
+ const int rf_intra = rfct[INTRA_FRAME];
+ const int rf_inter = rfct[LAST_FRAME] + rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME];
+ // Calculate the probabilities used to code the ref frame based on useage
+ if (!(cpi->prob_intra_coded = rf_intra * 255 / (rf_intra + rf_inter)))
+ cpi->prob_intra_coded = 1;
+
+ cpi->prob_last_coded = rf_inter ? (rfct[LAST_FRAME] * 255) / rf_inter : 128;
+
+ if (!cpi->prob_last_coded)
+ cpi->prob_last_coded = 1;
+
+ cpi->prob_gf_coded = (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME])
+ ? (rfct[GOLDEN_FRAME] * 255) / (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME]) : 128;
+
+ if (!cpi->prob_gf_coded)
+ cpi->prob_gf_coded = 1;
+
+}
static void pack_inter_mode_mvs(VP8_COMP *const cpi)
{
@@ -886,36 +907,17 @@
vp8_writer *const w = cpi->bc;
const MV_CONTEXT *mvc = pc->fc.mvc;
- const int *const rfct = cpi->count_mb_ref_frame_usage;
- const int rf_intra = rfct[INTRA_FRAME];
- const int rf_inter = rfct[LAST_FRAME] + rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME];
MODE_INFO *m = pc->mi, *ms;
const int mis = pc->mode_info_stride;
int mb_row = -1;
- int prob_last_coded;
- int prob_gf_coded;
int prob_skip_false = 0;
ms = pc->mi - 1;
cpi->mb.partition_info = cpi->mb.pi;
- // Calculate the probabilities to be used to code the reference frame based on actual useage this frame
- if (!(cpi->prob_intra_coded = rf_intra * 255 / (rf_intra + rf_inter)))
- cpi->prob_intra_coded = 1;
-
- prob_last_coded = rf_inter ? (rfct[LAST_FRAME] * 255) / rf_inter : 128;
-
- if (!prob_last_coded)
- prob_last_coded = 1;
-
- prob_gf_coded = (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME])
- ? (rfct[GOLDEN_FRAME] * 255) / (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME]) : 128;
-
- if (!prob_gf_coded)
- prob_gf_coded = 1;
-
+ vp8_convert_rfct_to_prob(cpi);
#ifdef ENTROPY_STATS
active_section = 1;
@@ -936,8 +938,8 @@
}
vp8_write_literal(w, cpi->prob_intra_coded, 8);
- vp8_write_literal(w, prob_last_coded, 8);
- vp8_write_literal(w, prob_gf_coded, 8);
+ vp8_write_literal(w, cpi->prob_last_coded, 8);
+ vp8_write_literal(w, cpi->prob_gf_coded, 8);
update_mbintra_mode_probs(cpi);
@@ -999,11 +1001,11 @@
vp8_write(w, 1, cpi->prob_intra_coded);
if (rf == LAST_FRAME)
- vp8_write(w, 0, prob_last_coded);
+ vp8_write(w, 0, cpi->prob_last_coded);
else
{
- vp8_write(w, 1, prob_last_coded);
- vp8_write(w, (rf == GOLDEN_FRAME) ? 0 : 1, prob_gf_coded);
+ vp8_write(w, 1, cpi->prob_last_coded);
+ vp8_write(w, (rf == GOLDEN_FRAME) ? 0 : 1, cpi->prob_gf_coded);
}
{
@@ -1204,7 +1206,7 @@
{
for (j=0; j < PREV_COEF_CONTEXTS; ++j)
{
- const int tmp = out[i];
+ const unsigned int tmp = out[i];
out[i] += probs[j][i];
/* check for wrap */
if (out[i] < tmp)
@@ -1355,6 +1357,24 @@
return savings;
}
+void vp8_calc_ref_frame_costs(int *ref_frame_cost,
+ int prob_intra,
+ int prob_last,
+ int prob_garf
+ )
+{
+ ref_frame_cost[INTRA_FRAME] = vp8_cost_zero(prob_intra);
+ ref_frame_cost[LAST_FRAME] = vp8_cost_one(prob_intra)
+ + vp8_cost_zero(prob_last);
+ ref_frame_cost[GOLDEN_FRAME] = vp8_cost_one(prob_intra)
+ + vp8_cost_one(prob_last)
+ + vp8_cost_zero(prob_garf);
+ ref_frame_cost[ALTREF_FRAME] = vp8_cost_one(prob_intra)
+ + vp8_cost_one(prob_last)
+ + vp8_cost_one(prob_garf);
+
+}
+
int vp8_estimate_entropy_savings(VP8_COMP *cpi)
{
int savings = 0;
@@ -1362,7 +1382,7 @@
const int *const rfct = cpi->count_mb_ref_frame_usage;
const int rf_intra = rfct[INTRA_FRAME];
const int rf_inter = rfct[LAST_FRAME] + rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME];
- int new_intra, new_last, gf_last, oldtotal, newtotal;
+ int new_intra, new_last, new_garf, oldtotal, newtotal;
int ref_frame_cost[MAX_REF_FRAMES];
vp8_clear_system_state(); //__asm emms;
@@ -1374,19 +1394,11 @@
new_last = rf_inter ? (rfct[LAST_FRAME] * 255) / rf_inter : 128;
- gf_last = (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME])
+ new_garf = (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME])
? (rfct[GOLDEN_FRAME] * 255) / (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME]) : 128;
- // new costs
- ref_frame_cost[INTRA_FRAME] = vp8_cost_zero(new_intra);
- ref_frame_cost[LAST_FRAME] = vp8_cost_one(new_intra)
- + vp8_cost_zero(new_last);
- ref_frame_cost[GOLDEN_FRAME] = vp8_cost_one(new_intra)
- + vp8_cost_one(new_last)
- + vp8_cost_zero(gf_last);
- ref_frame_cost[ALTREF_FRAME] = vp8_cost_one(new_intra)
- + vp8_cost_one(new_last)
- + vp8_cost_one(gf_last);
+
+ vp8_calc_ref_frame_costs(ref_frame_cost,new_intra,new_last,new_garf);
newtotal =
rfct[INTRA_FRAME] * ref_frame_cost[INTRA_FRAME] +
@@ -1396,15 +1408,8 @@
// old costs
- ref_frame_cost[INTRA_FRAME] = vp8_cost_zero(cpi->prob_intra_coded);
- ref_frame_cost[LAST_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_zero(cpi->prob_last_coded);
- ref_frame_cost[GOLDEN_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_one(cpi->prob_last_coded)
- + vp8_cost_zero(cpi->prob_gf_coded);
- ref_frame_cost[ALTREF_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_one(cpi->prob_last_coded)
- + vp8_cost_one(cpi->prob_gf_coded);
+ vp8_calc_ref_frame_costs(ref_frame_cost,cpi->prob_intra_coded,
+ cpi->prob_last_coded,cpi->prob_gf_coded);
oldtotal =
rfct[INTRA_FRAME] * ref_frame_cost[INTRA_FRAME] +
@@ -1639,20 +1644,20 @@
// Signal whether or not Segmentation is enabled
- vp8_write_bit(bc, (xd->segmentation_enabled) ? 1 : 0);
+ vp8_write_bit(bc, xd->segmentation_enabled);
// Indicate which features are enabled
if (xd->segmentation_enabled)
{
// Signal whether or not the segmentation map is being updated.
- vp8_write_bit(bc, (xd->update_mb_segmentation_map) ? 1 : 0);
- vp8_write_bit(bc, (xd->update_mb_segmentation_data) ? 1 : 0);
+ vp8_write_bit(bc, xd->update_mb_segmentation_map);
+ vp8_write_bit(bc, xd->update_mb_segmentation_data);
if (xd->update_mb_segmentation_data)
{
signed char Data;
- vp8_write_bit(bc, (xd->mb_segement_abs_delta) ? 1 : 0);
+ vp8_write_bit(bc, xd->mb_segement_abs_delta);
// For each segmentation feature (Quant and loop filter level)
for (i = 0; i < MB_LVL_MAX; i++)
@@ -1709,7 +1714,7 @@
vp8_write_literal(bc, pc->sharpness_level, 3);
// Write out loop filter deltas applied at the MB level based on mode or ref frame (if they are enabled).
- vp8_write_bit(bc, (xd->mode_ref_lf_delta_enabled) ? 1 : 0);
+ vp8_write_bit(bc, xd->mode_ref_lf_delta_enabled);
if (xd->mode_ref_lf_delta_enabled)
{
diff --git a/vp8/encoder/block.h b/vp8/encoder/block.h
index 5e5a60d..0a74ca4 100644
--- a/vp8/encoder/block.h
+++ b/vp8/encoder/block.h
@@ -45,10 +45,6 @@
unsigned char **base_src;
int src;
int src_stride;
-
-// MV enc_mv;
- int force_empty;
-
} BLOCK;
typedef struct
@@ -107,7 +103,6 @@
int mv_row_min;
int mv_row_max;
- int vector_range; // Used to monitor limiting range of recent vectors to guide search.
int skip;
int encode_breakout;
diff --git a/vp8/encoder/encodeframe.c b/vp8/encoder/encodeframe.c
index 0927f51..6a9ba29 100644
--- a/vp8/encoder/encodeframe.c
+++ b/vp8/encoder/encodeframe.c
@@ -39,7 +39,12 @@
#define IF_RTCD(x) NULL
#endif
extern void vp8_stuff_mb(VP8_COMP *cpi, MACROBLOCKD *x, TOKENEXTRA **t) ;
-
+extern void vp8_calc_ref_frame_costs(int *ref_frame_cost,
+ int prob_intra,
+ int prob_last,
+ int prob_garf
+ );
+extern void vp8_convert_rfct_to_prob(VP8_COMP *const cpi);
extern void vp8cx_initialize_me_consts(VP8_COMP *cpi, int QIndex);
extern void vp8_auto_select_speed(VP8_COMP *cpi);
extern void vp8cx_init_mbrthread_data(VP8_COMP *cpi,
@@ -590,8 +595,6 @@
// Activity map pointer
x->mb_activity_ptr = cpi->mb_activity_map;
- x->vector_range = 32;
-
x->act_zbin_adj = 0;
x->partition_info = x->pi;
@@ -636,55 +639,23 @@
vpx_memset(cm->above_context, 0,
sizeof(ENTROPY_CONTEXT_PLANES) * cm->mb_cols);
- xd->ref_frame_cost[INTRA_FRAME] = vp8_cost_zero(cpi->prob_intra_coded);
-
// Special case treatment when GF and ARF are not sensible options for reference
if (cpi->ref_frame_flags == VP8_LAST_FLAG)
- {
- xd->ref_frame_cost[LAST_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_zero(255);
- xd->ref_frame_cost[GOLDEN_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_one(255)
- + vp8_cost_zero(128);
- xd->ref_frame_cost[ALTREF_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_one(255)
- + vp8_cost_one(128);
- }
+ vp8_calc_ref_frame_costs(xd->ref_frame_cost,
+ cpi->prob_intra_coded,255,128);
else if ((cpi->oxcf.number_of_layers > 1) &&
(cpi->ref_frame_flags == VP8_GOLD_FLAG))
- {
- xd->ref_frame_cost[LAST_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_zero(1);
- xd->ref_frame_cost[GOLDEN_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_one(1)
- + vp8_cost_zero(255);
- xd->ref_frame_cost[ALTREF_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_one(1)
- + vp8_cost_one(255);
- }
+ vp8_calc_ref_frame_costs(xd->ref_frame_cost,
+ cpi->prob_intra_coded,1,255);
else if ((cpi->oxcf.number_of_layers > 1) &&
(cpi->ref_frame_flags == VP8_ALT_FLAG))
- {
- xd->ref_frame_cost[LAST_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_zero(1);
- xd->ref_frame_cost[GOLDEN_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_one(1)
- + vp8_cost_zero(1);
- xd->ref_frame_cost[ALTREF_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_one(1)
- + vp8_cost_one(1);
- }
+ vp8_calc_ref_frame_costs(xd->ref_frame_cost,
+ cpi->prob_intra_coded,1,1);
else
- {
- xd->ref_frame_cost[LAST_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_zero(cpi->prob_last_coded);
- xd->ref_frame_cost[GOLDEN_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_one(cpi->prob_last_coded)
- + vp8_cost_zero(cpi->prob_gf_coded);
- xd->ref_frame_cost[ALTREF_FRAME] = vp8_cost_one(cpi->prob_intra_coded)
- + vp8_cost_one(cpi->prob_last_coded)
- + vp8_cost_one(cpi->prob_gf_coded);
- }
+ vp8_calc_ref_frame_costs(xd->ref_frame_cost,
+ cpi->prob_intra_coded,
+ cpi->prob_last_coded,
+ cpi->prob_gf_coded);
xd->fullpixel_mask = 0xffffffff;
if(cm->full_pixel)
@@ -966,31 +937,7 @@
if ((cm->frame_type != KEY_FRAME) && ((cpi->oxcf.number_of_layers > 1) ||
(!cm->refresh_alt_ref_frame && !cm->refresh_golden_frame)))
{
- const int *const rfct = cpi->count_mb_ref_frame_usage;
- const int rf_intra = rfct[INTRA_FRAME];
- const int rf_inter = rfct[LAST_FRAME] + rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME];
-
- if ((rf_intra + rf_inter) > 0)
- {
- cpi->prob_intra_coded = (rf_intra * 255) / (rf_intra + rf_inter);
-
- if (cpi->prob_intra_coded < 1)
- cpi->prob_intra_coded = 1;
-
- if ((cm->frames_since_golden > 0) || cpi->source_alt_ref_active)
- {
- cpi->prob_last_coded = rf_inter ? (rfct[LAST_FRAME] * 255) / rf_inter : 128;
-
- if (cpi->prob_last_coded < 1)
- cpi->prob_last_coded = 1;
-
- cpi->prob_gf_coded = (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME])
- ? (rfct[GOLDEN_FRAME] * 255) / (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME]) : 128;
-
- if (cpi->prob_gf_coded < 1)
- cpi->prob_gf_coded = 1;
- }
- }
+ vp8_convert_rfct_to_prob(cpi);
}
#if 0
@@ -1142,8 +1089,10 @@
#endif
}
-int vp8cx_encode_intra_macro_block(VP8_COMP *cpi, MACROBLOCK *x, TOKENEXTRA **t, int mb_row, int mb_col)
+int vp8cx_encode_intra_macro_block(VP8_COMP *cpi, MACROBLOCK *x, TOKENEXTRA **t,
+ int mb_row, int mb_col)
{
+ MACROBLOCKD *xd = &x->e_mbd;
int rate;
if (cpi->sf.RD && cpi->compressor_speed != 2)
@@ -1163,14 +1112,17 @@
vp8_encode_intra16x16mby(IF_RTCD(&cpi->rtcd), x);
vp8_encode_intra16x16mbuv(IF_RTCD(&cpi->rtcd), x);
+
sum_intra_stats(cpi, x);
vp8_tokenize_mb(cpi, &x->e_mbd, t);
- if (x->e_mbd.mode_info_context->mbmi.mode != B_PRED)
- vp8_inverse_transform_mby(IF_RTCD(&cpi->rtcd.common->idct), &x->e_mbd);
+ if (xd->mode_info_context->mbmi.mode != B_PRED)
+ vp8_inverse_transform_mby(xd, IF_RTCD(&cpi->common.rtcd));
- vp8_inverse_transform_mbuv(IF_RTCD(&cpi->rtcd.common->idct), &x->e_mbd);
-
+ DEQUANT_INVOKE (&cpi->common.rtcd.dequant, idct_add_uv_block)
+ (xd->qcoeff+16*16, xd->block[16].dequant,
+ xd->dst.u_buffer, xd->dst.v_buffer,
+ xd->dst.uv_stride, xd->eobs+16);
return rate;
}
#ifdef SPEEDSTATS
@@ -1232,23 +1184,8 @@
}
else
{
-#if CONFIG_MULTI_RES_ENCODING
- if (cpi->oxcf.mr_encoder_id == 0)
- {
- /* Lowest-resolution encoding */
- vp8_pick_inter_mode(cpi, x, recon_yoffset, recon_uvoffset, &rate,
- &distortion, &intra_error);
-
- }else
- {
- /* Higher-resolution encoding */
- vp8_mr_pick_inter_mode(cpi, x, recon_yoffset, recon_uvoffset, &rate,
- &distortion, &intra_error, mb_row, mb_col);
- }
-#else
vp8_pick_inter_mode(cpi, x, recon_yoffset, recon_uvoffset, &rate,
- &distortion, &intra_error);
-#endif
+ &distortion, &intra_error, mb_row, mb_col);
}
cpi->prediction_error += distortion;
@@ -1363,12 +1300,14 @@
if (!x->skip)
{
vp8_tokenize_mb(cpi, xd, t);
- if (x->e_mbd.mode_info_context->mbmi.mode != B_PRED)
- {
- vp8_inverse_transform_mby(IF_RTCD(&cpi->rtcd.common->idct),
- &x->e_mbd);
- }
- vp8_inverse_transform_mbuv(IF_RTCD(&cpi->rtcd.common->idct), &x->e_mbd);
+
+ if (xd->mode_info_context->mbmi.mode != B_PRED)
+ vp8_inverse_transform_mby(xd, IF_RTCD(&cpi->common.rtcd));
+
+ DEQUANT_INVOKE (&cpi->common.rtcd.dequant, idct_add_uv_block)
+ (xd->qcoeff+16*16, xd->block[16].dequant,
+ xd->dst.u_buffer, xd->dst.v_buffer,
+ xd->dst.uv_stride, xd->eobs+16);
}
else
{
diff --git a/vp8/encoder/encodeintra.c b/vp8/encoder/encodeintra.c
index a3b800a..16393a1 100644
--- a/vp8/encoder/encodeintra.c
+++ b/vp8/encoder/encodeintra.c
@@ -18,7 +18,6 @@
#include "vp8/common/invtrans.h"
#include "vp8/common/recon.h"
#include "dct.h"
-#include "vp8/common/g_common.h"
#include "encodeintra.h"
@@ -45,7 +44,7 @@
vp8_encode_intra16x16mby(rtcd, x);
- vp8_inverse_transform_mby(IF_RTCD(&rtcd->common->idct), &x->e_mbd);
+ vp8_inverse_transform_mby(&x->e_mbd, IF_RTCD(&cpi->common.rtcd));
}
else
{
@@ -77,8 +76,17 @@
x->quantize_b(be, b);
- vp8_inverse_transform_b(IF_RTCD(&rtcd->common->idct), b, 16);
-
+ if (*b->eob > 1)
+ {
+ IDCT_INVOKE(IF_RTCD(&rtcd->common->idct), idct16)(b->dqcoeff,
+ b->predictor, 16, *(b->base_dst) + b->dst, b->dst_stride);
+ }
+ else
+ {
+ IDCT_INVOKE(IF_RTCD(&rtcd->common->idct), idct1_scalar_add)
+ (b->dqcoeff[0], b->predictor, 16, *(b->base_dst) + b->dst,
+ b->dst_stride);
+ }
}
void vp8_encode_intra4x4mby(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *mb)
@@ -96,11 +104,12 @@
void vp8_encode_intra16x16mby(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x)
{
BLOCK *b = &x->block[0];
+ MACROBLOCKD *xd = &x->e_mbd;
- RECON_INVOKE(&rtcd->common->recon, build_intra_predictors_mby)(&x->e_mbd);
+ RECON_INVOKE(&rtcd->common->recon, build_intra_predictors_mby_s)(&x->e_mbd);
- ENCODEMB_INVOKE(&rtcd->encodemb, submby)(x->src_diff, *(b->base_src),
- b->src_stride, x->e_mbd.predictor, 16);
+ ENCODEMB_INVOKE(&rtcd->encodemb, submby) (x->src_diff, *(b->base_src),
+ b->src_stride, xd->dst.y_buffer, xd->dst.y_stride);
vp8_transform_intra_mby(x);
@@ -108,16 +117,17 @@
if (x->optimize)
vp8_optimize_mby(x, rtcd);
-
}
void vp8_encode_intra16x16mbuv(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x)
{
- RECON_INVOKE(&rtcd->common->recon, build_intra_predictors_mbuv)(&x->e_mbd);
+ MACROBLOCKD *xd = &x->e_mbd;
+
+ RECON_INVOKE(&rtcd->common->recon, build_intra_predictors_mbuv_s)(&x->e_mbd);
ENCODEMB_INVOKE(&rtcd->encodemb, submbuv)(x->src_diff, x->src.u_buffer,
- x->src.v_buffer, x->src.uv_stride, &x->e_mbd.predictor[256],
- &x->e_mbd.predictor[320], 8);
+ x->src.v_buffer, x->src.uv_stride, xd->dst.u_buffer,
+ xd->dst.v_buffer, xd->dst.uv_stride);
vp8_transform_mbuv(x);
@@ -125,5 +135,4 @@
if (x->optimize)
vp8_optimize_mbuv(x, rtcd);
-
}
diff --git a/vp8/encoder/encodemb.c b/vp8/encoder/encodemb.c
index e9042e1..c9f7553 100644
--- a/vp8/encoder/encodemb.c
+++ b/vp8/encoder/encodemb.c
@@ -105,10 +105,10 @@
BLOCK *b = &x->block[0];
ENCODEMB_INVOKE(&rtcd->encodemb, submby)(x->src_diff, *(b->base_src),
- b->src_stride, x->e_mbd.predictor, 16);
+ b->src_stride, x->e_mbd.dst.y_buffer, x->e_mbd.dst.y_stride);
ENCODEMB_INVOKE(&rtcd->encodemb, submbuv)(x->src_diff, x->src.u_buffer,
- x->src.v_buffer, x->src.uv_stride, &x->e_mbd.predictor[256],
- &x->e_mbd.predictor[320], 8);
+ x->src.v_buffer, x->src.uv_stride, x->e_mbd.dst.u_buffer,
+ x->e_mbd.dst.v_buffer, x->e_mbd.dst.uv_stride);
}
static void build_dcblock(MACROBLOCK *x)
@@ -625,7 +625,7 @@
void vp8_encode_inter16x16(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x)
{
- vp8_build_inter_predictors_mb_e(&x->e_mbd);
+ vp8_build_inter_predictors_mb(&x->e_mbd);
vp8_subtract_mb(rtcd, x);
@@ -635,7 +635,6 @@
if (x->optimize)
optimize_mb(x, rtcd);
-
}
/* this funciton is used by first pass only */
@@ -643,15 +642,15 @@
{
BLOCK *b = &x->block[0];
- vp8_build_inter16x16_predictors_mby(&x->e_mbd);
+ vp8_build_inter16x16_predictors_mby(&x->e_mbd, x->e_mbd.dst.y_buffer,
+ x->e_mbd.dst.y_stride);
ENCODEMB_INVOKE(&rtcd->encodemb, submby)(x->src_diff, *(b->base_src),
- b->src_stride, x->e_mbd.predictor, 16);
+ b->src_stride, x->e_mbd.dst.y_buffer, x->e_mbd.dst.y_stride);
transform_mby(x);
vp8_quantize_mby(x);
- vp8_inverse_transform_mby(IF_RTCD(&rtcd->common->idct), &x->e_mbd);
-
+ vp8_inverse_transform_mby(&x->e_mbd, IF_RTCD(rtcd->common));
}
diff --git a/vp8/encoder/encodemb.h b/vp8/encoder/encodemb.h
index 597a57b..0fa87cf 100644
--- a/vp8/encoder/encodemb.h
+++ b/vp8/encoder/encodemb.h
@@ -12,6 +12,7 @@
#ifndef __INC_ENCODEMB_H
#define __INC_ENCODEMB_H
+
#include "vpx_config.h"
#include "block.h"
diff --git a/vp8/encoder/ethreading.c b/vp8/encoder/ethreading.c
index 8c49668..6965598 100644
--- a/vp8/encoder/ethreading.c
+++ b/vp8/encoder/ethreading.c
@@ -38,7 +38,7 @@
if (sem_wait(&cpi->h_event_start_lpf) == 0)
{
- if (cpi->b_multi_threaded == FALSE) // we're shutting down
+ if (cpi->b_multi_threaded == 0) // we're shutting down
break;
loopfilter_frame(cpi, cm);
@@ -78,7 +78,7 @@
int *segment_counts = mbri->segment_counts;
int *totalrate = &mbri->totalrate;
- if (cpi->b_multi_threaded == FALSE) // we're shutting down
+ if (cpi->b_multi_threaded == 0) // we're shutting down
break;
for (mb_row = ithread + 1; mb_row < cm->mb_rows; mb_row += (cpi->encoding_thread_count + 1))
@@ -302,7 +302,6 @@
z->mv_col_max = x->mv_col_max;
z->mv_row_min = x->mv_row_min;
z->mv_row_max = x->mv_row_max;
- z->vector_range = x->vector_range ;
*/
z->vp8_short_fdct4x4 = x->vp8_short_fdct4x4;
@@ -350,8 +349,6 @@
z->block[i].src = x->block[i].src;
*/
z->block[i].src_stride = x->block[i].src_stride;
- z->block[i].force_empty = x->block[i].force_empty;
-
}
{
@@ -421,8 +418,6 @@
#endif
mb->gf_active_ptr = x->gf_active_ptr;
- mb->vector_range = 32;
-
vpx_memset(mbr_ei[i].segment_counts, 0, sizeof(mbr_ei[i].segment_counts));
mbr_ei[i].totalrate = 0;
diff --git a/vp8/encoder/firstpass.c b/vp8/encoder/firstpass.c
index 8d19d1e..346c06f 100644
--- a/vp8/encoder/firstpass.c
+++ b/vp8/encoder/firstpass.c
@@ -267,8 +267,8 @@
// Calculate a modified Error used in distributing bits between easier and harder frames
static double calculate_modified_err(VP8_COMP *cpi, FIRSTPASS_STATS *this_frame)
{
- double av_err = ( cpi->twopass.total_stats->ssim_weighted_pred_err /
- cpi->twopass.total_stats->count );
+ double av_err = ( cpi->twopass.total_stats.ssim_weighted_pred_err /
+ cpi->twopass.total_stats.count );
double this_err = this_frame->ssim_weighted_pred_err;
double modified_err;
@@ -373,7 +373,7 @@
else
{
// For VBR base this on the bits and frames left plus the two_pass_vbrmax_section rate passed in by the user
- max_bits = (int)(((double)cpi->twopass.bits_left / (cpi->twopass.total_stats->count - (double)cpi->common.current_video_frame)) * ((double)cpi->oxcf.two_pass_vbrmax_section / 100.0));
+ max_bits = (int)(((double)cpi->twopass.bits_left / (cpi->twopass.total_stats.count - (double)cpi->common.current_video_frame)) * ((double)cpi->oxcf.two_pass_vbrmax_section / 100.0));
}
// Trap case where we are out of bits
@@ -385,12 +385,12 @@
void vp8_init_first_pass(VP8_COMP *cpi)
{
- zero_stats(cpi->twopass.total_stats);
+ zero_stats(&cpi->twopass.total_stats);
}
void vp8_end_first_pass(VP8_COMP *cpi)
{
- output_stats(cpi, cpi->output_pkt_list, cpi->twopass.total_stats);
+ output_stats(cpi, cpi->output_pkt_list, &cpi->twopass.total_stats);
}
static void zz_motion_search( VP8_COMP *cpi, MACROBLOCK * x, YV12_BUFFER_CONFIG * recon_buffer, int * best_motion_err, int recon_yoffset )
@@ -804,17 +804,17 @@
- cpi->source->ts_start;
// don't want to do output stats with a stack variable!
- memcpy(cpi->twopass.this_frame_stats,
+ memcpy(&cpi->twopass.this_frame_stats,
&fps,
sizeof(FIRSTPASS_STATS));
- output_stats(cpi, cpi->output_pkt_list, cpi->twopass.this_frame_stats);
- accumulate_stats(cpi->twopass.total_stats, &fps);
+ output_stats(cpi, cpi->output_pkt_list, &cpi->twopass.this_frame_stats);
+ accumulate_stats(&cpi->twopass.total_stats, &fps);
}
// Copy the previous Last Frame into the GF buffer if specific conditions for doing so are met
if ((cm->current_video_frame > 0) &&
- (cpi->twopass.this_frame_stats->pcnt_inter > 0.20) &&
- ((cpi->twopass.this_frame_stats->intra_error / cpi->twopass.this_frame_stats->coded_error) > 2.0))
+ (cpi->twopass.this_frame_stats.pcnt_inter > 0.20) &&
+ ((cpi->twopass.this_frame_stats.intra_error / cpi->twopass.this_frame_stats.coded_error) > 2.0))
{
vp8_yv12_copy_frame_ptr(lst_yv12, gld_yv12);
}
@@ -861,7 +861,7 @@
{
return -(log( prob ) / log( 2.0 ));
}
-static long long estimate_modemvcost(VP8_COMP *cpi,
+static int64_t estimate_modemvcost(VP8_COMP *cpi,
FIRSTPASS_STATS * fpstats)
{
int mv_cost;
@@ -1019,7 +1019,7 @@
// averaga q observed in clip for non kf/gf.arf frames
// Give average a chance to settle though.
if ( (cpi->ni_frames >
- ((unsigned int)cpi->twopass.total_stats->count >> 8)) &&
+ ((unsigned int)cpi->twopass.total_stats.count >> 8)) &&
(cpi->ni_frames > 150) )
{
cpi->twopass.maxq_max_limit = ((cpi->ni_av_qi + 32) < cpi->worst_quality)
@@ -1075,8 +1075,8 @@
}
// II ratio correction factor for clip as a whole
- clip_iiratio = cpi->twopass.total_stats->intra_error /
- DOUBLE_DIVIDE_CHECK(cpi->twopass.total_stats->coded_error);
+ clip_iiratio = cpi->twopass.total_stats.intra_error /
+ DOUBLE_DIVIDE_CHECK(cpi->twopass.total_stats.coded_error);
clip_iifactor = 1.0 - ((clip_iiratio - 10.0) * 0.025);
if (clip_iifactor < 0.80)
clip_iifactor = 0.80;
@@ -1260,25 +1260,25 @@
double two_pass_min_rate = (double)(cpi->oxcf.target_bandwidth * cpi->oxcf.two_pass_vbrmin_section / 100);
- zero_stats(cpi->twopass.total_stats);
- zero_stats(cpi->twopass.total_left_stats);
+ zero_stats(&cpi->twopass.total_stats);
+ zero_stats(&cpi->twopass.total_left_stats);
if (!cpi->twopass.stats_in_end)
return;
- *cpi->twopass.total_stats = *cpi->twopass.stats_in_end;
- *cpi->twopass.total_left_stats = *cpi->twopass.total_stats;
+ cpi->twopass.total_stats = *cpi->twopass.stats_in_end;
+ cpi->twopass.total_left_stats = cpi->twopass.total_stats;
// each frame can have a different duration, as the frame rate in the source
// isn't guaranteed to be constant. The frame rate prior to the first frame
// encoded in the second pass is a guess. However the sum duration is not.
// Its calculated based on the actual durations of all frames from the first
// pass.
- vp8_new_frame_rate(cpi, 10000000.0 * cpi->twopass.total_stats->count / cpi->twopass.total_stats->duration);
+ vp8_new_frame_rate(cpi, 10000000.0 * cpi->twopass.total_stats.count / cpi->twopass.total_stats.duration);
cpi->output_frame_rate = cpi->frame_rate;
- cpi->twopass.bits_left = (int64_t)(cpi->twopass.total_stats->duration * cpi->oxcf.target_bandwidth / 10000000.0) ;
- cpi->twopass.bits_left -= (int64_t)(cpi->twopass.total_stats->duration * two_pass_min_rate / 10000000.0);
+ cpi->twopass.bits_left = (int64_t)(cpi->twopass.total_stats.duration * cpi->oxcf.target_bandwidth / 10000000.0) ;
+ cpi->twopass.bits_left -= (int64_t)(cpi->twopass.total_stats.duration * two_pass_min_rate / 10000000.0);
// Calculate a minimum intra value to be used in determining the IIratio
// scores used in the second pass. We have this minimum to make sure
@@ -1301,7 +1301,7 @@
sum_iiratio += IIRatio;
}
- cpi->twopass.avg_iiratio = sum_iiratio / DOUBLE_DIVIDE_CHECK((double)cpi->twopass.total_stats->count);
+ cpi->twopass.avg_iiratio = sum_iiratio / DOUBLE_DIVIDE_CHECK((double)cpi->twopass.total_stats.count);
// Reset file position
reset_fpf_position(cpi, start_pos);
@@ -1376,7 +1376,7 @@
double loop_decay_rate,
double decay_accumulator )
{
- BOOL trans_to_still = FALSE;
+ int trans_to_still = 0;
// Break clause to detect very still sections after motion
// For example a static image after a fade or other transition
@@ -1406,7 +1406,7 @@
// Only if it does do we signal a transition to still
if ( j == still_interval )
- trans_to_still = TRUE;
+ trans_to_still = 1;
}
return trans_to_still;
@@ -1415,14 +1415,14 @@
// This function detects a flash through the high relative pcnt_second_ref
// score in the frame following a flash frame. The offset passed in should
// reflect this
-static BOOL detect_flash( VP8_COMP *cpi, int offset )
+static int detect_flash( VP8_COMP *cpi, int offset )
{
FIRSTPASS_STATS next_frame;
- BOOL flash_detected = FALSE;
+ int flash_detected = 0;
// Read the frame data.
- // The return is FALSE (no flash detected) if not a valid frame
+ // The return is 0 (no flash detected) if not a valid frame
if ( read_frame_stats(cpi, &next_frame, offset) != EOF )
{
// What we are looking for here is a situation where there is a
@@ -1433,7 +1433,7 @@
if ( (next_frame.pcnt_second_ref > next_frame.pcnt_inter) &&
(next_frame.pcnt_second_ref >= 0.5 ) )
{
- flash_detected = TRUE;
+ flash_detected = 1;
/*if (1)
{
@@ -1548,7 +1548,7 @@
double mv_in_out_accumulator = 0.0;
double abs_mv_in_out_accumulator = 0.0;
double r;
- BOOL flash_detected = FALSE;
+ int flash_detected = 0;
// Search forward from the proposed arf/next gf position
for ( i = 0; i < f_frames; i++ )
@@ -1677,7 +1677,7 @@
int alt_boost = 0;
int f_boost = 0;
int b_boost = 0;
- BOOL flash_detected;
+ int flash_detected;
cpi->twopass.gf_group_bits = 0;
cpi->twopass.gf_decay_rate = 0;
@@ -1751,7 +1751,7 @@
loop_decay_rate,
decay_accumulator ) )
{
- allow_alt_ref = FALSE;
+ allow_alt_ref = 0;
boost_score = old_boost_score;
break;
}
@@ -1923,7 +1923,7 @@
int frames_bwd = cpi->oxcf.arnr_max_frames - 1;
int frames_fwd = cpi->oxcf.arnr_max_frames - 1;
- cpi->source_alt_ref_pending = TRUE;
+ cpi->source_alt_ref_pending = 1;
// For alt ref frames the error score for the end frame of the
// group (the alt ref frame) should not contribute to the group
@@ -1949,7 +1949,7 @@
// Note: this_frame->frame has been updated in the loop
// so it now points at the ARF frame.
half_gf_int = cpi->baseline_gf_interval >> 1;
- frames_after_arf = cpi->twopass.total_stats->count -
+ frames_after_arf = cpi->twopass.total_stats.count -
this_frame->frame - 1;
switch (cpi->oxcf.arnr_type)
@@ -1989,13 +1989,13 @@
}
else
{
- cpi->source_alt_ref_pending = FALSE;
+ cpi->source_alt_ref_pending = 0;
cpi->baseline_gf_interval = i;
}
}
else
{
- cpi->source_alt_ref_pending = FALSE;
+ cpi->source_alt_ref_pending = 0;
cpi->baseline_gf_interval = i;
}
@@ -2005,7 +2005,7 @@
// where cpi->twopass.kf_group_bits is tied to cpi->twopass.bits_left.
// This is also important for short clips where there may only be one
// key frame.
- if (cpi->twopass.frames_to_key >= (int)(cpi->twopass.total_stats->count -
+ if (cpi->twopass.frames_to_key >= (int)(cpi->twopass.total_stats.count -
cpi->common.current_video_frame))
{
cpi->twopass.kf_group_bits =
@@ -2296,7 +2296,7 @@
void vp8_second_pass(VP8_COMP *cpi)
{
int tmp_q;
- int frames_left = (int)(cpi->twopass.total_stats->count - cpi->common.current_video_frame);
+ int frames_left = (int)(cpi->twopass.total_stats.count - cpi->common.current_video_frame);
FIRSTPASS_STATS this_frame = {0};
FIRSTPASS_STATS this_frame_copy;
@@ -2341,7 +2341,7 @@
cpi->twopass.gf_group_error_left = cpi->twopass.kf_group_error_left;
cpi->baseline_gf_interval = cpi->twopass.frames_to_key;
cpi->frames_till_gf_update_due = cpi->baseline_gf_interval;
- cpi->source_alt_ref_pending = FALSE;
+ cpi->source_alt_ref_pending = 0;
}
}
@@ -2411,7 +2411,7 @@
// Account for mv, mode and other overheads.
overhead_bits = estimate_modemvcost(
- cpi, cpi->twopass.total_left_stats );
+ cpi, &cpi->twopass.total_left_stats );
// Special case code for first frame.
if (cpi->common.current_video_frame == 0)
@@ -2425,7 +2425,7 @@
est_cq =
estimate_cq( cpi,
- cpi->twopass.total_left_stats,
+ &cpi->twopass.total_left_stats,
(int)(cpi->twopass.bits_left / frames_left),
overhead_bits );
@@ -2440,7 +2440,7 @@
tmp_q = estimate_max_q(
cpi,
- cpi->twopass.total_left_stats,
+ &cpi->twopass.total_left_stats,
(int)(cpi->twopass.bits_left / frames_left),
overhead_bits );
@@ -2463,16 +2463,16 @@
// radical adjustments to the allowed quantizer range just to use up a
// few surplus bits or get beneath the target rate.
else if ( (cpi->common.current_video_frame <
- (((unsigned int)cpi->twopass.total_stats->count * 255)>>8)) &&
+ (((unsigned int)cpi->twopass.total_stats.count * 255)>>8)) &&
((cpi->common.current_video_frame + cpi->baseline_gf_interval) <
- (unsigned int)cpi->twopass.total_stats->count) )
+ (unsigned int)cpi->twopass.total_stats.count) )
{
if (frames_left < 1)
frames_left = 1;
tmp_q = estimate_max_q(
cpi,
- cpi->twopass.total_left_stats,
+ &cpi->twopass.total_left_stats,
(int)(cpi->twopass.bits_left / frames_left),
overhead_bits );
@@ -2489,13 +2489,13 @@
cpi->twopass.frames_to_key --;
// Update the total stats remaining sturcture
- subtract_stats(cpi->twopass.total_left_stats, &this_frame );
+ subtract_stats(&cpi->twopass.total_left_stats, &this_frame );
}
-static BOOL test_candidate_kf(VP8_COMP *cpi, FIRSTPASS_STATS *last_frame, FIRSTPASS_STATS *this_frame, FIRSTPASS_STATS *next_frame)
+static int test_candidate_kf(VP8_COMP *cpi, FIRSTPASS_STATS *last_frame, FIRSTPASS_STATS *this_frame, FIRSTPASS_STATS *next_frame)
{
- BOOL is_viable_kf = FALSE;
+ int is_viable_kf = 0;
// Does the frame satisfy the primary criteria of a key frame
// If so, then examine how well it predicts subsequent frames
@@ -2569,13 +2569,13 @@
// If there is tolerable prediction for at least the next 3 frames then break out else discard this pottential key frame and move on
if (boost_score > 5.0 && (i > 3))
- is_viable_kf = TRUE;
+ is_viable_kf = 1;
else
{
// Reset the file position
reset_fpf_position(cpi, start_pos);
- is_viable_kf = FALSE;
+ is_viable_kf = 0;
}
}
@@ -2611,7 +2611,7 @@
cpi->this_key_frame_forced = cpi->next_key_frame_forced;
// Clear the alt ref active flag as this can never be active on a key frame
- cpi->source_alt_ref_active = FALSE;
+ cpi->source_alt_ref_active = 0;
// Kf is always a gf so clear frames till next gf counter
cpi->frames_till_gf_update_due = 0;
@@ -2727,10 +2727,10 @@
// Reset to the start of the group
reset_fpf_position(cpi, current_pos);
- cpi->next_key_frame_forced = TRUE;
+ cpi->next_key_frame_forced = 1;
}
else
- cpi->next_key_frame_forced = FALSE;
+ cpi->next_key_frame_forced = 0;
// Special case for the last frame of the file
if (cpi->twopass.stats_in >= cpi->twopass.stats_in_end)
@@ -3034,8 +3034,8 @@
if (cpi->oxcf.allow_spatial_resampling)
{
- int resample_trigger = FALSE;
- int last_kf_resampled = FALSE;
+ int resample_trigger = 0;
+ int last_kf_resampled = 0;
int kf_q;
int scale_val = 0;
int hr, hs, vr, vs;
@@ -3053,14 +3053,14 @@
double effective_size_ratio;
if ((cpi->common.Width != cpi->oxcf.Width) || (cpi->common.Height != cpi->oxcf.Height))
- last_kf_resampled = TRUE;
+ last_kf_resampled = 1;
// Set back to unscaled by defaults
cpi->common.horiz_scale = NORMAL;
cpi->common.vert_scale = NORMAL;
// Calculate Average bits per frame.
- //av_bits_per_frame = cpi->twopass.bits_left/(double)(cpi->twopass.total_stats->count - cpi->common.current_video_frame);
+ //av_bits_per_frame = cpi->twopass.bits_left/(double)(cpi->twopass.total_stats.count - cpi->common.current_video_frame);
av_bits_per_frame = cpi->oxcf.target_bandwidth / DOUBLE_DIVIDE_CHECK((double)cpi->frame_rate);
//if ( av_bits_per_frame < 0.0 )
// av_bits_per_frame = 0.0
@@ -3117,21 +3117,21 @@
(last_kf_resampled && (projected_buffer_level < (cpi->oxcf.resample_up_water_mark * cpi->oxcf.optimal_buffer_level / 100))))
//( ((cpi->buffer_level < (cpi->oxcf.resample_down_water_mark * cpi->oxcf.optimal_buffer_level / 100))) &&
// ((projected_buffer_level < (cpi->oxcf.resample_up_water_mark * cpi->oxcf.optimal_buffer_level / 100))) ))
- resample_trigger = TRUE;
+ resample_trigger = 1;
else
- resample_trigger = FALSE;
+ resample_trigger = 0;
}
else
{
- int64_t clip_bits = (int64_t)(cpi->twopass.total_stats->count * cpi->oxcf.target_bandwidth / DOUBLE_DIVIDE_CHECK((double)cpi->frame_rate));
+ int64_t clip_bits = (int64_t)(cpi->twopass.total_stats.count * cpi->oxcf.target_bandwidth / DOUBLE_DIVIDE_CHECK((double)cpi->frame_rate));
int64_t over_spend = cpi->oxcf.starting_buffer_level - cpi->buffer_level;
if ((last_kf_resampled && (kf_q > cpi->worst_quality)) || // If triggered last time the threshold for triggering again is reduced
((kf_q > cpi->worst_quality) && // Projected Q higher than allowed and ...
(over_spend > clip_bits / 20))) // ... Overspend > 5% of total bits
- resample_trigger = TRUE;
+ resample_trigger = 1;
else
- resample_trigger = FALSE;
+ resample_trigger = 0;
}
diff --git a/vp8/encoder/lookahead.c b/vp8/encoder/lookahead.c
index b92e82b..3e582e3 100644
--- a/vp8/encoder/lookahead.c
+++ b/vp8/encoder/lookahead.c
@@ -48,7 +48,7 @@
{
if(ctx->buf)
{
- int i;
+ unsigned int i;
for(i = 0; i < ctx->max_sz; i++)
vp8_yv12_de_alloc_frame_buffer(&ctx->buf[i].img);
@@ -65,7 +65,7 @@
unsigned int depth)
{
struct lookahead_ctx *ctx = NULL;
- int i;
+ unsigned int i;
/* Clamp the lookahead queue depth */
if(depth < 1)
@@ -188,7 +188,7 @@
struct lookahead_entry*
vp8_lookahead_peek(struct lookahead_ctx *ctx,
- int index)
+ unsigned int index)
{
struct lookahead_entry* buf = NULL;
diff --git a/vp8/encoder/lookahead.h b/vp8/encoder/lookahead.h
index afb3fd4..32bafcd 100644
--- a/vp8/encoder/lookahead.h
+++ b/vp8/encoder/lookahead.h
@@ -92,7 +92,7 @@
*/
struct lookahead_entry*
vp8_lookahead_peek(struct lookahead_ctx *ctx,
- int index);
+ unsigned int index);
/**\brief Get the number of frames currently in the lookahead queue
diff --git a/vp8/encoder/mcomp.c b/vp8/encoder/mcomp.c
index 9d96383..735af95 100644
--- a/vp8/encoder/mcomp.c
+++ b/vp8/encoder/mcomp.c
@@ -1156,7 +1156,7 @@
int tot_steps;
int_mv this_mv;
- int bestsad = INT_MAX;
+ unsigned int bestsad = UINT_MAX;
int best_site = 0;
int last_site = 0;
@@ -1397,7 +1397,7 @@
unsigned char *bestaddress;
int_mv *best_mv = &d->bmi.mv;
int_mv this_mv;
- int bestsad = INT_MAX;
+ unsigned int bestsad = UINT_MAX;
int r, c;
unsigned char *check_here;
@@ -1527,7 +1527,7 @@
unsigned char *bestaddress;
int_mv *best_mv = &d->bmi.mv;
int_mv this_mv;
- int bestsad = INT_MAX;
+ unsigned int bestsad = UINT_MAX;
int r, c;
unsigned char *check_here;
diff --git a/vp8/encoder/onyx_if.c b/vp8/encoder/onyx_if.c
index c3e7a9f..e3f9519 100644
--- a/vp8/encoder/onyx_if.c
+++ b/vp8/encoder/onyx_if.c
@@ -23,7 +23,6 @@
#include "ratectrl.h"
#include "vp8/common/quant_common.h"
#include "segmentation.h"
-#include "vp8/common/g_common.h"
#include "vpx_scale/yv12extend.h"
#if CONFIG_POSTPROC
#include "vp8/common/postproc.h"
@@ -70,6 +69,7 @@
#endif
int vp8_estimate_entropy_savings(VP8_COMP *cpi);
+
int vp8_calc_ss_err(YV12_BUFFER_CONFIG *source, YV12_BUFFER_CONFIG *dest, const vp8_variance_rtcd_vtable_t *rtcd);
extern void vp8_temporal_filter_prepare_c(VP8_COMP *cpi, int distance);
@@ -380,42 +380,25 @@
vpx_free(cpi->mb.pip);
cpi->mb.pip = 0;
-
-#if !(CONFIG_REALTIME_ONLY)
- vpx_free(cpi->twopass.total_stats);
- cpi->twopass.total_stats = 0;
-
- vpx_free(cpi->twopass.total_left_stats);
- cpi->twopass.total_left_stats = 0;
-
- vpx_free(cpi->twopass.this_frame_stats);
- cpi->twopass.this_frame_stats = 0;
-#endif
}
-static void enable_segmentation(VP8_PTR ptr)
+static void enable_segmentation(VP8_COMP *cpi)
{
- VP8_COMP *cpi = (VP8_COMP *)(ptr);
-
// Set the appropriate feature bit
cpi->mb.e_mbd.segmentation_enabled = 1;
cpi->mb.e_mbd.update_mb_segmentation_map = 1;
cpi->mb.e_mbd.update_mb_segmentation_data = 1;
}
-static void disable_segmentation(VP8_PTR ptr)
+static void disable_segmentation(VP8_COMP *cpi)
{
- VP8_COMP *cpi = (VP8_COMP *)(ptr);
-
// Clear the appropriate feature bit
cpi->mb.e_mbd.segmentation_enabled = 0;
}
// Valid values for a segment are 0 to 3
// Segmentation map is arrange as [Rows][Columns]
-static void set_segmentation_map(VP8_PTR ptr, unsigned char *segmentation_map)
+static void set_segmentation_map(VP8_COMP *cpi, unsigned char *segmentation_map)
{
- VP8_COMP *cpi = (VP8_COMP *)(ptr);
-
// Copy in the new segmentation map
vpx_memcpy(cpi->segmentation_map, segmentation_map, (cpi->common.mb_rows * cpi->common.mb_cols));
@@ -432,19 +415,15 @@
// abs_delta = SEGMENT_DELTADATA (deltas) abs_delta = SEGMENT_ABSDATA (use the absolute values given).
//
//
-static void set_segment_data(VP8_PTR ptr, signed char *feature_data, unsigned char abs_delta)
+static void set_segment_data(VP8_COMP *cpi, signed char *feature_data, unsigned char abs_delta)
{
- VP8_COMP *cpi = (VP8_COMP *)(ptr);
-
cpi->mb.e_mbd.mb_segement_abs_delta = abs_delta;
vpx_memcpy(cpi->segment_feature_data, feature_data, sizeof(cpi->segment_feature_data));
}
-static void segmentation_test_function(VP8_PTR ptr)
+static void segmentation_test_function(VP8_COMP *cpi)
{
- VP8_COMP *cpi = (VP8_COMP *)(ptr);
-
unsigned char *seg_map;
signed char feature_data[MB_LVL_MAX][MAX_MB_SEGMENTS];
@@ -472,10 +451,10 @@
}*/
// Set the segmentation Map
- set_segmentation_map(ptr, seg_map);
+ set_segmentation_map(cpi, seg_map);
// Activate segmentation.
- enable_segmentation(ptr);
+ enable_segmentation(cpi);
// Set up the quant segment data
feature_data[MB_LVL_ALT_Q][0] = 0;
@@ -490,7 +469,7 @@
// Initialise the feature data structure
// SEGMENT_DELTADATA 0, SEGMENT_ABSDATA 1
- set_segment_data(ptr, &feature_data[0][0], SEGMENT_DELTADATA);
+ set_segment_data(cpi, &feature_data[0][0], SEGMENT_DELTADATA);
// Delete sementation map
vpx_free(seg_map);
@@ -564,10 +543,10 @@
}
// Set the segmentation Map
- set_segmentation_map((VP8_PTR)cpi, seg_map);
+ set_segmentation_map(cpi, seg_map);
// Activate segmentation.
- enable_segmentation((VP8_PTR)cpi);
+ enable_segmentation(cpi);
// Set up the quant segment data
feature_data[MB_LVL_ALT_Q][0] = 0;
@@ -583,7 +562,7 @@
// Initialise the feature data structure
// SEGMENT_DELTADATA 0, SEGMENT_ABSDATA 1
- set_segment_data((VP8_PTR)cpi, &feature_data[0][0], SEGMENT_DELTADATA);
+ set_segment_data(cpi, &feature_data[0][0], SEGMENT_DELTADATA);
// Delete sementation map
vpx_free(seg_map);
@@ -612,6 +591,93 @@
cpi->mb.e_mbd.mode_lf_deltas[3] = 4; // Split mv
}
+/* Convenience macros for mapping speed and mode into a continuous
+ * range
+ */
+#define GOOD(x) (x+1)
+#define RT(x) (x+7)
+
+static int speed_map(int speed, int *map)
+{
+ int res;
+
+ do
+ {
+ res = *map++;
+ } while(speed >= *map++);
+ return res;
+}
+
+static int thresh_mult_map_znn[] = {
+ /* map common to zero, nearest, and near */
+ 0, GOOD(2), 1500, GOOD(3), 2000, RT(0), 1000, RT(2), 2000, INT_MAX
+};
+
+static int thresh_mult_map_vhpred[] = {
+ 1000, GOOD(2), 1500, GOOD(3), 2000, RT(0), 1000, RT(1), 2000,
+ RT(7), INT_MAX, INT_MAX
+};
+
+static int thresh_mult_map_bpred[] = {
+ 2000, GOOD(0), 2500, GOOD(2), 5000, GOOD(3), 7500, RT(0), 2500, RT(1), 5000,
+ RT(6), INT_MAX, INT_MAX
+};
+
+static int thresh_mult_map_tm[] = {
+ 1000, GOOD(2), 1500, GOOD(3), 2000, RT(0), 0, RT(1), 1000, RT(2), 2000,
+ RT(7), INT_MAX, INT_MAX
+};
+
+static int thresh_mult_map_new1[] = {
+ 1000, GOOD(2), 2000, RT(0), 2000, INT_MAX
+};
+
+static int thresh_mult_map_new2[] = {
+ 1000, GOOD(2), 2000, GOOD(3), 2500, GOOD(5), 4000, RT(0), 2000, RT(2), 2500,
+ RT(5), 4000, INT_MAX
+};
+
+static int thresh_mult_map_split1[] = {
+ 2500, GOOD(0), 1700, GOOD(2), 10000, GOOD(3), 25000, GOOD(4), INT_MAX,
+ RT(0), 5000, RT(1), 10000, RT(2), 25000, RT(3), INT_MAX, INT_MAX
+};
+
+static int thresh_mult_map_split2[] = {
+ 5000, GOOD(0), 4500, GOOD(2), 20000, GOOD(3), 50000, GOOD(4), INT_MAX,
+ RT(0), 10000, RT(1), 20000, RT(2), 50000, RT(3), INT_MAX, INT_MAX
+};
+
+static int mode_check_freq_map_zn2[] = {
+ /* {zero,nearest}{2,3} */
+ 0, RT(10), 1<<1, RT(11), 1<<2, RT(12), 1<<3, INT_MAX
+};
+
+static int mode_check_freq_map_vhbpred[] = {
+ 0, GOOD(5), 2, RT(0), 0, RT(3), 2, RT(5), 4, INT_MAX
+};
+
+static int mode_check_freq_map_near2[] = {
+ 0, GOOD(5), 2, RT(0), 0, RT(3), 2, RT(10), 1<<2, RT(11), 1<<3, RT(12), 1<<4,
+ INT_MAX
+};
+
+static int mode_check_freq_map_new1[] = {
+ 0, RT(10), 1<<1, RT(11), 1<<2, RT(12), 1<<3, INT_MAX
+};
+
+static int mode_check_freq_map_new2[] = {
+ 0, GOOD(5), 4, RT(0), 0, RT(3), 4, RT(10), 1<<3, RT(11), 1<<4, RT(12), 1<<5,
+ INT_MAX
+};
+
+static int mode_check_freq_map_split1[] = {
+ 0, GOOD(2), 2, GOOD(3), 7, RT(1), 2, RT(2), 7, INT_MAX
+};
+
+static int mode_check_freq_map_split2[] = {
+ 0, GOOD(1), 2, GOOD(2), 4, GOOD(3), 15, RT(1), 4, RT(2), 15, INT_MAX
+};
+
void vp8_set_speed_features(VP8_COMP *cpi)
{
SPEED_FEATURES *sf = &cpi->sf;
@@ -663,93 +729,81 @@
if (cpi->ref_frame_flags & VP8_ALT_FLAG)
ref_frames++;
+ /* Convert speed to continuous range, with clamping */
+ if (Mode == 0)
+ Speed = 0;
+ else if (Mode == 2)
+ Speed = RT(Speed);
+ else
+ {
+ if (Speed > 5)
+ Speed = 5;
+ Speed = GOOD(Speed);
+ }
+
+ sf->thresh_mult[THR_ZERO1] =
+ sf->thresh_mult[THR_NEAREST1] =
+ sf->thresh_mult[THR_NEAR1] =
+ sf->thresh_mult[THR_DC] = 0; /* always */
+
+ sf->thresh_mult[THR_ZERO2] =
+ sf->thresh_mult[THR_ZERO3] =
+ sf->thresh_mult[THR_NEAREST2] =
+ sf->thresh_mult[THR_NEAREST3] =
+ sf->thresh_mult[THR_NEAR2] =
+ sf->thresh_mult[THR_NEAR3] = speed_map(Speed, thresh_mult_map_znn);
+
+ sf->thresh_mult[THR_V_PRED] =
+ sf->thresh_mult[THR_H_PRED] = speed_map(Speed, thresh_mult_map_vhpred);
+ sf->thresh_mult[THR_B_PRED] = speed_map(Speed, thresh_mult_map_bpred);
+ sf->thresh_mult[THR_TM] = speed_map(Speed, thresh_mult_map_tm);
+ sf->thresh_mult[THR_NEW1] = speed_map(Speed, thresh_mult_map_new1);
+ sf->thresh_mult[THR_NEW2] =
+ sf->thresh_mult[THR_NEW3] = speed_map(Speed, thresh_mult_map_new2);
+ sf->thresh_mult[THR_SPLIT1] = speed_map(Speed, thresh_mult_map_split1);
+ sf->thresh_mult[THR_SPLIT2] =
+ sf->thresh_mult[THR_SPLIT3] = speed_map(Speed, thresh_mult_map_split2);
+
+ cpi->mode_check_freq[THR_ZERO1] =
+ cpi->mode_check_freq[THR_NEAREST1] =
+ cpi->mode_check_freq[THR_NEAR1] =
+ cpi->mode_check_freq[THR_TM] =
+ cpi->mode_check_freq[THR_DC] = 0; /* always */
+
+ cpi->mode_check_freq[THR_ZERO2] =
+ cpi->mode_check_freq[THR_ZERO3] =
+ cpi->mode_check_freq[THR_NEAREST2] =
+ cpi->mode_check_freq[THR_NEAREST3] = speed_map(Speed,
+ mode_check_freq_map_zn2);
+
+ cpi->mode_check_freq[THR_NEAR2] =
+ cpi->mode_check_freq[THR_NEAR3] = speed_map(Speed,
+ mode_check_freq_map_near2);
+
+ cpi->mode_check_freq[THR_V_PRED] =
+ cpi->mode_check_freq[THR_H_PRED] =
+ cpi->mode_check_freq[THR_B_PRED] = speed_map(Speed,
+ mode_check_freq_map_vhbpred);
+ cpi->mode_check_freq[THR_NEW1] = speed_map(Speed,
+ mode_check_freq_map_new1);
+ cpi->mode_check_freq[THR_NEW2] =
+ cpi->mode_check_freq[THR_NEW3] = speed_map(Speed,
+ mode_check_freq_map_new2);
+ cpi->mode_check_freq[THR_SPLIT1] = speed_map(Speed,
+ mode_check_freq_map_split1);
+ cpi->mode_check_freq[THR_SPLIT2] =
+ cpi->mode_check_freq[THR_SPLIT3] = speed_map(Speed,
+ mode_check_freq_map_split2);
+ Speed = cpi->Speed;
switch (Mode)
{
#if !(CONFIG_REALTIME_ONLY)
case 0: // best quality mode
- sf->thresh_mult[THR_ZERO1 ] = 0;
- sf->thresh_mult[THR_ZERO2 ] = 0;
- sf->thresh_mult[THR_ZERO3 ] = 0;
- sf->thresh_mult[THR_NEAREST1 ] = 0;
- sf->thresh_mult[THR_NEAREST2 ] = 0;
- sf->thresh_mult[THR_NEAREST3 ] = 0;
- sf->thresh_mult[THR_NEAR1 ] = 0;
- sf->thresh_mult[THR_NEAR2 ] = 0;
- sf->thresh_mult[THR_NEAR3 ] = 0;
-
- sf->thresh_mult[THR_DC ] = 0;
-
- sf->thresh_mult[THR_V_PRED ] = 1000;
- sf->thresh_mult[THR_H_PRED ] = 1000;
- sf->thresh_mult[THR_B_PRED ] = 2000;
- sf->thresh_mult[THR_TM ] = 1000;
-
- sf->thresh_mult[THR_NEW1 ] = 1000;
- sf->thresh_mult[THR_NEW2 ] = 1000;
- sf->thresh_mult[THR_NEW3 ] = 1000;
-
- sf->thresh_mult[THR_SPLIT1 ] = 2500;
- sf->thresh_mult[THR_SPLIT2 ] = 5000;
- sf->thresh_mult[THR_SPLIT3 ] = 5000;
-
-
sf->first_step = 0;
sf->max_step_search_steps = MAX_MVSEARCH_STEPS;
break;
case 1:
case 3:
- sf->thresh_mult[THR_NEAREST1 ] = 0;
- sf->thresh_mult[THR_ZERO1 ] = 0;
- sf->thresh_mult[THR_DC ] = 0;
- sf->thresh_mult[THR_NEAR1 ] = 0;
- sf->thresh_mult[THR_V_PRED ] = 1000;
- sf->thresh_mult[THR_H_PRED ] = 1000;
- sf->thresh_mult[THR_B_PRED ] = 2500;
- sf->thresh_mult[THR_TM ] = 1000;
-
- sf->thresh_mult[THR_NEAREST2 ] = 1000;
- sf->thresh_mult[THR_NEAREST3 ] = 1000;
-
- sf->thresh_mult[THR_ZERO2 ] = 1000;
- sf->thresh_mult[THR_ZERO3 ] = 1000;
- sf->thresh_mult[THR_NEAR2 ] = 1000;
- sf->thresh_mult[THR_NEAR3 ] = 1000;
-
-#if 1
- sf->thresh_mult[THR_ZERO1 ] = 0;
- sf->thresh_mult[THR_ZERO2 ] = 0;
- sf->thresh_mult[THR_ZERO3 ] = 0;
- sf->thresh_mult[THR_NEAREST1 ] = 0;
- sf->thresh_mult[THR_NEAREST2 ] = 0;
- sf->thresh_mult[THR_NEAREST3 ] = 0;
- sf->thresh_mult[THR_NEAR1 ] = 0;
- sf->thresh_mult[THR_NEAR2 ] = 0;
- sf->thresh_mult[THR_NEAR3 ] = 0;
-
-// sf->thresh_mult[THR_DC ] = 0;
-
-// sf->thresh_mult[THR_V_PRED ] = 1000;
-// sf->thresh_mult[THR_H_PRED ] = 1000;
-// sf->thresh_mult[THR_B_PRED ] = 2000;
-// sf->thresh_mult[THR_TM ] = 1000;
-
- sf->thresh_mult[THR_NEW1 ] = 1000;
- sf->thresh_mult[THR_NEW2 ] = 1000;
- sf->thresh_mult[THR_NEW3 ] = 1000;
-
- sf->thresh_mult[THR_SPLIT1 ] = 1700;
- sf->thresh_mult[THR_SPLIT2 ] = 4500;
- sf->thresh_mult[THR_SPLIT3 ] = 4500;
-#else
- sf->thresh_mult[THR_NEW1 ] = 1500;
- sf->thresh_mult[THR_NEW2 ] = 1500;
- sf->thresh_mult[THR_NEW3 ] = 1500;
-
- sf->thresh_mult[THR_SPLIT1 ] = 5000;
- sf->thresh_mult[THR_SPLIT2 ] = 10000;
- sf->thresh_mult[THR_SPLIT3 ] = 10000;
-#endif
-
if (Speed > 0)
{
/* Disable coefficient optimization above speed 0 */
@@ -758,83 +812,10 @@
sf->no_skip_block4x4_search = 0;
sf->first_step = 1;
-
- cpi->mode_check_freq[THR_SPLIT2] = 2;
- cpi->mode_check_freq[THR_SPLIT3] = 2;
- cpi->mode_check_freq[THR_SPLIT1 ] = 0;
- }
-
- if (Speed > 1)
- {
- cpi->mode_check_freq[THR_SPLIT2] = 4;
- cpi->mode_check_freq[THR_SPLIT3] = 4;
- cpi->mode_check_freq[THR_SPLIT1 ] = 2;
-
- sf->thresh_mult[THR_TM ] = 1500;
- sf->thresh_mult[THR_V_PRED ] = 1500;
- sf->thresh_mult[THR_H_PRED ] = 1500;
- sf->thresh_mult[THR_B_PRED ] = 5000;
-
- if (ref_frames > 1)
- {
- sf->thresh_mult[THR_NEW1 ] = 2000;
- sf->thresh_mult[THR_SPLIT1 ] = 10000;
- }
-
- if (ref_frames > 2)
- {
- sf->thresh_mult[THR_NEAREST2 ] = 1500;
- sf->thresh_mult[THR_ZERO2 ] = 1500;
- sf->thresh_mult[THR_NEAR2 ] = 1500;
- sf->thresh_mult[THR_NEW2 ] = 2000;
- sf->thresh_mult[THR_SPLIT2 ] = 20000;
- }
-
- if (ref_frames > 3)
- {
- sf->thresh_mult[THR_NEAREST3 ] = 1500;
- sf->thresh_mult[THR_ZERO3 ] = 1500;
- sf->thresh_mult[THR_NEAR3 ] = 1500;
- sf->thresh_mult[THR_NEW3 ] = 2000;
- sf->thresh_mult[THR_SPLIT3 ] = 20000;
- }
}
if (Speed > 2)
{
- cpi->mode_check_freq[THR_SPLIT2] = 15;
- cpi->mode_check_freq[THR_SPLIT3] = 15;
- cpi->mode_check_freq[THR_SPLIT1 ] = 7;
-
- sf->thresh_mult[THR_TM ] = 2000;
- sf->thresh_mult[THR_V_PRED ] = 2000;
- sf->thresh_mult[THR_H_PRED ] = 2000;
- sf->thresh_mult[THR_B_PRED ] = 7500;
-
- if (ref_frames > 1)
- {
- sf->thresh_mult[THR_NEW1 ] = 2000;
- sf->thresh_mult[THR_SPLIT1 ] = 25000;
- }
-
- if (ref_frames > 2)
- {
- sf->thresh_mult[THR_NEAREST2 ] = 2000;
- sf->thresh_mult[THR_ZERO2 ] = 2000;
- sf->thresh_mult[THR_NEAR2 ] = 2000;
- sf->thresh_mult[THR_NEW2 ] = 2500;
- sf->thresh_mult[THR_SPLIT2 ] = 50000;
- }
-
- if (ref_frames > 3)
- {
- sf->thresh_mult[THR_NEAREST3 ] = 2000;
- sf->thresh_mult[THR_ZERO3 ] = 2000;
- sf->thresh_mult[THR_NEAR3 ] = 2000;
- sf->thresh_mult[THR_NEW3 ] = 2500;
- sf->thresh_mult[THR_SPLIT3 ] = 50000;
- }
-
sf->improved_quant = 0;
sf->improved_dct = 0;
@@ -846,18 +827,6 @@
if (Speed > 3)
{
- sf->thresh_mult[THR_SPLIT3 ] = INT_MAX;
- sf->thresh_mult[THR_SPLIT2 ] = INT_MAX;
- sf->thresh_mult[THR_SPLIT1 ] = INT_MAX;
-
- cpi->mode_check_freq[THR_V_PRED] = 0;
- cpi->mode_check_freq[THR_H_PRED] = 0;
- cpi->mode_check_freq[THR_B_PRED] = 0;
- cpi->mode_check_freq[THR_NEAR2] = 0;
- cpi->mode_check_freq[THR_NEW2] = 0;
- cpi->mode_check_freq[THR_NEAR3] = 0;
- cpi->mode_check_freq[THR_NEW3] = 0;
-
sf->auto_filter = 1;
sf->recode_loop = 0; // recode loop off
sf->RD = 0; // Turn rd off
@@ -867,38 +836,6 @@
if (Speed > 4)
{
sf->auto_filter = 0; // Faster selection of loop filter
-
- cpi->mode_check_freq[THR_V_PRED] = 2;
- cpi->mode_check_freq[THR_H_PRED] = 2;
- cpi->mode_check_freq[THR_B_PRED] = 2;
-
- if (ref_frames > 2)
- {
- cpi->mode_check_freq[THR_NEAR2] = 2;
- cpi->mode_check_freq[THR_NEW2] = 4;
- }
-
- if (ref_frames > 3)
- {
- cpi->mode_check_freq[THR_NEAR3] = 2;
- cpi->mode_check_freq[THR_NEW3] = 4;
- }
-
- if (ref_frames > 2)
- {
- sf->thresh_mult[THR_NEAREST2 ] = 2000;
- sf->thresh_mult[THR_ZERO2 ] = 2000;
- sf->thresh_mult[THR_NEAR2 ] = 2000;
- sf->thresh_mult[THR_NEW2 ] = 4000;
- }
-
- if (ref_frames > 3)
- {
- sf->thresh_mult[THR_NEAREST3 ] = 2000;
- sf->thresh_mult[THR_ZERO3 ] = 2000;
- sf->thresh_mult[THR_NEAR3 ] = 2000;
- sf->thresh_mult[THR_NEW3 ] = 4000;
- }
}
break;
@@ -908,67 +845,10 @@
sf->recode_loop = 0;
sf->auto_filter = 1;
sf->iterative_sub_pixel = 1;
- sf->thresh_mult[THR_NEAREST1 ] = 0;
- sf->thresh_mult[THR_ZERO1 ] = 0;
- sf->thresh_mult[THR_DC ] = 0;
- sf->thresh_mult[THR_TM ] = 0;
- sf->thresh_mult[THR_NEAR1 ] = 0;
- sf->thresh_mult[THR_V_PRED ] = 1000;
- sf->thresh_mult[THR_H_PRED ] = 1000;
- sf->thresh_mult[THR_B_PRED ] = 2500;
- sf->thresh_mult[THR_NEAREST2 ] = 1000;
- sf->thresh_mult[THR_ZERO2 ] = 1000;
- sf->thresh_mult[THR_NEAR2 ] = 1000;
- sf->thresh_mult[THR_NEAREST3 ] = 1000;
- sf->thresh_mult[THR_ZERO3 ] = 1000;
- sf->thresh_mult[THR_NEAR3 ] = 1000;
- sf->thresh_mult[THR_NEW1 ] = 2000;
- sf->thresh_mult[THR_NEW2 ] = 2000;
- sf->thresh_mult[THR_NEW3 ] = 2000;
- sf->thresh_mult[THR_SPLIT1 ] = 5000;
- sf->thresh_mult[THR_SPLIT2 ] = 10000;
- sf->thresh_mult[THR_SPLIT3 ] = 10000;
sf->search_method = NSTEP;
if (Speed > 0)
{
- cpi->mode_check_freq[THR_SPLIT2] = 4;
- cpi->mode_check_freq[THR_SPLIT3] = 4;
- cpi->mode_check_freq[THR_SPLIT1 ] = 2;
-
- sf->thresh_mult[THR_DC ] = 0;
- sf->thresh_mult[THR_TM ] = 1000;
- sf->thresh_mult[THR_V_PRED ] = 2000;
- sf->thresh_mult[THR_H_PRED ] = 2000;
- sf->thresh_mult[THR_B_PRED ] = 5000;
-
- if (ref_frames > 1)
- {
- sf->thresh_mult[THR_NEAREST1 ] = 0;
- sf->thresh_mult[THR_ZERO1 ] = 0;
- sf->thresh_mult[THR_NEAR1 ] = 0;
- sf->thresh_mult[THR_NEW1 ] = 2000;
- sf->thresh_mult[THR_SPLIT1 ] = 10000;
- }
-
- if (ref_frames > 2)
- {
- sf->thresh_mult[THR_NEAREST2 ] = 1000;
- sf->thresh_mult[THR_ZERO2 ] = 1000;
- sf->thresh_mult[THR_NEAR2 ] = 1000;
- sf->thresh_mult[THR_NEW2 ] = 2000;
- sf->thresh_mult[THR_SPLIT2 ] = 20000;
- }
-
- if (ref_frames > 3)
- {
- sf->thresh_mult[THR_NEAREST3 ] = 1000;
- sf->thresh_mult[THR_ZERO3 ] = 1000;
- sf->thresh_mult[THR_NEAR3 ] = 1000;
- sf->thresh_mult[THR_NEW3 ] = 2000;
- sf->thresh_mult[THR_SPLIT3 ] = 20000;
- }
-
sf->improved_quant = 0;
sf->improved_dct = 0;
@@ -977,133 +857,28 @@
sf->first_step = 1;
}
- if (Speed > 1)
- {
- cpi->mode_check_freq[THR_SPLIT1 ] = 7;
- cpi->mode_check_freq[THR_SPLIT2] = 15;
- cpi->mode_check_freq[THR_SPLIT3] = 15;
-
- sf->thresh_mult[THR_TM ] = 2000;
- sf->thresh_mult[THR_V_PRED ] = 2000;
- sf->thresh_mult[THR_H_PRED ] = 2000;
- sf->thresh_mult[THR_B_PRED ] = 5000;
-
- if (ref_frames > 1)
- {
- sf->thresh_mult[THR_NEW1 ] = 2000;
- sf->thresh_mult[THR_SPLIT1 ] = 25000;
- }
-
- if (ref_frames > 2)
- {
- sf->thresh_mult[THR_NEAREST2 ] = 2000;
- sf->thresh_mult[THR_ZERO2 ] = 2000;
- sf->thresh_mult[THR_NEAR2 ] = 2000;
- sf->thresh_mult[THR_NEW2 ] = 2500;
- sf->thresh_mult[THR_SPLIT2 ] = 50000;
- }
-
- if (ref_frames > 3)
- {
- sf->thresh_mult[THR_NEAREST3 ] = 2000;
- sf->thresh_mult[THR_ZERO3 ] = 2000;
- sf->thresh_mult[THR_NEAR3 ] = 2000;
- sf->thresh_mult[THR_NEW3 ] = 2500;
- sf->thresh_mult[THR_SPLIT3 ] = 50000;
- }
-
- }
-
if (Speed > 2)
- {
sf->auto_filter = 0; // Faster selection of loop filter
- cpi->mode_check_freq[THR_V_PRED] = 2;
- cpi->mode_check_freq[THR_H_PRED] = 2;
- cpi->mode_check_freq[THR_B_PRED] = 2;
-
- if (ref_frames > 2)
- {
- cpi->mode_check_freq[THR_NEAR2] = 2;
- cpi->mode_check_freq[THR_NEW2] = 4;
- }
-
- if (ref_frames > 3)
- {
- cpi->mode_check_freq[THR_NEAR3] = 2;
- cpi->mode_check_freq[THR_NEW3] = 4;
- }
-
- sf->thresh_mult[THR_SPLIT1 ] = INT_MAX;
- sf->thresh_mult[THR_SPLIT2 ] = INT_MAX;
- sf->thresh_mult[THR_SPLIT3 ] = INT_MAX;
-
- }
-
if (Speed > 3)
{
sf->RD = 0;
-
sf->auto_filter = 1;
}
if (Speed > 4)
{
sf->auto_filter = 0; // Faster selection of loop filter
-
sf->search_method = HEX;
- //sf->search_method = DIAMOND;
-
sf->iterative_sub_pixel = 0;
-
- cpi->mode_check_freq[THR_V_PRED] = 4;
- cpi->mode_check_freq[THR_H_PRED] = 4;
- cpi->mode_check_freq[THR_B_PRED] = 4;
-
- if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
- {
- cpi->mode_check_freq[THR_NEAR2] = 2;
- cpi->mode_check_freq[THR_NEW2] = 4;
- }
-
- if (cpi->ref_frame_flags & VP8_ALT_FLAG)
- {
- cpi->mode_check_freq[THR_NEAR3] = 2;
- cpi->mode_check_freq[THR_NEW3] = 4;
- }
-
- sf->thresh_mult[THR_TM ] = 2000;
- sf->thresh_mult[THR_B_PRED ] = 5000;
-
- if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
- {
- sf->thresh_mult[THR_NEAREST2 ] = 2000;
- sf->thresh_mult[THR_ZERO2 ] = 2000;
- sf->thresh_mult[THR_NEAR2 ] = 2000;
- sf->thresh_mult[THR_NEW2 ] = 4000;
- }
-
- if (cpi->ref_frame_flags & VP8_ALT_FLAG)
- {
- sf->thresh_mult[THR_NEAREST3 ] = 2000;
- sf->thresh_mult[THR_ZERO3 ] = 2000;
- sf->thresh_mult[THR_NEAR3 ] = 2000;
- sf->thresh_mult[THR_NEW3 ] = 4000;
- }
- }
-
- if (Speed > 5)
- {
- // Disable split MB intra prediction mode
- sf->thresh_mult[THR_B_PRED] = INT_MAX;
}
if (Speed > 6)
{
- unsigned int i, sum = 0;
+ unsigned int sum = 0;
unsigned int total_mbs = cm->MBs;
- int thresh;
- int total_skip;
+ int i, thresh;
+ unsigned int total_skip;
int min = 2000;
@@ -1156,55 +931,28 @@
sf->thresh_mult[THR_NEAR3 ] = thresh;
}
- // Disable other intra prediction modes
- sf->thresh_mult[THR_TM] = INT_MAX;
- sf->thresh_mult[THR_V_PRED] = INT_MAX;
- sf->thresh_mult[THR_H_PRED] = INT_MAX;
-
sf->improved_mv_pred = 0;
}
if (Speed > 8)
- {
sf->quarter_pixel_search = 0;
- }
- if (Speed > 9)
+ if(cm->version == 0)
{
- int Tmp = cpi->Speed - 8;
+ cm->filter_type = NORMAL_LOOPFILTER;
- if (Tmp > 4)
- Tmp = 4;
-
- if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
- {
- cpi->mode_check_freq[THR_ZERO2] = 1 << (Tmp - 1);
- cpi->mode_check_freq[THR_NEAREST2] = 1 << (Tmp - 1);
- cpi->mode_check_freq[THR_NEAR2] = 1 << Tmp;
- cpi->mode_check_freq[THR_NEW2] = 1 << (Tmp + 1);
- }
-
- if (cpi->ref_frame_flags & VP8_ALT_FLAG)
- {
- cpi->mode_check_freq[THR_ZERO3] = 1 << (Tmp - 1);
- cpi->mode_check_freq[THR_NEAREST3] = 1 << (Tmp - 1);
- cpi->mode_check_freq[THR_NEAR3] = 1 << Tmp;
- cpi->mode_check_freq[THR_NEW3] = 1 << (Tmp + 1);
- }
-
- cpi->mode_check_freq[THR_NEW1 ] = 1 << (Tmp - 1);
+ if (Speed >= 14)
+ cm->filter_type = SIMPLE_LOOPFILTER;
}
-
- cm->filter_type = NORMAL_LOOPFILTER;
-
- if (Speed >= 14)
+ else
+ {
cm->filter_type = SIMPLE_LOOPFILTER;
-
- if (Speed >= 15)
- {
- sf->half_pixel_search = 0; // This has a big hit on quality. Last resort
}
+ // This has a big hit on quality. Last resort
+ if (Speed >= 15)
+ sf->half_pixel_search = 0;
+
vpx_memset(cpi->error_bins, 0, sizeof(cpi->error_bins));
}; /* switch */
@@ -1290,6 +1038,9 @@
frames_at_speed[cpi->Speed]++;
#endif
}
+#undef GOOD
+#undef RT
+
static void alloc_raw_frame_buffers(VP8_COMP *cpi)
{
int width = (cpi->oxcf.Width + 15) & ~15;
@@ -1390,25 +1141,6 @@
vpx_calloc(sizeof(unsigned int),
cm->mb_rows * cm->mb_cols));
-#if !(CONFIG_REALTIME_ONLY)
- vpx_free(cpi->twopass.total_stats);
-
- cpi->twopass.total_stats = vpx_calloc(1, sizeof(FIRSTPASS_STATS));
-
- vpx_free(cpi->twopass.total_left_stats);
- cpi->twopass.total_left_stats = vpx_calloc(1, sizeof(FIRSTPASS_STATS));
-
- vpx_free(cpi->twopass.this_frame_stats);
-
- cpi->twopass.this_frame_stats = vpx_calloc(1, sizeof(FIRSTPASS_STATS));
-
- if( !cpi->twopass.total_stats ||
- !cpi->twopass.total_left_stats ||
- !cpi->twopass.this_frame_stats)
- vpx_internal_error(&cpi->common.error, VPX_CODEC_MEM_ERROR,
- "Failed to allocate firstpass stats");
-#endif
-
#if CONFIG_MULTITHREAD
if (width < 640)
cpi->mt_sync_range = 1;
@@ -1497,9 +1229,8 @@
}
-static void init_config(VP8_PTR ptr, VP8_CONFIG *oxcf)
+static void init_config(VP8_COMP *cpi, VP8_CONFIG *oxcf)
{
- VP8_COMP *cpi = (VP8_COMP *)(ptr);
VP8_COMMON *cm = &cpi->common;
cpi->oxcf = *oxcf;
@@ -1524,7 +1255,7 @@
cpi->frame_rate = 30;
// change includes all joint functionality
- vp8_change_config(ptr, oxcf);
+ vp8_change_config(cpi, oxcf);
// Initialize active best and worst q and average q values.
cpi->active_worst_quality = cpi->oxcf.worst_allowed_q;
@@ -1546,8 +1277,8 @@
// Temporal scalabilty
if (cpi->oxcf.number_of_layers > 1)
{
- int i;
- int prev_layer_frame_rate=0;
+ unsigned int i;
+ double prev_layer_frame_rate=0;
for (i=0; i<cpi->oxcf.number_of_layers; i++)
{
@@ -1615,9 +1346,8 @@
}
-void vp8_change_config(VP8_PTR ptr, VP8_CONFIG *oxcf)
+void vp8_change_config(VP8_COMP *cpi, VP8_CONFIG *oxcf)
{
- VP8_COMP *cpi = (VP8_COMP *)(ptr);
VP8_COMMON *cm = &cpi->common;
if (!cpi)
@@ -1809,7 +1539,7 @@
cpi->active_best_quality = cpi->oxcf.worst_allowed_q;
}
- cpi->buffered_mode = (cpi->oxcf.optimal_buffer_level > 0) ? TRUE : FALSE;
+ cpi->buffered_mode = cpi->oxcf.optimal_buffer_level > 0;
cpi->cq_target_quality = cpi->oxcf.cq_level;
@@ -1908,19 +1638,14 @@
while (++i <= mvfp_max);
}
-VP8_PTR vp8_create_compressor(VP8_CONFIG *oxcf)
+struct VP8_COMP* vp8_create_compressor(VP8_CONFIG *oxcf)
{
int i;
- volatile union
- {
- VP8_COMP *cpi;
- VP8_PTR ptr;
- } ctx;
VP8_COMP *cpi;
VP8_COMMON *cm;
- cpi = ctx.cpi = vpx_memalign(32, sizeof(VP8_COMP));
+ cpi = vpx_memalign(32, sizeof(VP8_COMP));
// Check that the CPI instance is valid
if (!cpi)
return 0;
@@ -1931,10 +1656,8 @@
if (setjmp(cm->error.jmp))
{
- VP8_PTR ptr = ctx.ptr;
-
- ctx.cpi->common.error.setjmp = 0;
- vp8_remove_compressor(&ptr);
+ cpi->common.error.setjmp = 0;
+ vp8_remove_compressor(&cpi);
return 0;
}
@@ -1945,7 +1668,7 @@
vp8_create_common(&cpi->common);
vp8_cmachine_specific_config(cpi);
- init_config((VP8_PTR)cpi, oxcf);
+ init_config(cpi, oxcf);
memcpy(cpi->base_skip_false_prob, vp8cx_base_skip_false_prob, sizeof(vp8cx_base_skip_false_prob));
cpi->common.current_video_frame = 0;
@@ -2024,7 +1747,7 @@
cpi->cyclic_refresh_map = (signed char *) NULL;
// Test function for segmentation
- //segmentation_test_function((VP8_PTR) cpi);
+ //segmentation_test_function( cpi);
#ifdef ENTROPY_STATS
init_context_counters();
@@ -2035,11 +1758,11 @@
cpi->frames_since_key = 8; // Give a sensible default for the first frame.
cpi->key_frame_frequency = cpi->oxcf.key_freq;
- cpi->this_key_frame_forced = FALSE;
- cpi->next_key_frame_forced = FALSE;
+ cpi->this_key_frame_forced = 0;
+ cpi->next_key_frame_forced = 0;
- cpi->source_alt_ref_pending = FALSE;
- cpi->source_alt_ref_active = FALSE;
+ cpi->source_alt_ref_pending = 0;
+ cpi->source_alt_ref_active = 0;
cpi->common.refresh_alt_ref_frame = 0;
cpi->b_calculate_psnr = CONFIG_INTERNAL_STATS;
@@ -2244,14 +1967,14 @@
vp8_cal_low_res_mb_cols(cpi);
#endif
- return (VP8_PTR) cpi;
+ return cpi;
}
-void vp8_remove_compressor(VP8_PTR *ptr)
+void vp8_remove_compressor(VP8_COMP **ptr)
{
- VP8_COMP *cpi = (VP8_COMP *)(*ptr);
+ VP8_COMP *cpi = *ptr;
if (!cpi)
return;
@@ -2662,20 +2385,16 @@
}
-int vp8_use_as_reference(VP8_PTR ptr, int ref_frame_flags)
+int vp8_use_as_reference(VP8_COMP *cpi, int ref_frame_flags)
{
- VP8_COMP *cpi = (VP8_COMP *)(ptr);
-
if (ref_frame_flags > 7)
return -1 ;
cpi->ref_frame_flags = ref_frame_flags;
return 0;
}
-int vp8_update_reference(VP8_PTR ptr, int ref_frame_flags)
+int vp8_update_reference(VP8_COMP *cpi, int ref_frame_flags)
{
- VP8_COMP *cpi = (VP8_COMP *)(ptr);
-
if (ref_frame_flags > 7)
return -1 ;
@@ -2695,9 +2414,8 @@
return 0;
}
-int vp8_get_reference(VP8_PTR ptr, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd)
+int vp8_get_reference(VP8_COMP *cpi, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd)
{
- VP8_COMP *cpi = (VP8_COMP *)(ptr);
VP8_COMMON *cm = &cpi->common;
int ref_fb_idx;
@@ -2714,9 +2432,8 @@
return 0;
}
-int vp8_set_reference(VP8_PTR ptr, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd)
+int vp8_set_reference(VP8_COMP *cpi, VP8_REFFRAME ref_frame_flag, YV12_BUFFER_CONFIG *sd)
{
- VP8_COMP *cpi = (VP8_COMP *)(ptr);
VP8_COMMON *cm = &cpi->common;
int ref_fb_idx;
@@ -2734,9 +2451,8 @@
return 0;
}
-int vp8_update_entropy(VP8_PTR comp, int update)
+int vp8_update_entropy(VP8_COMP *cpi, int update)
{
- VP8_COMP *cpi = (VP8_COMP *) comp;
VP8_COMMON *cm = &cpi->common;
cm->refresh_entropy_probs = update;
@@ -2892,10 +2608,10 @@
cpi->common.frames_since_golden = 0;
// Clear the alternate reference update pending flag.
- cpi->source_alt_ref_pending = FALSE;
+ cpi->source_alt_ref_pending = 0;
// Set the alternate refernce frame active flag
- cpi->source_alt_ref_active = TRUE;
+ cpi->source_alt_ref_active = 1;
}
@@ -2958,12 +2674,12 @@
if (cpi->oxcf.fixed_q >= 0 &&
cpi->oxcf.play_alternate && !cpi->common.refresh_alt_ref_frame)
{
- cpi->source_alt_ref_pending = TRUE;
+ cpi->source_alt_ref_pending = 1;
cpi->frames_till_gf_update_due = cpi->baseline_gf_interval;
}
if (!cpi->source_alt_ref_pending)
- cpi->source_alt_ref_active = FALSE;
+ cpi->source_alt_ref_active = 0;
// Decrement count down till next gf
if (cpi->frames_till_gf_update_due > 0)
@@ -2997,47 +2713,6 @@
{
VP8_COMMON *cm = &cpi->common;
-#if 0
- const int *const rfct = cpi->recent_ref_frame_usage;
- const int rf_intra = rfct[INTRA_FRAME];
- const int rf_inter = rfct[LAST_FRAME] + rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME];
-
- if (cm->frame_type == KEY_FRAME)
- {
- cpi->prob_intra_coded = 255;
- cpi->prob_last_coded = 128;
- cpi->prob_gf_coded = 128;
- }
- else if (!(rf_intra + rf_inter))
- {
- // This is a trap in case this function is called with cpi->recent_ref_frame_usage[] blank.
- cpi->prob_intra_coded = 63;
- cpi->prob_last_coded = 128;
- cpi->prob_gf_coded = 128;
- }
- else
- {
- cpi->prob_intra_coded = (rf_intra * 255) / (rf_intra + rf_inter);
-
- if (cpi->prob_intra_coded < 1)
- cpi->prob_intra_coded = 1;
-
- if ((cm->frames_since_golden > 0) || cpi->source_alt_ref_active)
- {
- cpi->prob_last_coded = rf_inter ? (rfct[LAST_FRAME] * 255) / rf_inter : 128;
-
- if (cpi->prob_last_coded < 1)
- cpi->prob_last_coded = 1;
-
- cpi->prob_gf_coded = (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME])
- ? (rfct[GOLDEN_FRAME] * 255) / (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME]) : 128;
-
- if (cpi->prob_gf_coded < 1)
- cpi->prob_gf_coded = 1;
- }
- }
-
-#else
const int *const rfct = cpi->count_mb_ref_frame_usage;
const int rf_intra = rfct[INTRA_FRAME];
const int rf_inter = rfct[LAST_FRAME] + rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME];
@@ -3050,59 +2725,9 @@
}
else if (!(rf_intra + rf_inter))
{
- if (cpi->oxcf.number_of_layers > 1)
- {
- if (cpi->ref_frame_flags == VP8_LAST_FLAG)
- {
- cpi->prob_intra_coded = 63;
- cpi->prob_last_coded = 255;
- cpi->prob_gf_coded = 128;
- }
- else if (cpi->ref_frame_flags == VP8_GOLD_FLAG)
- {
- cpi->prob_intra_coded = 63;
- cpi->prob_last_coded = 1;
- cpi->prob_gf_coded = 255;
- }
- else if (cpi->ref_frame_flags == VP8_ALT_FLAG)
- {
- cpi->prob_intra_coded = 63;
- cpi->prob_last_coded = 1;
- cpi->prob_gf_coded = 1;
- }
- else
- {
- cpi->prob_intra_coded = 63;
- cpi->prob_last_coded = 128;
- cpi->prob_gf_coded = 128;
- }
- }
- else
- {
- // This is a trap in case this function is called with
- // cpi->recent_ref_frame_usage[] blank.
- cpi->prob_intra_coded = 63;
- cpi->prob_last_coded = 128;
- cpi->prob_gf_coded = 128;
- }
- }
- else
- {
- cpi->prob_intra_coded = (rf_intra * 255) / (rf_intra + rf_inter);
-
- if (cpi->prob_intra_coded < 1)
- cpi->prob_intra_coded = 1;
-
- cpi->prob_last_coded = rf_inter ? (rfct[LAST_FRAME] * 255) / rf_inter : 128;
-
- if (cpi->prob_last_coded < 1)
- cpi->prob_last_coded = 1;
-
- cpi->prob_gf_coded = (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME])
- ? (rfct[GOLDEN_FRAME] * 255) / (rfct[GOLDEN_FRAME] + rfct[ALTREF_FRAME]) : 128;
-
- if (cpi->prob_gf_coded < 1)
- cpi->prob_gf_coded = 1;
+ cpi->prob_intra_coded = 63;
+ cpi->prob_last_coded = 128;
+ cpi->prob_gf_coded = 128;
}
// update reference frame costs since we can do better than what we got last frame.
@@ -3117,7 +2742,6 @@
else if (cpi->common.frames_since_golden == 0)
{
cpi->prob_last_coded = 214;
- cpi->prob_gf_coded = 1;
}
else if (cpi->common.frames_since_golden == 1)
{
@@ -3126,14 +2750,14 @@
}
else if (cpi->source_alt_ref_active)
{
- //int dist = cpi->common.frames_till_alt_ref_frame + cpi->common.frames_since_golden;
cpi->prob_gf_coded -= 20;
if (cpi->prob_gf_coded < 10)
cpi->prob_gf_coded = 10;
}
+ if (!cpi->source_alt_ref_active)
+ cpi->prob_gf_coded = 255;
}
-#endif
}
@@ -3142,12 +2766,12 @@
{
VP8_COMMON *cm = &cpi->common;
- int code_key_frame = FALSE;
+ int code_key_frame = 0;
cpi->kf_boost = 0;
if (cpi->Speed > 11)
- return FALSE;
+ return 0;
// Clear down mmx registers
vp8_clear_system_state(); //__asm emms;
@@ -3189,10 +2813,10 @@
&& (change > .25 || change2 > .25))
{
/*(change > 1.4 || change < .75)&& cpi->this_frame_percent_intra > cpi->last_frame_percent_intra + 3*/
- return TRUE;
+ return 1;
}
- return FALSE;
+ return 0;
}
@@ -3202,7 +2826,7 @@
((cpi->this_frame_percent_intra > 95) &&
(cpi->this_frame_percent_intra >= (cpi->last_frame_percent_intra + 5))))
{
- code_key_frame = TRUE;
+ code_key_frame = 1;
}
// in addition if the following are true and this is not a golden frame then code a key frame
// Note that on golden frames there often seems to be a pop in intra useage anyway hence this
@@ -3215,7 +2839,7 @@
(cpi->this_frame_percent_intra > (cpi->last_frame_percent_intra + 10))))
{
if (!cm->refresh_golden_frame)
- code_key_frame = TRUE;
+ code_key_frame = 1;
}
return code_key_frame;
@@ -3271,11 +2895,11 @@
// Function to test for conditions that indeicate we should loop
// back and recode a frame.
-static BOOL recode_loop_test( VP8_COMP *cpi,
+static int recode_loop_test( VP8_COMP *cpi,
int high_limit, int low_limit,
int q, int maxq, int minq )
{
- BOOL force_recode = FALSE;
+ int force_recode = 0;
VP8_COMMON *cm = &cpi->common;
// Is frame recode allowed at all
@@ -3291,7 +2915,7 @@
if ( ((cpi->projected_frame_size > high_limit) && (q < maxq)) ||
((cpi->projected_frame_size < low_limit) && (q > minq)) )
{
- force_recode = TRUE;
+ force_recode = 1;
}
// Special Constrained quality tests
else if (cpi->oxcf.end_usage == USAGE_CONSTRAINED_QUALITY)
@@ -3301,14 +2925,14 @@
(cpi->projected_frame_size <
((cpi->this_frame_target * 7) >> 3)))
{
- force_recode = TRUE;
+ force_recode = 1;
}
// Severe undershoot and between auto and user cq level
else if ( (q > cpi->oxcf.cq_level) &&
(cpi->projected_frame_size < cpi->min_frame_bandwidth) &&
(cpi->active_best_quality > cpi->oxcf.cq_level))
{
- force_recode = TRUE;
+ force_recode = 1;
cpi->active_best_quality = cpi->oxcf.cq_level;
}
}
@@ -3459,7 +3083,7 @@
int frame_over_shoot_limit;
int frame_under_shoot_limit;
- int Loop = FALSE;
+ int Loop = 0;
int loop_count;
int this_q;
int last_zbin_oq;
@@ -3471,10 +3095,10 @@
int top_index;
int bottom_index;
VP8_COMMON *cm = &cpi->common;
- int active_worst_qchanged = FALSE;
+ int active_worst_qchanged = 0;
- int overshoot_seen = FALSE;
- int undershoot_seen = FALSE;
+ int overshoot_seen = 0;
+ int undershoot_seen = 0;
int drop_mark = cpi->oxcf.drop_frames_water_mark * cpi->oxcf.optimal_buffer_level / 100;
int drop_mark75 = drop_mark * 2 / 3;
int drop_mark50 = drop_mark / 4;
@@ -3485,7 +3109,7 @@
vp8_clear_system_state();
// Test code for segmentation of gf/arf (0,0)
- //segmentation_test_function((VP8_PTR) cpi);
+ //segmentation_test_function( cpi);
if (cpi->compressor_speed == 2)
{
@@ -3525,12 +3149,12 @@
// Enable or disable mode based tweaking of the zbin
// For 2 Pass Only used where GF/ARF prediction quality
// is above a threshold
- cpi->zbin_mode_boost_enabled = TRUE;
+ cpi->zbin_mode_boost_enabled = 1;
if (cpi->pass == 2)
{
if ( cpi->gfu_boost <= 400 )
{
- cpi->zbin_mode_boost_enabled = FALSE;
+ cpi->zbin_mode_boost_enabled = 0;
}
}
@@ -3571,7 +3195,7 @@
}
// The alternate reference frame cannot be active for a key frame
- cpi->source_alt_ref_active = FALSE;
+ cpi->source_alt_ref_active = 0;
// Reset the RD threshold multipliers to default of * 1 (128)
for (i = 0; i < MAX_MODES; i++)
@@ -3583,9 +3207,9 @@
// Test code for segmentation
//if ( (cm->frame_type == KEY_FRAME) || ((cm->current_video_frame % 2) == 0))
//if ( (cm->current_video_frame % 2) == 0 )
- // enable_segmentation((VP8_PTR)cpi);
+ // enable_segmentation(cpi);
//else
- // disable_segmentation((VP8_PTR)cpi);
+ // disable_segmentation(cpi);
#if 0
// Experimental code for lagged compress and one pass
@@ -3679,7 +3303,7 @@
if (cpi->oxcf.number_of_layers > 1)
{
- int i;
+ unsigned int i;
// Propagate bits saved by dropping the frame to higher layers
for (i=cpi->current_layer+1; i<cpi->oxcf.number_of_layers; i++)
@@ -4083,7 +3707,7 @@
vp8_pick_frame_size(cpi);
// Clear the Alt reference frame active flag when we have a key frame
- cpi->source_alt_ref_active = FALSE;
+ cpi->source_alt_ref_active = 0;
// Reset the loop filter deltas and segmentation map
setup_features(cpi);
@@ -4108,7 +3732,7 @@
q_high = cpi->active_worst_quality;
loop_count++;
- Loop = TRUE;
+ Loop = 1;
continue;
}
@@ -4136,10 +3760,10 @@
}
// If we have updated the active max Q do not call vp8_update_rate_correction_factors() this loop.
- active_worst_qchanged = TRUE;
+ active_worst_qchanged = 1;
}
else
- active_worst_qchanged = FALSE;
+ active_worst_qchanged = 0;
#if !(CONFIG_REALTIME_ONLY)
// Special case handling for forced key frames
@@ -4175,7 +3799,7 @@
else if (Q < q_low)
Q = q_low;
- Loop = ((Q != last_q)) ? TRUE : FALSE;
+ Loop = Q != last_q;
}
// Is the projected frame size out of range and are we allowed to attempt to recode.
@@ -4232,7 +3856,7 @@
}
}
- overshoot_seen = TRUE;
+ overshoot_seen = 1;
}
// Frame is too small
else
@@ -4282,7 +3906,7 @@
}
}
- undershoot_seen = TRUE;
+ undershoot_seen = 1;
}
// Clamp Q to upper and lower limits:
@@ -4294,18 +3918,18 @@
// Clamp cpi->zbin_over_quant
cpi->zbin_over_quant = (cpi->zbin_over_quant < zbin_oq_low) ? zbin_oq_low : (cpi->zbin_over_quant > zbin_oq_high) ? zbin_oq_high : cpi->zbin_over_quant;
- //Loop = ((Q != last_q) || (last_zbin_oq != cpi->zbin_over_quant)) ? TRUE : FALSE;
- Loop = ((Q != last_q)) ? TRUE : FALSE;
+ //Loop = (Q != last_q) || (last_zbin_oq != cpi->zbin_over_quant);
+ Loop = Q != last_q;
last_zbin_oq = cpi->zbin_over_quant;
}
else
#endif
- Loop = FALSE;
+ Loop = 0;
if (cpi->is_src_frame_alt_ref)
- Loop = FALSE;
+ Loop = 0;
- if (Loop == TRUE)
+ if (Loop == 1)
{
vp8_restore_coding_context(cpi);
loop_count++;
@@ -4314,7 +3938,7 @@
#endif
}
}
- while (Loop == TRUE);
+ while (Loop == 1);
#if 0
// Experimental code for lagged and one pass
@@ -4456,7 +4080,7 @@
if (cpi->oxcf.number_of_layers > 1)
{
- int i;
+ unsigned int i;
for (i=cpi->current_layer+1; i<cpi->oxcf.number_of_layers; i++)
cpi->layer_context[i].total_byte_count += (*size);
}
@@ -4523,7 +4147,7 @@
(cpi->buffer_level < cpi->oxcf.drop_frames_water_mark * cpi->oxcf.optimal_buffer_level / 100) &&
(cpi->projected_frame_size > (4 * cpi->this_frame_target)))
{
- cpi->drop_frame = TRUE;
+ cpi->drop_frame = 1;
}
#endif
@@ -4567,7 +4191,7 @@
// Propagate values to higher temporal layers
if (cpi->oxcf.number_of_layers > 1)
{
- int i;
+ unsigned int i;
for (i=cpi->current_layer+1; i<cpi->oxcf.number_of_layers; i++)
{
@@ -4882,12 +4506,11 @@
#endif
-int vp8_receive_raw_frame(VP8_PTR ptr, unsigned int frame_flags, YV12_BUFFER_CONFIG *sd, int64_t time_stamp, int64_t end_time)
+int vp8_receive_raw_frame(VP8_COMP *cpi, unsigned int frame_flags, YV12_BUFFER_CONFIG *sd, int64_t time_stamp, int64_t end_time)
{
#if HAVE_ARMV7
int64_t store_reg[8];
#endif
- VP8_COMP *cpi = (VP8_COMP *) ptr;
VP8_COMMON *cm = &cpi->common;
struct vpx_usec_timer timer;
int res = 0;
@@ -4936,12 +4559,11 @@
}
-int vp8_get_compressed_data(VP8_PTR ptr, unsigned int *frame_flags, unsigned long *size, unsigned char *dest, unsigned char *dest_end, int64_t *time_stamp, int64_t *time_end, int flush)
+int vp8_get_compressed_data(VP8_COMP *cpi, unsigned int *frame_flags, unsigned long *size, unsigned char *dest, unsigned char *dest_end, int64_t *time_stamp, int64_t *time_end, int flush)
{
#if HAVE_ARMV7
int64_t store_reg[8];
#endif
- VP8_COMP *cpi = (VP8_COMP *) ptr;
VP8_COMMON *cm;
struct vpx_usec_timer tsctimer;
struct vpx_usec_timer ticktimer;
@@ -4995,7 +4617,7 @@
cm->refresh_golden_frame = 0;
cm->refresh_last_frame = 0;
cm->show_frame = 0;
- cpi->source_alt_ref_pending = FALSE; // Clear Pending alt Ref flag.
+ cpi->source_alt_ref_pending = 0; // Clear Pending alt Ref flag.
cpi->is_src_frame_alt_ref = 0;
}
}
@@ -5216,6 +4838,17 @@
vpx_memcpy(&cm->fc, &cm->lfc, sizeof(cm->fc));
}
+ // Save the contexts separately for alt ref, gold and last.
+ // (TODO jbb -> Optimize this with pointers to avoid extra copies. )
+ if(cm->refresh_alt_ref_frame)
+ vpx_memcpy(&cpi->lfc_a, &cm->fc, sizeof(cm->fc));
+
+ if(cm->refresh_golden_frame)
+ vpx_memcpy(&cpi->lfc_g, &cm->fc, sizeof(cm->fc));
+
+ if(cm->refresh_last_frame)
+ vpx_memcpy(&cpi->lfc_n, &cm->fc, sizeof(cm->fc));
+
// if its a dropped frame honor the requests on subsequent frames
if (*size > 0)
{
@@ -5416,10 +5049,8 @@
return 0;
}
-int vp8_get_preview_raw_frame(VP8_PTR comp, YV12_BUFFER_CONFIG *dest, vp8_ppflags_t *flags)
+int vp8_get_preview_raw_frame(VP8_COMP *cpi, YV12_BUFFER_CONFIG *dest, vp8_ppflags_t *flags)
{
- VP8_COMP *cpi = (VP8_COMP *) comp;
-
if (cpi->common.refresh_alt_ref_frame)
return -1;
else
@@ -5448,9 +5079,8 @@
}
}
-int vp8_set_roimap(VP8_PTR comp, unsigned char *map, unsigned int rows, unsigned int cols, int delta_q[4], int delta_lf[4], unsigned int threshold[4])
+int vp8_set_roimap(VP8_COMP *cpi, unsigned char *map, unsigned int rows, unsigned int cols, int delta_q[4], int delta_lf[4], unsigned int threshold[4])
{
- VP8_COMP *cpi = (VP8_COMP *) comp;
signed char feature_data[MB_LVL_MAX][MAX_MB_SEGMENTS];
if (cpi->common.mb_rows != rows || cpi->common.mb_cols != cols)
@@ -5458,15 +5088,15 @@
if (!map)
{
- disable_segmentation((VP8_PTR)cpi);
+ disable_segmentation(cpi);
return 0;
}
// Set the segmentation Map
- set_segmentation_map((VP8_PTR)cpi, map);
+ set_segmentation_map(cpi, map);
// Activate segmentation.
- enable_segmentation((VP8_PTR)cpi);
+ enable_segmentation(cpi);
// Set up the quant segment data
feature_data[MB_LVL_ALT_Q][0] = delta_q[0];
@@ -5487,15 +5117,13 @@
// Initialise the feature data structure
// SEGMENT_DELTADATA 0, SEGMENT_ABSDATA 1
- set_segment_data((VP8_PTR)cpi, &feature_data[0][0], SEGMENT_DELTADATA);
+ set_segment_data(cpi, &feature_data[0][0], SEGMENT_DELTADATA);
return 0;
}
-int vp8_set_active_map(VP8_PTR comp, unsigned char *map, unsigned int rows, unsigned int cols)
+int vp8_set_active_map(VP8_COMP *cpi, unsigned char *map, unsigned int rows, unsigned int cols)
{
- VP8_COMP *cpi = (VP8_COMP *) comp;
-
if (rows == cpi->common.mb_rows && cols == cpi->common.mb_cols)
{
if (map)
@@ -5515,10 +5143,8 @@
}
}
-int vp8_set_internal_size(VP8_PTR comp, VPX_SCALING horiz_mode, VPX_SCALING vert_mode)
+int vp8_set_internal_size(VP8_COMP *cpi, VPX_SCALING horiz_mode, VPX_SCALING vert_mode)
{
- VP8_COMP *cpi = (VP8_COMP *) comp;
-
if (horiz_mode <= ONETWO)
cpi->common.horiz_scale = horiz_mode;
else
@@ -5560,8 +5186,7 @@
}
-int vp8_get_quantizer(VP8_PTR c)
+int vp8_get_quantizer(VP8_COMP *cpi)
{
- VP8_COMP *cpi = (VP8_COMP *) c;
return cpi->common.base_qindex;
}
diff --git a/vp8/encoder/onyx_int.h b/vp8/encoder/onyx_int.h
index 68ec3ab..46951e3 100644
--- a/vp8/encoder/onyx_int.h
+++ b/vp8/encoder/onyx_int.h
@@ -259,7 +259,7 @@
int buffer_level;
int bits_off_target;
- long long total_actual_bits;
+ int64_t total_actual_bits;
int total_target_vs_actual;
int worst_quality;
@@ -279,7 +279,7 @@
int zbin_over_quant;
int inter_frame_target;
- INT64 total_byte_count;
+ int64_t total_byte_count;
int filter_level;
@@ -568,16 +568,21 @@
int base_skip_false_prob[128];
+ FRAME_CONTEXT lfc_n; /* last frame entropy */
+ FRAME_CONTEXT lfc_a; /* last alt ref entropy */
+ FRAME_CONTEXT lfc_g; /* last gold ref entropy */
+
+
struct twopass_rc
{
unsigned int section_intra_rating;
double section_max_qfactor;
unsigned int next_iiratio;
unsigned int this_iiratio;
- FIRSTPASS_STATS *total_stats;
- FIRSTPASS_STATS *this_frame_stats;
+ FIRSTPASS_STATS total_stats;
+ FIRSTPASS_STATS this_frame_stats;
FIRSTPASS_STATS *stats_in, *stats_in_end, *stats_in_start;
- FIRSTPASS_STATS *total_left_stats;
+ FIRSTPASS_STATS total_left_stats;
int first_pass_done;
int64_t bits_left;
int64_t clip_bits_total;
@@ -668,8 +673,8 @@
unsigned int current_layer;
LAYER_CONTEXT layer_context[MAX_LAYERS];
- long long frames_in_layer[MAX_LAYERS];
- long long bytes_in_layer[MAX_LAYERS];
+ int64_t frames_in_layer[MAX_LAYERS];
+ int64_t bytes_in_layer[MAX_LAYERS];
double sum_psnr[MAX_LAYERS];
double sum_psnr_p[MAX_LAYERS];
double total_error2[MAX_LAYERS];
diff --git a/vp8/encoder/pickinter.c b/vp8/encoder/pickinter.c
index d9c8975..46f53a1 100644
--- a/vp8/encoder/pickinter.c
+++ b/vp8/encoder/pickinter.c
@@ -21,7 +21,6 @@
#include "vp8/common/reconinter.h"
#include "vp8/common/reconintra.h"
#include "vp8/common/reconintra4x4.h"
-#include "vp8/common/g_common.h"
#include "variance.h"
#include "mcomp.h"
#include "rdopt.h"
@@ -402,9 +401,68 @@
}
}
+
+#if CONFIG_MULTI_RES_ENCODING
+static
+void get_lower_res_motion_info(VP8_COMP *cpi, MACROBLOCKD *xd, int *dissim,
+ int *parent_ref_frame,
+ MB_PREDICTION_MODE *parent_mode,
+ int_mv *parent_ref_mv, int mb_row, int mb_col)
+{
+ LOWER_RES_INFO* store_mode_info
+ = (LOWER_RES_INFO*)cpi->oxcf.mr_low_res_mode_info;
+ unsigned int parent_mb_index;
+ //unsigned int parent_mb_index = map_640x480_to_320x240[mb_row][mb_col];
+
+ /* Consider different down_sampling_factor. */
+ {
+ /* TODO: Removed the loop that supports special down_sampling_factor
+ * such as 2, 4, 8. Will revisit it if needed.
+ * Should also try using a look-up table to see if it helps
+ * performance. */
+ int round = cpi->oxcf.mr_down_sampling_factor.num/2;
+ int parent_mb_row, parent_mb_col;
+
+ parent_mb_row = (mb_row*cpi->oxcf.mr_down_sampling_factor.den+round)
+ /cpi->oxcf.mr_down_sampling_factor.num;
+ parent_mb_col = (mb_col*cpi->oxcf.mr_down_sampling_factor.den+round)
+ /cpi->oxcf.mr_down_sampling_factor.num;
+ parent_mb_index = parent_mb_row*cpi->mr_low_res_mb_cols + parent_mb_col;
+ }
+
+ /* Read lower-resolution mode & motion result from memory.*/
+ *parent_ref_frame = store_mode_info[parent_mb_index].ref_frame;
+ *parent_mode = store_mode_info[parent_mb_index].mode;
+ *dissim = store_mode_info[parent_mb_index].dissim;
+
+ /* For highest-resolution encoder, adjust dissim value. Lower its quality
+ * for good performance. */
+ if (cpi->oxcf.mr_encoder_id == (cpi->oxcf.mr_total_resolutions - 1))
+ *dissim>>=1;
+
+ if(*parent_ref_frame != INTRA_FRAME)
+ {
+ /* Consider different down_sampling_factor.
+ * The result can be rounded to be more precise, but it takes more time.
+ */
+ //int round = cpi->oxcf.mr_down_sampling_factor.den/2;
+ (*parent_ref_mv).as_mv.row = store_mode_info[parent_mb_index].mv.as_mv.row
+ *cpi->oxcf.mr_down_sampling_factor.num
+ /cpi->oxcf.mr_down_sampling_factor.den;
+ (*parent_ref_mv).as_mv.col = store_mode_info[parent_mb_index].mv.as_mv.col
+ *cpi->oxcf.mr_down_sampling_factor.num
+ /cpi->oxcf.mr_down_sampling_factor.den;
+
+ vp8_clamp_mv2(parent_ref_mv, xd);
+ }
+}
+#endif
+
+
void vp8_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset,
int recon_uvoffset, int *returnrate,
- int *returndistortion, int *returnintra)
+ int *returndistortion, int *returnintra, int mb_row,
+ int mb_col)
{
BLOCK *b = &x->block[0];
BLOCKD *d = &x->e_mbd.block[0];
@@ -422,33 +480,38 @@
int rate;
int rate2;
int distortion2;
- int bestsme;
- //int all_rds[MAX_MODES]; // Experimental debug code.
+ int bestsme = INT_MAX;
int best_mode_index = 0;
unsigned int sse = INT_MAX, best_sse = INT_MAX;
int_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)
- int_mv nearest_mv[4];
- int_mv near_mv[4];
- int_mv frame_best_ref_mv[4];
- int MDCounts[4][4];
unsigned char *y_buffer[4];
unsigned char *u_buffer[4];
unsigned char *v_buffer[4];
int i;
int ref_frame_map[4];
+ int sign_bias = 0;
- int found_near_mvs[4] = {0, 0, 0, 0};
+ int have_subp_search = cpi->sf.half_pixel_search; /* In real-time mode,
+ when Speed >= 15, no sub-pixel search. */
- int have_subp_search = cpi->sf.half_pixel_search; /* In real-time mode, when Speed >= 15, no sub-pixel search. */
+#if CONFIG_MULTI_RES_ENCODING
+ int dissim = INT_MAX;
+ int parent_ref_frame = 0;
+ int_mv parent_ref_mv;
+ MB_PREDICTION_MODE parent_mode = 0;
+
+ if (cpi->oxcf.mr_encoder_id)
+ get_lower_res_motion_info(cpi, xd, &dissim, &parent_ref_frame,
+ &parent_mode, &parent_ref_mv, mb_row, mb_col);
+#endif
vpx_memset(mode_mv, 0, sizeof(mode_mv));
- vpx_memset(nearest_mv, 0, sizeof(nearest_mv));
- vpx_memset(near_mv, 0, sizeof(near_mv));
vpx_memset(&best_mbmode, 0, sizeof(best_mbmode));
/* Setup search priorities */
@@ -463,6 +526,22 @@
for(; i<4; i++)
ref_frame_map[i] = -1;
+ /* Check to see if there is at least 1 valid reference frame that we need
+ * to calculate near_mvs.
+ */
+ if (ref_frame_map[1] > 0)
+ {
+ vp8_find_near_mvs(&x->e_mbd,
+ x->e_mbd.mode_info_context,
+ &mode_mv[NEARESTMV], &mode_mv[NEARMV],
+ &best_ref_mv,
+ mdcounts,
+ ref_frame_map[1],
+ cpi->common.ref_frame_sign_bias);
+
+ sign_bias = cpi->common.ref_frame_sign_bias[ref_frame_map[1]];
+ }
+
// set up all the refframe dependent pointers.
if (cpi->ref_frame_flags & VP8_LAST_FLAG)
{
@@ -488,7 +567,7 @@
v_buffer[ALTREF_FRAME] = alt_yv12->v_buffer + recon_uvoffset;
}
- cpi->mbs_tested_so_far++; // Count of the number of MBs tested so far this frame
+ cpi->mbs_tested_so_far++; // Count of the number of MBs tested so far this frame
*returnintra = INT_MAX;
x->skip = 0;
@@ -511,59 +590,21 @@
x->e_mbd.mode_info_context->mbmi.ref_frame = this_ref_frame;
-
- // Check to see if the testing frequency for this mode is at its max
- // If so then prevent it from being tested and increase the threshold for its testing
- if (cpi->mode_test_hit_counts[mode_index] && (cpi->mode_check_freq[mode_index] > 1))
+#if CONFIG_MULTI_RES_ENCODING
+ if (cpi->oxcf.mr_encoder_id)
{
- //if ( (cpi->mbs_tested_so_far / cpi->mode_test_hit_counts[mode_index]) <= cpi->mode_check_freq[mode_index] )
- if (cpi->mbs_tested_so_far <= (cpi->mode_check_freq[mode_index] * cpi->mode_test_hit_counts[mode_index]))
- {
- // Increase the threshold for coding this mode to make it less likely to be chosen
- cpi->rd_thresh_mult[mode_index] += 4;
-
- if (cpi->rd_thresh_mult[mode_index] > MAX_THRESHMULT)
- cpi->rd_thresh_mult[mode_index] = MAX_THRESHMULT;
-
- cpi->rd_threshes[mode_index] = (cpi->rd_baseline_thresh[mode_index] >> 7) * cpi->rd_thresh_mult[mode_index];
-
+ /* If parent MB is intra, child MB is intra. */
+ if (!parent_ref_frame && this_ref_frame)
continue;
- }
+
+ /* If parent MB is inter, and it is unlikely there are multiple
+ * objects in parent MB, we use parent ref frame as child MB's
+ * ref frame. */
+ if (parent_ref_frame && dissim < 8
+ && parent_ref_frame != this_ref_frame)
+ continue;
}
-
- // If nearby MVs haven't been found for this reference frame then do it now.
- if (x->e_mbd.mode_info_context->mbmi.ref_frame != INTRA_FRAME &&
- !found_near_mvs[x->e_mbd.mode_info_context->mbmi.ref_frame])
- {
- int ref_frame = x->e_mbd.mode_info_context->mbmi.ref_frame;
- vp8_find_near_mvs(&x->e_mbd,
- x->e_mbd.mode_info_context,
- &nearest_mv[ref_frame], &near_mv[ref_frame],
- &frame_best_ref_mv[ref_frame],
- MDCounts[ref_frame],
- ref_frame,
- cpi->common.ref_frame_sign_bias);
- found_near_mvs[ref_frame] = 1;
- }
-
- // We have now reached the point where we are going to test the current mode so increment the counter for the number of times it has been tested
- cpi->mode_test_hit_counts[mode_index] ++;
-
- rate2 = 0;
- distortion2 = 0;
-
- this_mode = vp8_mode_order[mode_index];
-
- // Experimental debug code.
- //all_rds[mode_index] = -1;
-
- x->e_mbd.mode_info_context->mbmi.mode = this_mode;
- x->e_mbd.mode_info_context->mbmi.uv_mode = DC_PRED;
-
- // Work out the cost assosciated with selecting the reference frame
- frame_cost =
- x->e_mbd.ref_frame_cost[x->e_mbd.mode_info_context->mbmi.ref_frame];
- rate2 += frame_cost;
+#endif
// everything but intra
if (x->e_mbd.mode_info_context->mbmi.ref_frame)
@@ -571,25 +612,95 @@
x->e_mbd.pre.y_buffer = y_buffer[x->e_mbd.mode_info_context->mbmi.ref_frame];
x->e_mbd.pre.u_buffer = u_buffer[x->e_mbd.mode_info_context->mbmi.ref_frame];
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_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));
+
+ if (sign_bias !=
+ cpi->common.ref_frame_sign_bias[x->e_mbd.mode_info_context->mbmi.ref_frame])
+ {
+ mode_mv[NEARESTMV].as_mv.row *= -1;
+ mode_mv[NEARESTMV].as_mv.col *= -1;
+ mode_mv[NEARMV].as_mv.row *= -1;
+ mode_mv[NEARMV].as_mv.col *= -1;
+ best_ref_mv.as_mv.row *= -1;
+ best_ref_mv.as_mv.col *= -1;
+ sign_bias
+ = cpi->common.ref_frame_sign_bias[x->e_mbd.mode_info_context->mbmi.ref_frame];
+ }
+
+#if CONFIG_MULTI_RES_ENCODING
+ if (cpi->oxcf.mr_encoder_id)
+ {
+ if (vp8_mode_order[mode_index] == NEARESTMV &&
+ mode_mv[NEARESTMV].as_int ==0)
+ continue;
+ if (vp8_mode_order[mode_index] == NEARMV &&
+ mode_mv[NEARMV].as_int ==0)
+ continue;
+
+ if (vp8_mode_order[mode_index] == NEWMV && parent_mode == ZEROMV
+ && best_ref_mv.as_int==0) //&& dissim==0
+ continue;
+ else if(vp8_mode_order[mode_index] == NEWMV && dissim==0
+ && best_ref_mv.as_int==parent_ref_mv.as_int)
+ continue;
+ }
+#endif
}
- // Only consider ZEROMV/ALTREF_FRAME for alt ref frame,
- // unless ARNR filtering is enabled in which case we want
- // an unfiltered alternative
+ /* Check to see if the testing frequency for this mode is at its max
+ * If so then prevent it from being tested and increase the threshold
+ * for its testing */
+ if (cpi->mode_test_hit_counts[mode_index] &&
+ (cpi->mode_check_freq[mode_index] > 1))
+ {
+ if (cpi->mbs_tested_so_far <= (cpi->mode_check_freq[mode_index] *
+ cpi->mode_test_hit_counts[mode_index]))
+ {
+ /* Increase the threshold for coding this mode to make it less
+ * likely to be chosen */
+ cpi->rd_thresh_mult[mode_index] += 4;
+
+ if (cpi->rd_thresh_mult[mode_index] > MAX_THRESHMULT)
+ cpi->rd_thresh_mult[mode_index] = MAX_THRESHMULT;
+
+ cpi->rd_threshes[mode_index] =
+ (cpi->rd_baseline_thresh[mode_index] >> 7) *
+ cpi->rd_thresh_mult[mode_index];
+ continue;
+ }
+ }
+
+ /* We have now reached the point where we are going to test the current
+ * mode so increment the counter for the number of times it has been
+ * tested */
+ cpi->mode_test_hit_counts[mode_index] ++;
+
+ rate2 = 0;
+ distortion2 = 0;
+
+ this_mode = vp8_mode_order[mode_index];
+
+ x->e_mbd.mode_info_context->mbmi.mode = this_mode;
+ x->e_mbd.mode_info_context->mbmi.uv_mode = DC_PRED;
+
+ /* Work out the cost assosciated with selecting the reference frame */
+ frame_cost =
+ x->e_mbd.ref_frame_cost[x->e_mbd.mode_info_context->mbmi.ref_frame];
+ rate2 += frame_cost;
+
+ /* Only consider ZEROMV/ALTREF_FRAME for alt ref frame,
+ * unless ARNR filtering is enabled in which case we want
+ * an unfiltered alternative */
if (cpi->is_src_frame_alt_ref && (cpi->oxcf.arnr_max_frames == 0))
{
- if (this_mode != ZEROMV || x->e_mbd.mode_info_context->mbmi.ref_frame != ALTREF_FRAME)
+ if (this_mode != ZEROMV ||
+ x->e_mbd.mode_info_context->mbmi.ref_frame != ALTREF_FRAME)
continue;
}
switch (this_mode)
{
case B_PRED:
- // Pass best so far to pick_intra4x4mby_modes to use as breakout
+ /* Pass best so far to pick_intra4x4mby_modes to use as breakout */
distortion2 = best_sse;
pick_intra4x4mby_modes(IF_RTCD(&cpi->rtcd), x, &rate, &distortion2);
@@ -648,10 +759,12 @@
int sadpb = x->sadperbit16;
int_mv mvp_full;
- int col_min = (best_ref_mv.as_mv.col>>3) - MAX_FULL_PEL_VAL + ((best_ref_mv.as_mv.col & 7)?1:0);
- int row_min = (best_ref_mv.as_mv.row>>3) - MAX_FULL_PEL_VAL + ((best_ref_mv.as_mv.row & 7)?1:0);
- int col_max = (best_ref_mv.as_mv.col>>3) + MAX_FULL_PEL_VAL;
- int row_max = (best_ref_mv.as_mv.row>>3) + MAX_FULL_PEL_VAL;
+ int col_min = ((best_ref_mv.as_mv.col+7)>>3) - MAX_FULL_PEL_VAL;
+ int row_min = ((best_ref_mv.as_mv.row+7)>>3) - MAX_FULL_PEL_VAL;
+ int col_max = (best_ref_mv.as_mv.col>>3)
+ + MAX_FULL_PEL_VAL;
+ int row_max = (best_ref_mv.as_mv.row>>3)
+ + MAX_FULL_PEL_VAL;
int tmp_col_min = x->mv_col_min;
int tmp_col_max = x->mv_col_max;
@@ -663,118 +776,156 @@
// Further step/diamond searches as necessary
step_param = cpi->sf.first_step + speed_adjust;
- if(cpi->sf.improved_mv_pred)
+#if CONFIG_MULTI_RES_ENCODING
+ if (cpi->oxcf.mr_encoder_id)
{
- if(!saddone)
- {
- vp8_cal_sad(cpi,xd,x, recon_yoffset ,&near_sadidx[0] );
- saddone = 1;
- }
+ // Use parent MV as predictor. Adjust search range accordingly.
+ mvp.as_int = parent_ref_mv.as_int;
+ mvp_full.as_mv.col = parent_ref_mv.as_mv.col>>3;
+ mvp_full.as_mv.row = parent_ref_mv.as_mv.row>>3;
- 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]);
-
- sr += speed_adjust;
- //adjust search range according to sr from mv prediction
- if(sr > step_param)
- step_param = sr;
-
- mvp_full.as_mv.col = mvp.as_mv.col>>3;
- mvp_full.as_mv.row = mvp.as_mv.row>>3;
-
+ if(dissim <=32) step_param += 3;
+ else if(dissim <=128) step_param += 2;
+ else step_param += 1;
}else
+#endif
{
- mvp.as_int = best_ref_mv.as_int;
- mvp_full.as_mv.col = best_ref_mv.as_mv.col>>3;
- mvp_full.as_mv.row = best_ref_mv.as_mv.row>>3;
+ if(cpi->sf.improved_mv_pred)
+ {
+ 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]);
+
+ sr += speed_adjust;
+ //adjust search range according to sr from mv prediction
+ if(sr > step_param)
+ step_param = sr;
+
+ mvp_full.as_mv.col = mvp.as_mv.col>>3;
+ mvp_full.as_mv.row = mvp.as_mv.row>>3;
+ }else
+ {
+ mvp.as_int = best_ref_mv.as_int;
+ mvp_full.as_mv.col = best_ref_mv.as_mv.col>>3;
+ mvp_full.as_mv.row = best_ref_mv.as_mv.row>>3;
+ }
}
- // 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_steps = (cpi->Speed >= 8)? 0: (cpi->sf.max_step_search_steps - 1 - step_param);
-
- if (cpi->sf.search_method == HEX)
+#if CONFIG_MULTI_RES_ENCODING
+ if (cpi->oxcf.mr_encoder_id && dissim <= 2 &&
+ MAX(abs(best_ref_mv.as_mv.row - parent_ref_mv.as_mv.row),
+ abs(best_ref_mv.as_mv.col - parent_ref_mv.as_mv.col)) <= 4)
{
+ d->bmi.mv.as_int = mvp_full.as_int;
+ mode_mv[NEWMV].as_int = mvp_full.as_int;
+
+ cpi->find_fractional_mv_step(x, b, d, &d->bmi.mv, &best_ref_mv,
+ x->errorperbit,
+ &cpi->fn_ptr[BLOCK_16X16],
+ cpi->mb.mvcost,
+ &distortion2,&sse);
+ }else
+#endif
+ {
+ /* 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_steps = (cpi->Speed >= 8)?
+ 0: (cpi->sf.max_step_search_steps - 1 - step_param);
+
+ if (cpi->sf.search_method == HEX)
+ {
#if CONFIG_MULTI_RES_ENCODING
/* TODO: In higher-res pick_inter_mode, step_param is used to
* modify hex search range. Here, set step_param to 0 not to
* change the behavior in lowest-resolution encoder.
* Will improve it later.
*/
- step_param = 0;
+ if (!cpi->oxcf.mr_encoder_id)
+ step_param = 0;
#endif
- bestsme = vp8_hex_search(x, b, d, &mvp_full, &d->bmi.mv, step_param,
- sadpb, &cpi->fn_ptr[BLOCK_16X16],
- x->mvsadcost, x->mvcost, &best_ref_mv);
- mode_mv[NEWMV].as_int = d->bmi.mv.as_int;
- }
- else
- {
- bestsme = cpi->diamond_search_sad(x, b, d, &mvp_full, &d->bmi.mv,
- step_param, sadpb, &num00,
- &cpi->fn_ptr[BLOCK_16X16],
- x->mvcost, &best_ref_mv);
- mode_mv[NEWMV].as_int = d->bmi.mv.as_int;
-
- // Further step/diamond searches as necessary
- n = 0;
- //further_steps = (cpi->sf.max_step_search_steps - 1) - step_param;
-
- n = num00;
- num00 = 0;
-
- while (n < further_steps)
+ bestsme = vp8_hex_search(x, b, d, &mvp_full, &d->bmi.mv,
+ step_param, sadpb,
+ &cpi->fn_ptr[BLOCK_16X16],
+ x->mvsadcost, x->mvcost, &best_ref_mv);
+ mode_mv[NEWMV].as_int = d->bmi.mv.as_int;
+ }
+ else
{
- n++;
+ bestsme = cpi->diamond_search_sad(x, b, d, &mvp_full,
+ &d->bmi.mv, step_param, sadpb, &num00,
+ &cpi->fn_ptr[BLOCK_16X16],
+ x->mvcost, &best_ref_mv);
+ mode_mv[NEWMV].as_int = d->bmi.mv.as_int;
- if (num00)
- num00--;
- else
+ // Further step/diamond searches as necessary
+ n = 0;
+ //further_steps = (cpi->sf.max_step_search_steps - 1) - step_param;
+
+ n = num00;
+ num00 = 0;
+
+ while (n < further_steps)
{
- thissme =
- cpi->diamond_search_sad(x, b, d, &mvp_full,
- &d->bmi.mv,
- step_param + n,
- sadpb, &num00,
- &cpi->fn_ptr[BLOCK_16X16],
- x->mvcost, &best_ref_mv);
- if (thissme < bestsme)
- {
- bestsme = thissme;
- mode_mv[NEWMV].as_int = d->bmi.mv.as_int;
- }
+ n++;
+
+ if (num00)
+ num00--;
else
{
- d->bmi.mv.as_int = mode_mv[NEWMV].as_int;
+ thissme =
+ cpi->diamond_search_sad(x, b, d, &mvp_full,
+ &d->bmi.mv,
+ step_param + n,
+ sadpb, &num00,
+ &cpi->fn_ptr[BLOCK_16X16],
+ x->mvcost, &best_ref_mv);
+ if (thissme < bestsme)
+ {
+ bestsme = thissme;
+ mode_mv[NEWMV].as_int = d->bmi.mv.as_int;
+ }
+ else
+ {
+ d->bmi.mv.as_int = mode_mv[NEWMV].as_int;
+ }
}
}
}
- }
- 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;
+ 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, &best_ref_mv,
- x->errorperbit,
+ if (bestsme < INT_MAX)
+ cpi->find_fractional_mv_step(x, b, d, &d->bmi.mv,
+ &best_ref_mv, x->errorperbit,
&cpi->fn_ptr[BLOCK_16X16],
cpi->mb.mvcost,
&distortion2,&sse);
+ }
mode_mv[NEWMV].as_int = d->bmi.mv.as_int;
// mv cost;
- rate2 += vp8_mv_bit_cost(&mode_mv[NEWMV], &best_ref_mv, cpi->mb.mvcost, 128);
+ rate2 += vp8_mv_bit_cost(&mode_mv[NEWMV], &best_ref_mv,
+ cpi->mb.mvcost, 128);
}
case NEARESTMV:
@@ -785,18 +936,23 @@
case ZEROMV:
- // Trap vectors that reach beyond the UMV borders
- // Note that ALL New MV, Nearest MV Near MV and Zero MV code drops through to this point
- // because of the lack of break statements in the previous two cases.
- if (((mode_mv[this_mode].as_mv.row >> 3) < x->mv_row_min) || ((mode_mv[this_mode].as_mv.row >> 3) > x->mv_row_max) ||
- ((mode_mv[this_mode].as_mv.col >> 3) < x->mv_col_min) || ((mode_mv[this_mode].as_mv.col >> 3) > x->mv_col_max))
+ /* Trap vectors that reach beyond the UMV borders
+ * Note that ALL New MV, Nearest MV Near MV and Zero MV code drops
+ * through to this point because of the lack of break statements
+ * in the previous two cases.
+ */
+ if (((mode_mv[this_mode].as_mv.row >> 3) < x->mv_row_min) ||
+ ((mode_mv[this_mode].as_mv.row >> 3) > x->mv_row_max) ||
+ ((mode_mv[this_mode].as_mv.col >> 3) < x->mv_col_min) ||
+ ((mode_mv[this_mode].as_mv.col >> 3) > x->mv_col_max))
continue;
rate2 += vp8_cost_mv_ref(this_mode, mdcounts);
x->e_mbd.mode_info_context->mbmi.mv.as_int =
mode_mv[this_mode].as_int;
- /* Exit early and don't compute the distortion if this macroblock is marked inactive. */
+ /* Exit early and don't compute the distortion if this macroblock
+ * is marked inactive. */
if (cpi->active_map_enabled && x->active_ptr[0] == 0)
{
sse = 0;
@@ -831,9 +987,6 @@
break;
}
- // Experimental debug code.
- //all_rds[mode_index] = this_rd;
-
if (this_rd < best_rd || x->skip)
{
// Note index of best mode
@@ -843,14 +996,23 @@
*returndistortion = distortion2;
best_sse = sse;
best_rd = this_rd;
- vpx_memcpy(&best_mbmode, &x->e_mbd.mode_info_context->mbmi, sizeof(MB_MODE_INFO));
+ vpx_memcpy(&best_mbmode, &x->e_mbd.mode_info_context->mbmi,
+ sizeof(MB_MODE_INFO));
- // Testing this mode gave rise to an improvement in best error score. Lower threshold a bit for next time
- cpi->rd_thresh_mult[mode_index] = (cpi->rd_thresh_mult[mode_index] >= (MIN_THRESHMULT + 2)) ? cpi->rd_thresh_mult[mode_index] - 2 : MIN_THRESHMULT;
- cpi->rd_threshes[mode_index] = (cpi->rd_baseline_thresh[mode_index] >> 7) * cpi->rd_thresh_mult[mode_index];
+ /* Testing this mode gave rise to an improvement in best error
+ * score. Lower threshold a bit for next time
+ */
+ cpi->rd_thresh_mult[mode_index] =
+ (cpi->rd_thresh_mult[mode_index] >= (MIN_THRESHMULT + 2)) ?
+ cpi->rd_thresh_mult[mode_index] - 2 : MIN_THRESHMULT;
+ cpi->rd_threshes[mode_index] =
+ (cpi->rd_baseline_thresh[mode_index] >> 7) *
+ cpi->rd_thresh_mult[mode_index];
}
- // If the mode did not help improve the best error case then raise the threshold for testing that mode next time around.
+ /* If the mode did not help improve the best error case then raise the
+ * threshold for testing that mode next time around.
+ */
else
{
cpi->rd_thresh_mult[mode_index] += 4;
@@ -858,7 +1020,9 @@
if (cpi->rd_thresh_mult[mode_index] > MAX_THRESHMULT)
cpi->rd_thresh_mult[mode_index] = MAX_THRESHMULT;
- cpi->rd_threshes[mode_index] = (cpi->rd_baseline_thresh[mode_index] >> 7) * cpi->rd_thresh_mult[mode_index];
+ cpi->rd_threshes[mode_index] =
+ (cpi->rd_baseline_thresh[mode_index] >> 7) *
+ cpi->rd_thresh_mult[mode_index];
}
if (x->skip)
@@ -870,8 +1034,14 @@
{
int best_adjustment = (cpi->rd_thresh_mult[best_mode_index] >> 3);
- cpi->rd_thresh_mult[best_mode_index] = (cpi->rd_thresh_mult[best_mode_index] >= (MIN_THRESHMULT + best_adjustment)) ? cpi->rd_thresh_mult[best_mode_index] - best_adjustment : MIN_THRESHMULT;
- cpi->rd_threshes[best_mode_index] = (cpi->rd_baseline_thresh[best_mode_index] >> 7) * cpi->rd_thresh_mult[best_mode_index];
+ cpi->rd_thresh_mult[best_mode_index] =
+ (cpi->rd_thresh_mult[best_mode_index]
+ >= (MIN_THRESHMULT + best_adjustment)) ?
+ cpi->rd_thresh_mult[best_mode_index] - best_adjustment :
+ MIN_THRESHMULT;
+ cpi->rd_threshes[best_mode_index] =
+ (cpi->rd_baseline_thresh[best_mode_index] >> 7) *
+ cpi->rd_thresh_mult[best_mode_index];
}
@@ -894,15 +1064,17 @@
x->e_mbd.mode_info_context->mbmi.mv.as_int = 0;
x->e_mbd.mode_info_context->mbmi.uv_mode = DC_PRED;
x->e_mbd.mode_info_context->mbmi.mb_skip_coeff =
- (cpi->common.mb_no_coeff_skip) ? 1 : 0;
+ (cpi->common.mb_no_coeff_skip);
x->e_mbd.mode_info_context->mbmi.partitioning = 0;
return;
}
- /* set to the best mb mode, this copy can be skip if x->skip since it already has the right content */
+ /* set to the best mb mode, this copy can be skip if x->skip since it
+ * already has the right content */
if (!x->skip)
- vpx_memcpy(&x->e_mbd.mode_info_context->mbmi, &best_mbmode, sizeof(MB_MODE_INFO));
+ vpx_memcpy(&x->e_mbd.mode_info_context->mbmi, &best_mbmode,
+ sizeof(MB_MODE_INFO));
if (best_mbmode.mode <= B_PRED)
{
@@ -910,7 +1082,14 @@
pick_intra_mbuv_mode(x);
}
- update_mvcount(cpi, &x->e_mbd, &frame_best_ref_mv[xd->mode_info_context->mbmi.ref_frame]);
+ if (sign_bias
+ != cpi->common.ref_frame_sign_bias[xd->mode_info_context->mbmi.ref_frame])
+ {
+ best_ref_mv.as_mv.row *= -1;
+ best_ref_mv.as_mv.col *= -1;
+ }
+
+ update_mvcount(cpi, &x->e_mbd, &best_ref_mv);
}
@@ -957,568 +1136,3 @@
*rate_ = best_rate;
}
-
-#if CONFIG_MULTI_RES_ENCODING
-void vp8_mr_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset,
- int recon_uvoffset, int *returnrate,
- int *returndistortion, int *returnintra, int mb_row,
- int mb_col)
-{
- BLOCK *b = &x->block[0];
- BLOCKD *d = &x->e_mbd.block[0];
- MACROBLOCKD *xd = &x->e_mbd;
- MB_MODE_INFO best_mbmode;
-
- int_mv best_ref_mv;
- int_mv mode_mv[MB_MODE_COUNT];
- MB_PREDICTION_MODE this_mode;
- int num00;
- int mdcounts[4];
- int best_rd = INT_MAX; // 1 << 30;
- int best_intra_rd = INT_MAX;
- int mode_index;
- int rate;
- int rate2;
- int distortion2;
- int bestsme;
- int best_mode_index = 0;
- unsigned int sse = INT_MAX, best_sse = INT_MAX;
-
- int_mv mvp;
- int_mv nearest_mv[4];
- int_mv near_mv[4];
- int_mv frame_best_ref_mv[4];
- int MDCounts[4][4];
- unsigned char *y_buffer[4];
- unsigned char *u_buffer[4];
- unsigned char *v_buffer[4];
- int skip_mode[4] = {0, 0, 0, 0};
- int have_subp_search = cpi->sf.half_pixel_search; /* In real-time mode,
- when Speed >= 15, no sub-pixel search. */
- int lfdone=0, gfdone=0, afdone=0;
-
- LOWER_RES_INFO* store_mode_info
- = (LOWER_RES_INFO*)cpi->oxcf.mr_low_res_mode_info;
- unsigned int parent_mb_index;
- //unsigned int parent_mb_index = map_640x480_to_320x240[mb_row][mb_col];
- int dissim;
- int parent_ref_frame;
- int_mv parent_ref_mv;
- MB_PREDICTION_MODE parent_mode;
-
- /* Consider different down_sampling_factor. */
- {
- /* TODO: Removed the loop that supports special down_sampling_factor
- * such as 2, 4, 8. Will revisit it if needed.
- * Should also try using a look-up table to see if it helps
- * performance. */
- int round = cpi->oxcf.mr_down_sampling_factor.num/2;
- int parent_mb_row, parent_mb_col;
-
- parent_mb_row = (mb_row*cpi->oxcf.mr_down_sampling_factor.den+round)
- /cpi->oxcf.mr_down_sampling_factor.num;
- parent_mb_col = (mb_col*cpi->oxcf.mr_down_sampling_factor.den+round)
- /cpi->oxcf.mr_down_sampling_factor.num;
- parent_mb_index = parent_mb_row*cpi->mr_low_res_mb_cols + parent_mb_col;
- }
-
- /* Read lower-resolution mode & motion result from memory.*/
- parent_ref_frame = store_mode_info[parent_mb_index].ref_frame;
- parent_mode = store_mode_info[parent_mb_index].mode;
- dissim = store_mode_info[parent_mb_index].dissim;
-
- /* For highest-resolution encoder, adjust dissim value. Lower its quality
- * for good performance. */
- if (cpi->oxcf.mr_encoder_id == (cpi->oxcf.mr_total_resolutions - 1))
- dissim>>=1;
-
- if(parent_ref_frame != INTRA_FRAME)
- {
- /* Consider different down_sampling_factor.
- * The result can be rounded to be more precise, but it takes more time.
- */
- //int round = cpi->oxcf.mr_down_sampling_factor.den/2;
- parent_ref_mv.as_mv.row = store_mode_info[parent_mb_index].mv.as_mv.row
- *cpi->oxcf.mr_down_sampling_factor.num
- /cpi->oxcf.mr_down_sampling_factor.den;
- parent_ref_mv.as_mv.col = store_mode_info[parent_mb_index].mv.as_mv.col
- *cpi->oxcf.mr_down_sampling_factor.num
- /cpi->oxcf.mr_down_sampling_factor.den;
-
- vp8_clamp_mv2(&parent_ref_mv, xd);
- }
-
- vpx_memset(mode_mv, 0, sizeof(mode_mv));
- vpx_memset(nearest_mv, 0, sizeof(nearest_mv));
- vpx_memset(near_mv, 0, sizeof(near_mv));
- vpx_memset(&best_mbmode, 0, sizeof(best_mbmode));
-
- cpi->mbs_tested_so_far++;
-
- *returnintra = INT_MAX;
- x->skip = 0;
-
- x->e_mbd.mode_info_context->mbmi.ref_frame = INTRA_FRAME;
-
- // if we encode a new mv this is important
- // find the best new motion vector
- for (mode_index = 0; mode_index < MAX_MODES; mode_index++)
- {
- int frame_cost;
- int this_rd = INT_MAX;
-
- if (best_rd <= cpi->rd_threshes[mode_index])
- continue;
-
- /* If parent MB is intra, child MB is intra. */
- if (!parent_ref_frame && vp8_ref_frame_order[mode_index])
- continue;
-
- /* If parent MB is inter, and it is unlikely there are multiple objects
- * in parent MB, we use parent ref frame as child MB's ref frame. */
- if (parent_ref_frame && dissim < 8
- && parent_ref_frame != vp8_ref_frame_order[mode_index])
- continue;
-
- x->e_mbd.mode_info_context->mbmi.ref_frame = vp8_ref_frame_order[mode_index];
-
- if(x->e_mbd.mode_info_context->mbmi.ref_frame)
- {
- if(x->e_mbd.mode_info_context->mbmi.ref_frame==LAST_FRAME && !lfdone)
- {
- // set up all the refframe dependent pointers.
- //if (x->e_mbd.mode_info_context->mbmi.ref_frame == LAST_FRAME
- //&& (cpi->ref_frame_flags & VP8_LAST_FLAG))
- if (cpi->ref_frame_flags & VP8_LAST_FLAG)
- {
- 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],
- &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;
- v_buffer[LAST_FRAME] = lst_yv12->v_buffer + recon_uvoffset;
- }
- else
- skip_mode[LAST_FRAME] = 1;
-
- lfdone = 1;
- }
-
- if(x->e_mbd.mode_info_context->mbmi.ref_frame==GOLDEN_FRAME && !gfdone)
- {
- //if (x->e_mbd.mode_info_context->mbmi.ref_frame == GOLDEN_FRAME
- //&& (cpi->ref_frame_flags & VP8_GOLD_FLAG))
- if (cpi->ref_frame_flags & VP8_GOLD_FLAG)
- {
- 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],
- &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;
- v_buffer[GOLDEN_FRAME] = gld_yv12->v_buffer + recon_uvoffset;
- }
- else
- skip_mode[GOLDEN_FRAME] = 1;
-
- gfdone = 1;
- }
-
- if(x->e_mbd.mode_info_context->mbmi.ref_frame==ALTREF_FRAME && !afdone)
- {
- //if (x->e_mbd.mode_info_context->mbmi.ref_frame == ALTREF_FRAME
- //&& (cpi->ref_frame_flags & VP8_ALT_FLAG && cpi->source_alt_ref_active))
- if (cpi->ref_frame_flags & VP8_ALT_FLAG && cpi->source_alt_ref_active)
- {
- 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],
- &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;
- v_buffer[ALTREF_FRAME] = alt_yv12->v_buffer + recon_uvoffset;
- }
- else
- skip_mode[ALTREF_FRAME] = 1;
-
- afdone = 1;
- }
-
- if (skip_mode[x->e_mbd.mode_info_context->mbmi.ref_frame])
- continue;
-
- x->e_mbd.pre.y_buffer = y_buffer[x->e_mbd.mode_info_context->mbmi.ref_frame];
- x->e_mbd.pre.u_buffer = u_buffer[x->e_mbd.mode_info_context->mbmi.ref_frame];
- 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_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));
-
- if (vp8_mode_order[mode_index] == NEARESTMV && mode_mv[NEARESTMV].as_int ==0)
- continue;
- if (vp8_mode_order[mode_index] == NEARMV && mode_mv[NEARMV].as_int ==0)
- continue;
-
- if (vp8_mode_order[mode_index] == NEWMV && parent_mode == ZEROMV
- && best_ref_mv.as_int==0) //&& dissim==0
- continue;
- else if(vp8_mode_order[mode_index] == NEWMV && dissim==0
- && best_ref_mv.as_int==parent_ref_mv.as_int)
- continue;
- }
-
- // Check to see if the testing frequency for this mode is at its max
- // If so then prevent it from being tested and increase the threshold for its testing
- if (cpi->mode_test_hit_counts[mode_index] && (cpi->mode_check_freq[mode_index] > 1))
- {
- //if ( (cpi->mbs_tested_so_far / cpi->mode_test_hit_counts[mode_index]) <= cpi->mode_check_freq[mode_index] )
- if (cpi->mbs_tested_so_far <= (cpi->mode_check_freq[mode_index] * cpi->mode_test_hit_counts[mode_index]))
- {
- // Increase the threshold for coding this mode to make it less likely to be chosen
- cpi->rd_thresh_mult[mode_index] += 4;
-
- if (cpi->rd_thresh_mult[mode_index] > MAX_THRESHMULT)
- cpi->rd_thresh_mult[mode_index] = MAX_THRESHMULT;
-
- cpi->rd_threshes[mode_index] = (cpi->rd_baseline_thresh[mode_index] >> 7) * cpi->rd_thresh_mult[mode_index];
-
- continue;
- }
- }
-
- // We have now reached the point where we are going to test the current
- //mode so increment the counter for the number of times it has been tested
- cpi->mode_test_hit_counts[mode_index] ++;
-
- rate2 = 0;
- distortion2 = 0;
-
- this_mode = vp8_mode_order[mode_index];
-
- x->e_mbd.mode_info_context->mbmi.mode = this_mode;
- x->e_mbd.mode_info_context->mbmi.uv_mode = DC_PRED;
-
- // Work out the cost assosciated with selecting the reference frame
- frame_cost =
- x->e_mbd.ref_frame_cost[x->e_mbd.mode_info_context->mbmi.ref_frame];
- rate2 += frame_cost;
-
- // Only consider ZEROMV/ALTREF_FRAME for alt ref frame,
- // unless ARNR filtering is enabled in which case we want
- // an unfiltered alternative
- if (cpi->is_src_frame_alt_ref && (cpi->oxcf.arnr_max_frames == 0))
- {
- if (this_mode != ZEROMV || x->e_mbd.mode_info_context->mbmi.ref_frame != ALTREF_FRAME)
- continue;
- }
-
- switch (this_mode)
- {
- case B_PRED:
- // Pass best so far to pick_intra4x4mby_modes to use as breakout
- distortion2 = best_sse;
- pick_intra4x4mby_modes(IF_RTCD(&cpi->rtcd), x, &rate, &distortion2);
-
- if (distortion2 == INT_MAX)
- {
- this_rd = INT_MAX;
- }
- else
- {
- rate2 += rate;
- distortion2 = VARIANCE_INVOKE
- (&cpi->rtcd.variance, var16x16)(
- *(b->base_src), b->src_stride,
- x->e_mbd.predictor, 16, &sse);
- this_rd = RDCOST(x->rdmult, x->rddiv, rate2, distortion2);
-
- if (this_rd < best_intra_rd)
- {
- best_intra_rd = this_rd;
- *returnintra = distortion2;
- }
- }
-
- break;
-
- case DC_PRED:
- case V_PRED:
- case H_PRED:
- case TM_PRED:
- RECON_INVOKE(&cpi->common.rtcd.recon, build_intra_predictors_mby)
- (&x->e_mbd);
- distortion2 = VARIANCE_INVOKE(&cpi->rtcd.variance, var16x16)
- (*(b->base_src), b->src_stride,
- x->e_mbd.predictor, 16, &sse);
- rate2 += x->mbmode_cost[x->e_mbd.frame_type][x->e_mbd.mode_info_context->mbmi.mode];
- this_rd = RDCOST(x->rdmult, x->rddiv, rate2, distortion2);
-
- if (this_rd < best_intra_rd)
- {
- best_intra_rd = this_rd;
- *returnintra = distortion2;
- }
- break;
-
- case NEWMV:
- {
- int thissme;
- int step_param;
- int further_steps;
- int n = 0;
- int sadpb = x->sadperbit16;
- int_mv mvp_full;
-
- int col_min = (best_ref_mv.as_mv.col>>3) - MAX_FULL_PEL_VAL
- + ((best_ref_mv.as_mv.col & 7)?1:0);
- int row_min = (best_ref_mv.as_mv.row>>3) - MAX_FULL_PEL_VAL
- + ((best_ref_mv.as_mv.row & 7)?1:0);
- int col_max = (best_ref_mv.as_mv.col>>3) + MAX_FULL_PEL_VAL;
- int row_max = (best_ref_mv.as_mv.row>>3) + MAX_FULL_PEL_VAL;
-
- 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;
-
- int speed_adjust = (cpi->Speed > 5) ? ((cpi->Speed >= 8)? 3 : 2) : 1;
- int diff_mv = MAX(abs(best_ref_mv.as_mv.row - parent_ref_mv.as_mv.row),
- abs(best_ref_mv.as_mv.col - parent_ref_mv.as_mv.col));
-
- // Further step/diamond searches as necessary
- step_param = cpi->sf.first_step + speed_adjust; //sf->first_step = 1; for -6 step_param =3;
-
- // Use parent MV as predictor. Adjust search range accordingly.
- mvp.as_int = parent_ref_mv.as_int;
- mvp_full.as_mv.col = parent_ref_mv.as_mv.col>>3;
- mvp_full.as_mv.row = parent_ref_mv.as_mv.row>>3;
-
- if(dissim <=32) step_param += 3;
- else if(dissim <=128) step_param += 2;
- else step_param += 1;
-
- if(dissim >2 || diff_mv >4)
- {
- /* 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_steps = (cpi->Speed >= 8)?
- 0: (cpi->sf.max_step_search_steps - 1 - step_param);
-
- if (cpi->sf.search_method == HEX)
- {
- bestsme = vp8_hex_search(x, b, d, &mvp_full, &d->bmi.mv, step_param,
- sadpb, &cpi->fn_ptr[BLOCK_16X16],
- x->mvsadcost, x->mvcost, &best_ref_mv);
- mode_mv[NEWMV].as_int = d->bmi.mv.as_int;
- }
- else
- {
- bestsme = cpi->diamond_search_sad(x, b, d, &mvp_full, &d->bmi.mv,
- step_param, sadpb, &num00,
- &cpi->fn_ptr[BLOCK_16X16],
- x->mvcost, &best_ref_mv);
- mode_mv[NEWMV].as_int = d->bmi.mv.as_int;
-
- // Further step/diamond searches as necessary
- n = 0;
- //further_steps = (cpi->sf.max_step_search_steps - 1) - step_param;
-
- n = num00;
- num00 = 0;
-
- while (n < further_steps)
- {
- n++;
-
- if (num00)
- num00--;
- else
- {
- thissme =
- cpi->diamond_search_sad(x, b, d, &mvp_full,
- &d->bmi.mv,
- step_param + n,
- sadpb, &num00,
- &cpi->fn_ptr[BLOCK_16X16],
- x->mvcost, &best_ref_mv);
- if (thissme < bestsme)
- {
- bestsme = thissme;
- mode_mv[NEWMV].as_int = d->bmi.mv.as_int;
- }
- else
- {
- d->bmi.mv.as_int = mode_mv[NEWMV].as_int;
- }
- }
- }
- }
-
- 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;
- }else
- {
- d->bmi.mv.as_int = mvp_full.as_int;
- mode_mv[NEWMV].as_int = mvp_full.as_int;
- }
-
- // This is not needed.
- //if (bestsme < INT_MAX)
- cpi->find_fractional_mv_step(x, b, d, &d->bmi.mv, &best_ref_mv,
- x->errorperbit,
- &cpi->fn_ptr[BLOCK_16X16],
- cpi->mb.mvcost,
- &distortion2,&sse);
-
- mode_mv[NEWMV].as_int = d->bmi.mv.as_int;
-
- // mv cost;
- rate2 += vp8_mv_bit_cost(&mode_mv[NEWMV], &best_ref_mv, cpi->mb.mvcost, 128);
- }
-
- case NEARESTMV:
- case NEARMV:
- // Trap vectors that reach beyond the UMV borders
- // Note that ALL New MV, Nearest MV Near MV and Zero MV code drops
- // through to this point because of the lack of break statements
- // in the previous two cases.
- if (((mode_mv[this_mode].as_mv.row >> 3) < x->mv_row_min) ||
- ((mode_mv[this_mode].as_mv.row >> 3) > x->mv_row_max) ||
- ((mode_mv[this_mode].as_mv.col >> 3) < x->mv_col_min) ||
- ((mode_mv[this_mode].as_mv.col >> 3) > x->mv_col_max))
- continue;
-
- case ZEROMV:
- rate2 += vp8_cost_mv_ref(this_mode, mdcounts);
- x->e_mbd.mode_info_context->mbmi.mv.as_int =
- mode_mv[this_mode].as_int;
-
- if((this_mode != NEWMV) ||
- !(have_subp_search) || cpi->common.full_pixel==1)
- distortion2 = get_inter_mbpred_error(x,
- &cpi->fn_ptr[BLOCK_16X16],
- &sse, mode_mv[this_mode]);
-
- this_rd = RDCOST(x->rdmult, x->rddiv, rate2, distortion2);
-
- if (cpi->active_map_enabled && x->active_ptr[0] == 0)
- {
- x->skip = 1;
- }
- else if (sse < x->encode_breakout)
- {
- // Check u and v to make sure skip is ok
- int sse2 = 0;
-
- sse2 = VP8_UVSSE(x, IF_RTCD(&cpi->rtcd.variance));
-
- if (sse2 * 2 < x->encode_breakout)
- x->skip = 1;
- else
- x->skip = 0;
- }
-
- break;
- default:
- break;
- }
-
- if (this_rd < best_rd || x->skip)
- {
- // Note index of best mode
- best_mode_index = mode_index;
-
- *returnrate = rate2;
- *returndistortion = distortion2;
- best_sse = sse;
- best_rd = this_rd;
- vpx_memcpy(&best_mbmode, &x->e_mbd.mode_info_context->mbmi, sizeof(MB_MODE_INFO));
-
- // Testing this mode gave rise to an improvement in best error score. Lower threshold a bit for next time
- cpi->rd_thresh_mult[mode_index] = (cpi->rd_thresh_mult[mode_index] >= (MIN_THRESHMULT + 2)) ? cpi->rd_thresh_mult[mode_index] - 2 : MIN_THRESHMULT;
- cpi->rd_threshes[mode_index] = (cpi->rd_baseline_thresh[mode_index] >> 7) * cpi->rd_thresh_mult[mode_index];
- }
- // If the mode did not help improve the best error case then raise the threshold for testing that mode next time around.
- else
- {
- cpi->rd_thresh_mult[mode_index] += 4;
-
- if (cpi->rd_thresh_mult[mode_index] > MAX_THRESHMULT)
- cpi->rd_thresh_mult[mode_index] = MAX_THRESHMULT;
-
- cpi->rd_threshes[mode_index] = (cpi->rd_baseline_thresh[mode_index] >> 7) * cpi->rd_thresh_mult[mode_index];
- }
-
- if (x->skip)
- break;
- }
-
- // Reduce the activation RD thresholds for the best choice mode
- if ((cpi->rd_baseline_thresh[best_mode_index] > 0) && (cpi->rd_baseline_thresh[best_mode_index] < (INT_MAX >> 2)))
- {
- int best_adjustment = (cpi->rd_thresh_mult[best_mode_index] >> 3);
-
- cpi->rd_thresh_mult[best_mode_index] = (cpi->rd_thresh_mult[best_mode_index] >= (MIN_THRESHMULT + best_adjustment)) ? cpi->rd_thresh_mult[best_mode_index] - best_adjustment : MIN_THRESHMULT;
- cpi->rd_threshes[best_mode_index] = (cpi->rd_baseline_thresh[best_mode_index] >> 7) * cpi->rd_thresh_mult[best_mode_index];
- }
-
-
- {
- int this_rdbin = (*returndistortion >> 7);
-
- if (this_rdbin >= 1024)
- {
- this_rdbin = 1023;
- }
-
- cpi->error_bins[this_rdbin] ++;
- }
-
- if (cpi->is_src_frame_alt_ref &&
- (best_mbmode.mode != ZEROMV || best_mbmode.ref_frame != ALTREF_FRAME))
- {
- x->e_mbd.mode_info_context->mbmi.mode = ZEROMV;
- x->e_mbd.mode_info_context->mbmi.ref_frame = ALTREF_FRAME;
- x->e_mbd.mode_info_context->mbmi.mv.as_int = 0;
- x->e_mbd.mode_info_context->mbmi.uv_mode = DC_PRED;
- x->e_mbd.mode_info_context->mbmi.mb_skip_coeff =
- (cpi->common.mb_no_coeff_skip) ? 1 : 0;
- x->e_mbd.mode_info_context->mbmi.partitioning = 0;
-
- return;
- }
-
- /* set to the best mb mode */
- vpx_memcpy(&x->e_mbd.mode_info_context->mbmi, &best_mbmode, sizeof(MB_MODE_INFO));
-
- if (best_mbmode.mode <= B_PRED)
- {
- /* set mode_info_context->mbmi.uv_mode */
- pick_intra_mbuv_mode(x);
- }
-
- update_mvcount(cpi, &x->e_mbd, &frame_best_ref_mv[xd->mode_info_context->mbmi.ref_frame]);
-}
-#endif
diff --git a/vp8/encoder/pickinter.h b/vp8/encoder/pickinter.h
index df6042f..3d83782 100644
--- a/vp8/encoder/pickinter.h
+++ b/vp8/encoder/pickinter.h
@@ -16,14 +16,8 @@
extern void vp8_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset,
int recon_uvoffset, int *returnrate,
- int *returndistortion, int *returnintra);
+ int *returndistortion, int *returnintra,
+ int mb_row, int mb_col);
extern void vp8_pick_intra_mode(VP8_COMP *cpi, MACROBLOCK *x, int *rate);
-#if CONFIG_MULTI_RES_ENCODING
-extern void vp8_mr_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x,
- int recon_yoffset, int recon_uvoffset,
- int *returnrate, int *returndistortion,
- int *returnintra, int mb_row, int mb_col);
-#endif
-
#endif
diff --git a/vp8/encoder/picklpf.c b/vp8/encoder/picklpf.c
index e97a560..2449ae5 100644
--- a/vp8/encoder/picklpf.c
+++ b/vp8/encoder/picklpf.c
@@ -186,7 +186,7 @@
best_err = calc_partial_ssl_err(sd, cm->frame_to_show,
IF_RTCD(&cpi->rtcd.variance));
- filt_val -= (1 + ((filt_val > 10) ? 1 : 0));
+ filt_val -= 1 + (filt_val > 10);
// Search lower filter levels
while (filt_val >= min_filter_level)
@@ -209,11 +209,11 @@
break;
// Adjust filter level
- filt_val -= (1 + ((filt_val > 10) ? 1 : 0));
+ filt_val -= 1 + (filt_val > 10);
}
// Search up (note that we have already done filt_val = cm->filter_level)
- filt_val = cm->filter_level + (1 + ((filt_val > 10) ? 1 : 0));
+ filt_val = cm->filter_level + 1 + (filt_val > 10);
if (best_filt_val == cm->filter_level)
{
@@ -243,7 +243,7 @@
break;
// Adjust filter level
- filt_val += (1 + ((filt_val > 10) ? 1 : 0));
+ filt_val += 1 + (filt_val > 10);
}
}
@@ -289,8 +289,12 @@
int Bias = 0; // Bias against raising loop filter and in favor of lowering it
+ int ss_err[MAX_LOOP_FILTER + 1];
+
YV12_BUFFER_CONFIG * saved_frame = cm->frame_to_show;
+ vpx_memset(ss_err, 0, sizeof(ss_err));
+
/* Replace unfiltered frame buffer with a new one */
cm->frame_to_show = &cpi->pick_lf_lvl_frame;
@@ -320,6 +324,9 @@
best_err = vp8_calc_ss_err(sd, cm->frame_to_show,
IF_RTCD(&cpi->rtcd.variance));
+
+ ss_err[filt_mid] = best_err;
+
filt_best = filt_mid;
while (filter_step > 0)
@@ -335,13 +342,19 @@
if ((filt_direction <= 0) && (filt_low != filt_mid))
{
- // Get Low filter error score
- vp8_yv12_copy_y_ptr(saved_frame, cm->frame_to_show);
- vp8cx_set_alt_lf_level(cpi, filt_low);
- vp8_loop_filter_frame_yonly(cm, &cpi->mb.e_mbd, filt_low);
+ if(ss_err[filt_low] == 0)
+ {
+ // Get Low filter error score
+ vp8_yv12_copy_y_ptr(saved_frame, cm->frame_to_show);
+ vp8cx_set_alt_lf_level(cpi, filt_low);
+ vp8_loop_filter_frame_yonly(cm, &cpi->mb.e_mbd, filt_low);
- filt_err = vp8_calc_ss_err(sd, cm->frame_to_show,
- IF_RTCD(&cpi->rtcd.variance));
+ filt_err = vp8_calc_ss_err(sd, cm->frame_to_show,
+ IF_RTCD(&cpi->rtcd.variance));
+ ss_err[filt_low] = filt_err;
+ }
+ else
+ filt_err = ss_err[filt_low];
// If value is close to the best so far then bias towards a lower loop filter value.
if ((filt_err - Bias) < best_err)
@@ -357,12 +370,18 @@
// Now look at filt_high
if ((filt_direction >= 0) && (filt_high != filt_mid))
{
- vp8_yv12_copy_y_ptr(saved_frame, cm->frame_to_show);
- vp8cx_set_alt_lf_level(cpi, filt_high);
- vp8_loop_filter_frame_yonly(cm, &cpi->mb.e_mbd, filt_high);
+ if(ss_err[filt_high] == 0)
+ {
+ vp8_yv12_copy_y_ptr(saved_frame, cm->frame_to_show);
+ vp8cx_set_alt_lf_level(cpi, filt_high);
+ vp8_loop_filter_frame_yonly(cm, &cpi->mb.e_mbd, filt_high);
- filt_err = vp8_calc_ss_err(sd, cm->frame_to_show,
- IF_RTCD(&cpi->rtcd.variance));
+ filt_err = vp8_calc_ss_err(sd, cm->frame_to_show,
+ IF_RTCD(&cpi->rtcd.variance));
+ ss_err[filt_high] = filt_err;
+ }
+ else
+ filt_err = ss_err[filt_high];
// Was it better than the previous best?
if (filt_err < (best_err - Bias))
diff --git a/vp8/encoder/quantize.c b/vp8/encoder/quantize.c
index 3ca8758..ce04212 100644
--- a/vp8/encoder/quantize.c
+++ b/vp8/encoder/quantize.c
@@ -436,7 +436,8 @@
int quant_val;
int Q;
- int zbin_boost[16] = {0, 0, 8, 10, 12, 14, 16, 20, 24, 28, 32, 36, 40, 44, 44, 44};
+ int zbin_boost[16] = {0, 0, 8, 10, 12, 14, 16, 20, 24, 28, 32, 36, 40, 44,
+ 44, 44};
for (Q = 0; Q < QINDEX_RANGE; Q++)
{
@@ -469,36 +470,61 @@
cpi->zrun_zbin_boost_uv[Q][0] = (quant_val * zbin_boost[0]) >> 7;
// all the ac values = ;
- for (i = 1; i < 16; i++)
+ quant_val = vp8_ac_yquant(Q);
+ cpi->Y1quant_fast[Q][1] = (1 << 16) / quant_val;
+ invert_quant(cpi->sf.improved_quant, cpi->Y1quant[Q] + 1,
+ cpi->Y1quant_shift[Q] + 1, quant_val);
+ cpi->Y1zbin[Q][1] = ((qzbin_factors[Q] * quant_val) + 64) >> 7;
+ cpi->Y1round[Q][1] = (qrounding_factors[Q] * quant_val) >> 7;
+ cpi->common.Y1dequant[Q][1] = quant_val;
+ cpi->zrun_zbin_boost_y1[Q][1] = (quant_val * zbin_boost[1]) >> 7;
+
+ quant_val = vp8_ac2quant(Q, cpi->common.y2ac_delta_q);
+ cpi->Y2quant_fast[Q][1] = (1 << 16) / quant_val;
+ invert_quant(cpi->sf.improved_quant, cpi->Y2quant[Q] + 1,
+ cpi->Y2quant_shift[Q] + 1, quant_val);
+ cpi->Y2zbin[Q][1] = ((qzbin_factors_y2[Q] * quant_val) + 64) >> 7;
+ cpi->Y2round[Q][1] = (qrounding_factors_y2[Q] * quant_val) >> 7;
+ cpi->common.Y2dequant[Q][1] = quant_val;
+ cpi->zrun_zbin_boost_y2[Q][1] = (quant_val * zbin_boost[1]) >> 7;
+
+ quant_val = vp8_ac_uv_quant(Q, cpi->common.uvac_delta_q);
+ cpi->UVquant_fast[Q][1] = (1 << 16) / quant_val;
+ invert_quant(cpi->sf.improved_quant, cpi->UVquant[Q] + 1,
+ cpi->UVquant_shift[Q] + 1, quant_val);
+ cpi->UVzbin[Q][1] = ((qzbin_factors[Q] * quant_val) + 64) >> 7;
+ cpi->UVround[Q][1] = (qrounding_factors[Q] * quant_val) >> 7;
+ cpi->common.UVdequant[Q][1] = quant_val;
+ cpi->zrun_zbin_boost_uv[Q][1] = (quant_val * zbin_boost[1]) >> 7;
+
+ for (i = 2; i < 16; i++)
{
- int rc = vp8_default_zig_zag1d[i];
+ cpi->Y1quant_fast[Q][i] = cpi->Y1quant_fast[Q][1];
+ cpi->Y1quant[Q][i] = cpi->Y1quant[Q][1];
+ cpi->Y1quant_shift[Q][i] = cpi->Y1quant_shift[Q][1];
+ cpi->Y1zbin[Q][i] = cpi->Y1zbin[Q][1];
+ cpi->Y1round[Q][i] = cpi->Y1round[Q][1];
+ cpi->common.Y1dequant[Q][i] = cpi->common.Y1dequant[Q][1];
+ cpi->zrun_zbin_boost_y1[Q][i] = (cpi->common.Y1dequant[Q][1] *
+ zbin_boost[i]) >> 7;
- quant_val = vp8_ac_yquant(Q);
- cpi->Y1quant_fast[Q][rc] = (1 << 16) / quant_val;
- invert_quant(cpi->sf.improved_quant, cpi->Y1quant[Q] + rc,
- cpi->Y1quant_shift[Q] + rc, quant_val);
- cpi->Y1zbin[Q][rc] = ((qzbin_factors[Q] * quant_val) + 64) >> 7;
- cpi->Y1round[Q][rc] = (qrounding_factors[Q] * quant_val) >> 7;
- cpi->common.Y1dequant[Q][rc] = quant_val;
- cpi->zrun_zbin_boost_y1[Q][i] = (quant_val * zbin_boost[i]) >> 7;
+ cpi->Y2quant_fast[Q][i] = cpi->Y2quant_fast[Q][1];
+ cpi->Y2quant[Q][i] = cpi->Y2quant[Q][1];
+ cpi->Y2quant_shift[Q][i] = cpi->Y2quant_shift[Q][1];
+ cpi->Y2zbin[Q][i] = cpi->Y2zbin[Q][1];
+ cpi->Y2round[Q][i] = cpi->Y2round[Q][1];
+ cpi->common.Y2dequant[Q][i] = cpi->common.Y2dequant[Q][1];
+ cpi->zrun_zbin_boost_y2[Q][i] = (cpi->common.Y2dequant[Q][1] *
+ zbin_boost[i]) >> 7;
- quant_val = vp8_ac2quant(Q, cpi->common.y2ac_delta_q);
- cpi->Y2quant_fast[Q][rc] = (1 << 16) / quant_val;
- invert_quant(cpi->sf.improved_quant, cpi->Y2quant[Q] + rc,
- cpi->Y2quant_shift[Q] + rc, quant_val);
- cpi->Y2zbin[Q][rc] = ((qzbin_factors_y2[Q] * quant_val) + 64) >> 7;
- cpi->Y2round[Q][rc] = (qrounding_factors_y2[Q] * quant_val) >> 7;
- cpi->common.Y2dequant[Q][rc] = quant_val;
- cpi->zrun_zbin_boost_y2[Q][i] = (quant_val * zbin_boost[i]) >> 7;
-
- quant_val = vp8_ac_uv_quant(Q, cpi->common.uvac_delta_q);
- cpi->UVquant_fast[Q][rc] = (1 << 16) / quant_val;
- invert_quant(cpi->sf.improved_quant, cpi->UVquant[Q] + rc,
- cpi->UVquant_shift[Q] + rc, quant_val);
- cpi->UVzbin[Q][rc] = ((qzbin_factors[Q] * quant_val) + 64) >> 7;
- cpi->UVround[Q][rc] = (qrounding_factors[Q] * quant_val) >> 7;
- cpi->common.UVdequant[Q][rc] = quant_val;
- cpi->zrun_zbin_boost_uv[Q][i] = (quant_val * zbin_boost[i]) >> 7;
+ cpi->UVquant_fast[Q][i] = cpi->UVquant_fast[Q][1];
+ cpi->UVquant[Q][i] = cpi->UVquant[Q][1];
+ cpi->UVquant_shift[Q][i] = cpi->UVquant_shift[Q][1];
+ cpi->UVzbin[Q][i] = cpi->UVzbin[Q][1];
+ cpi->UVround[Q][i] = cpi->UVround[Q][1];
+ cpi->common.UVdequant[Q][i] = cpi->common.UVdequant[Q][1];
+ cpi->zrun_zbin_boost_uv[Q][i] = (cpi->common.UVdequant[Q][1] *
+ zbin_boost[i]) >> 7;
}
}
}
diff --git a/vp8/encoder/ratectrl.c b/vp8/encoder/ratectrl.c
index db1a327..1c43c11 100644
--- a/vp8/encoder/ratectrl.c
+++ b/vp8/encoder/ratectrl.c
@@ -304,6 +304,8 @@
// Setup for Key frame:
vp8_default_coef_probs(& cpi->common);
+
+
vp8_kf_default_bmode_probs(cpi->common.kf_bmode_prob);
vpx_memcpy(cpi->common.fc.mvc, vp8_default_mv_context, sizeof(vp8_default_mv_context));
@@ -314,6 +316,12 @@
vpx_memset(cpi->common.fc.pre_mvc, 0, sizeof(cpi->common.fc.pre_mvc)); //initialize pre_mvc to all zero.
+ // Make sure we initialize separate contexts for altref,gold, and normal.
+ // TODO shouldn't need 3 different copies of structure to do this!
+ vpx_memcpy(&cpi->lfc_a, &cpi->common.fc, sizeof(cpi->common.fc));
+ vpx_memcpy(&cpi->lfc_g, &cpi->common.fc, sizeof(cpi->common.fc));
+ vpx_memcpy(&cpi->lfc_n, &cpi->common.fc, sizeof(cpi->common.fc));
+
//cpi->common.filter_level = 0; // Reset every key frame.
cpi->common.filter_level = cpi->common.base_qindex * 3 / 8 ;
@@ -324,8 +332,8 @@
else
cpi->frames_till_gf_update_due = cpi->goldfreq;
- cpi->common.refresh_golden_frame = TRUE;
- cpi->common.refresh_alt_ref_frame = TRUE;
+ cpi->common.refresh_golden_frame = 1;
+ cpi->common.refresh_alt_ref_frame = 1;
}
@@ -463,7 +471,7 @@
if (cpi->pass != 2)
{
// Single Pass lagged mode: TBD
- if (FALSE)
+ if (0)
{
}
@@ -590,14 +598,14 @@
if (cpi->pass != 2)
{
// For now Alt ref is not allowed except in 2 pass modes.
- cpi->source_alt_ref_pending = FALSE;
+ cpi->source_alt_ref_pending = 0;
/*if ( cpi->oxcf.fixed_q == -1)
{
if ( cpi->oxcf.play_alternate && (cpi->last_boost > (100 + (AF_THRESH*cpi->frames_till_gf_update_due)) ) )
- cpi->source_alt_ref_pending = TRUE;
+ cpi->source_alt_ref_pending = 1;
else
- cpi->source_alt_ref_pending = FALSE;
+ cpi->source_alt_ref_pending = 0;
}*/
}
}
@@ -932,6 +940,8 @@
if (cpi->active_worst_quality <= cpi->active_best_quality)
cpi->active_worst_quality = cpi->active_best_quality + 1;
+ if(cpi->active_worst_quality > 127)
+ cpi->active_worst_quality = 127;
}
// Unbuffered mode (eg. video conferencing)
else
@@ -972,7 +982,7 @@
#endif
//vpx_log("Decoder: Drop frame due to bandwidth: %d \n",cpi->buffer_level, cpi->av_per_frame_bandwidth);
- cpi->drop_frame = TRUE;
+ cpi->drop_frame = 1;
}
#if 0
@@ -980,7 +990,7 @@
else if ((cpi->buffer_level < cpi->oxcf.drop_frames_water_mark * cpi->oxcf.optimal_buffer_level / 100) &&
(cpi->drop_count < cpi->max_drop_count) && (cpi->pass == 0))
{
- cpi->drop_frame = TRUE;
+ cpi->drop_frame = 1;
}
#endif
@@ -1026,11 +1036,11 @@
{
// For one pass throw a GF if recent frame intra useage is low or the GF useage is high
if ((cpi->pass == 0) && (cpi->this_frame_percent_intra < 15 || gf_frame_useage >= 5))
- cpi->common.refresh_golden_frame = TRUE;
+ cpi->common.refresh_golden_frame = 1;
// Two pass GF descision
else if (cpi->pass == 2)
- cpi->common.refresh_golden_frame = TRUE;
+ cpi->common.refresh_golden_frame = 1;
}
#if 0
@@ -1048,7 +1058,7 @@
#endif
- if (cpi->common.refresh_golden_frame == TRUE)
+ if (cpi->common.refresh_golden_frame == 1)
{
#if 0
@@ -1533,7 +1543,7 @@
// Check if we're dropping the frame:
if (cpi->drop_frame)
{
- cpi->drop_frame = FALSE;
+ cpi->drop_frame = 0;
cpi->drop_count++;
return 0;
}
diff --git a/vp8/encoder/rdopt.c b/vp8/encoder/rdopt.c
index 62a2cfc..ce97961 100644
--- a/vp8/encoder/rdopt.c
+++ b/vp8/encoder/rdopt.c
@@ -28,7 +28,6 @@
#include "encodemb.h"
#include "quantize.h"
#include "vp8/common/idct.h"
-#include "vp8/common/g_common.h"
#include "variance.h"
#include "mcomp.h"
#include "rdopt.h"
@@ -288,12 +287,33 @@
}
}
- fill_token_costs(
- cpi->mb.token_costs,
- (const vp8_prob( *)[8][3][11]) cpi->common.fc.coef_probs
- );
+ {
+ // build token cost array for the type of frame we have now
+ FRAME_CONTEXT *l = &cpi->lfc_n;
- vp8_init_mode_costs(cpi);
+ if(cpi->common.refresh_alt_ref_frame)
+ l = &cpi->lfc_a;
+ else if(cpi->common.refresh_golden_frame)
+ l = &cpi->lfc_g;
+
+ fill_token_costs(
+ cpi->mb.token_costs,
+ (const vp8_prob( *)[8][3][11]) l->coef_probs
+ );
+ /*
+ fill_token_costs(
+ cpi->mb.token_costs,
+ (const vp8_prob( *)[8][3][11]) cpi->common.fc.coef_probs);
+ */
+
+
+ // TODO make these mode costs depend on last,alt or gold too. (jbb)
+ vp8_init_mode_costs(cpi);
+
+ // TODO figure onnnnuut why making mv cost frame type dependent didn't help (jbb)
+ //vp8_build_component_cost_table(cpi->mb.mvcost, (const MV_CONTEXT *) l->mvc, flags);
+
+ }
}
@@ -1364,8 +1384,8 @@
if (bsi.segment_rd < best_rd)
{
- int col_min = (best_ref_mv->as_mv.col>>3) - MAX_FULL_PEL_VAL + ((best_ref_mv->as_mv.col & 7)?1:0);
- int row_min = (best_ref_mv->as_mv.row>>3) - MAX_FULL_PEL_VAL + ((best_ref_mv->as_mv.row & 7)?1:0);
+ int col_min = ((best_ref_mv->as_mv.col+7)>>3) - MAX_FULL_PEL_VAL;
+ int row_min = ((best_ref_mv->as_mv.row+7)>>3) - MAX_FULL_PEL_VAL;
int col_max = (best_ref_mv->as_mv.col>>3) + MAX_FULL_PEL_VAL;
int row_max = (best_ref_mv->as_mv.row>>3) + MAX_FULL_PEL_VAL;
@@ -1695,7 +1715,9 @@
}
}
-void vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int recon_uvoffset, int *returnrate, int *returndistortion, int *returnintra)
+void 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];
BLOCKD *d = &x->e_mbd.block[0];
@@ -1722,27 +1744,20 @@
int distortion_uv;
int best_yrd = INT_MAX;
- //int all_rds[MAX_MODES]; // Experimental debug code.
- //int all_rates[MAX_MODES];
- //int all_dist[MAX_MODES];
- //int intermodecost[MAX_MODES];
-
MB_PREDICTION_MODE uv_intra_mode;
int_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)
- int_mv frame_nearest_mv[4];
- int_mv frame_near_mv[4];
- int_mv frame_best_ref_mv[4];
- int frame_mdcounts[4][4];
int frame_lf_or_gf[4];
unsigned char *y_buffer[4];
unsigned char *u_buffer[4];
unsigned char *v_buffer[4];
int ref_frame_map[4];
+ int sign_bias = 0;
+ vpx_memset(mode_mv, 0, sizeof(mode_mv));
vpx_memset(&best_mbmode, 0, sizeof(best_mbmode));
vpx_memset(&best_bmodes, 0, sizeof(best_bmodes));
@@ -1758,17 +1773,29 @@
for(; i<4; i++)
ref_frame_map[i] = -1;
+ /* Check to see if there is at least 1 valid reference frame that we need
+ * to calculate near_mvs.
+ */
+ if (ref_frame_map[1] > 0)
+ {
+ vp8_find_near_mvs(&x->e_mbd,
+ x->e_mbd.mode_info_context,
+ &mode_mv[NEARESTMV], &mode_mv[NEARMV],
+ &best_ref_mv,
+ mdcounts,
+ ref_frame_map[1],
+ cpi->common.ref_frame_sign_bias);
+
+ sign_bias = cpi->common.ref_frame_sign_bias[ref_frame_map[1]];
+ }
+
if (cpi->ref_frame_flags & VP8_LAST_FLAG)
{
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, &frame_nearest_mv[LAST_FRAME], &frame_near_mv[LAST_FRAME],
- &frame_best_ref_mv[LAST_FRAME], 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;
v_buffer[LAST_FRAME] = lst_yv12->v_buffer + recon_uvoffset;
-
frame_lf_or_gf[LAST_FRAME] = 0;
}
@@ -1776,13 +1803,9 @@
{
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, &frame_nearest_mv[GOLDEN_FRAME], &frame_near_mv[GOLDEN_FRAME],
- &frame_best_ref_mv[GOLDEN_FRAME], 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;
v_buffer[GOLDEN_FRAME] = gld_yv12->v_buffer + recon_uvoffset;
-
frame_lf_or_gf[GOLDEN_FRAME] = 1;
}
@@ -1790,13 +1813,9 @@
{
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, &frame_nearest_mv[ALTREF_FRAME], &frame_near_mv[ALTREF_FRAME],
- &frame_best_ref_mv[ALTREF_FRAME], 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;
v_buffer[ALTREF_FRAME] = alt_yv12->v_buffer + recon_uvoffset;
-
frame_lf_or_gf[ALTREF_FRAME] = 1;
}
@@ -1805,8 +1824,6 @@
x->skip = 0;
- vpx_memset(mode_mv, 0, sizeof(mode_mv));
-
x->e_mbd.mode_info_context->mbmi.ref_frame = INTRA_FRAME;
rd_pick_intra_mbuv_mode(cpi, x, &uv_intra_rate, &uv_intra_rate_tokenonly, &uv_intra_distortion);
uv_intra_mode = x->e_mbd.mode_info_context->mbmi.uv_mode;
@@ -1819,13 +1836,6 @@
int other_cost = 0;
int this_ref_frame = ref_frame_map[vp8_ref_frame_order[mode_index]];
- // Experimental debug code.
- // Record of rd values recorded for this MB. -1 indicates not measured
- //all_rds[mode_index] = -1;
- //all_rates[mode_index] = -1;
- //all_dist[mode_index] = -1;
- //intermodecost[mode_index] = -1;
-
// Test best rd so far against threshold for trying this mode.
if (best_rd <= cpi->rd_threshes[mode_index])
continue;
@@ -1858,10 +1868,20 @@
x->e_mbd.pre.y_buffer = y_buffer[x->e_mbd.mode_info_context->mbmi.ref_frame];
x->e_mbd.pre.u_buffer = u_buffer[x->e_mbd.mode_info_context->mbmi.ref_frame];
x->e_mbd.pre.v_buffer = v_buffer[x->e_mbd.mode_info_context->mbmi.ref_frame];
- mode_mv[NEARESTMV] = frame_nearest_mv[x->e_mbd.mode_info_context->mbmi.ref_frame];
- mode_mv[NEARMV] = frame_near_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];
- vpx_memcpy(mdcounts, frame_mdcounts[x->e_mbd.mode_info_context->mbmi.ref_frame], sizeof(mdcounts));
+
+ if (sign_bias !=
+ cpi->common.ref_frame_sign_bias[x->e_mbd.mode_info_context->mbmi.ref_frame])
+ {
+ mode_mv[NEARESTMV].as_mv.row *= -1;
+ mode_mv[NEARESTMV].as_mv.col *= -1;
+ mode_mv[NEARMV].as_mv.row *= -1;
+ mode_mv[NEARMV].as_mv.col *= -1;
+ best_ref_mv.as_mv.row *= -1;
+ best_ref_mv.as_mv.col *= -1;
+ sign_bias
+ = cpi->common.ref_frame_sign_bias[x->e_mbd.mode_info_context->mbmi.ref_frame];
+ }
+
lf_or_gf = frame_lf_or_gf[x->e_mbd.mode_info_context->mbmi.ref_frame];
}
@@ -1995,8 +2015,8 @@
int sadpb = x->sadperbit16;
int_mv mvp_full;
- int col_min = (best_ref_mv.as_mv.col>>3) - MAX_FULL_PEL_VAL + ((best_ref_mv.as_mv.col & 7)?1:0);
- int row_min = (best_ref_mv.as_mv.row>>3) - MAX_FULL_PEL_VAL + ((best_ref_mv.as_mv.row & 7)?1:0);
+ int col_min = ((best_ref_mv.as_mv.col+7)>>3) - MAX_FULL_PEL_VAL;
+ int row_min = ((best_ref_mv.as_mv.row+7)>>3) - MAX_FULL_PEL_VAL;
int col_max = (best_ref_mv.as_mv.col>>3) + MAX_FULL_PEL_VAL;
int row_max = (best_ref_mv.as_mv.row>>3) + MAX_FULL_PEL_VAL;
@@ -2145,7 +2165,7 @@
continue;
vp8_set_mbmode_and_mvs(x, this_mode, &mode_mv[this_mode]);
- vp8_build_inter16x16_predictors_mby(&x->e_mbd);
+ vp8_build_inter16x16_predictors_mby(&x->e_mbd, x->e_mbd.predictor, 16);
if (cpi->active_map_enabled && x->active_ptr[0] == 0) {
x->skip = 1;
@@ -2265,11 +2285,6 @@
this_rd = RDCOST(x->rdmult, x->rddiv, rate2, distortion2);
}
- // Experimental debug code.
- //all_rds[mode_index] = this_rd;
- //all_rates[mode_index] = rate2;
- //all_dist[mode_index] = distortion2;
-
// Keep record of best intra distortion
if ((x->e_mbd.mode_info_context->mbmi.ref_frame == INTRA_FRAME) &&
(this_rd < best_intra_rd) )
@@ -2370,7 +2385,7 @@
x->e_mbd.mode_info_context->mbmi.mv.as_int = 0;
x->e_mbd.mode_info_context->mbmi.uv_mode = DC_PRED;
x->e_mbd.mode_info_context->mbmi.mb_skip_coeff =
- (cpi->common.mb_no_coeff_skip) ? 1 : 0;
+ (cpi->common.mb_no_coeff_skip);
x->e_mbd.mode_info_context->mbmi.partitioning = 0;
return;
@@ -2397,10 +2412,14 @@
x->partition_info->bmi[15].mv.as_int;
}
- rd_update_mvcount(cpi, x, &frame_best_ref_mv[xd->mode_info_context->mbmi.ref_frame]);
+ if (sign_bias
+ != cpi->common.ref_frame_sign_bias[xd->mode_info_context->mbmi.ref_frame])
+ {
+ best_ref_mv.as_mv.row *= -1;
+ best_ref_mv.as_mv.col *= -1;
+ }
-
-
+ rd_update_mvcount(cpi, x, &best_ref_mv);
}
void vp8_rd_pick_intra_mode(VP8_COMP *cpi, MACROBLOCK *x, int *rate_)
diff --git a/vp8/encoder/temporal_filter.c b/vp8/encoder/temporal_filter.c
index b9ade1c..545e4f2 100644
--- a/vp8/encoder/temporal_filter.c
+++ b/vp8/encoder/temporal_filter.c
@@ -22,7 +22,6 @@
#include "ratectrl.h"
#include "vp8/common/quant_common.h"
#include "segmentation.h"
-#include "vp8/common/g_common.h"
#include "vpx_scale/yv12extend.h"
#include "vpx_mem/vpx_mem.h"
#include "vp8/common/swapyv12buffer.h"
@@ -98,7 +97,7 @@
unsigned short *count
)
{
- int i, j, k;
+ unsigned int i, j, k;
int modifier;
int byte = 0;
@@ -186,7 +185,7 @@
if (cpi->Speed < 8)
{
step_param = cpi->sf.first_step +
- ((cpi->Speed > 5) ? 1 : 0);
+ (cpi->Speed > 5);
further_steps =
(cpi->sf.max_step_search_steps - 1)-step_param;
}
diff --git a/vp8/encoder/tokenize.c b/vp8/encoder/tokenize.c
index e819485..8bfc47f 100644
--- a/vp8/encoder/tokenize.c
+++ b/vp8/encoder/tokenize.c
@@ -514,17 +514,19 @@
TOKENEXTRA **tp,
ENTROPY_CONTEXT *a,
ENTROPY_CONTEXT *l,
+ int type,
VP8_COMP *cpi
)
{
int pt; /* near block/prev token context index */
+ int band;
TOKENEXTRA *t = *tp; /* store tokens starting here */
VP8_COMBINEENTROPYCONTEXTS(pt, *a, *l);
-
+ band = type ? 0 : 1;
t->Token = DCT_EOB_TOKEN;
- t->context_tree = cpi->common.fc.coef_probs [0] [1] [pt];
+ t->context_tree = cpi->common.fc.coef_probs [type] [band] [pt];
t->skip_eob_node = 0;
- ++cpi->coef_counts [0] [1] [pt] [DCT_EOB_TOKEN];
+ ++cpi->coef_counts [type] [band] [pt] [DCT_EOB_TOKEN];
++t;
*tp = t;
pt = 0; /* 0 <-> all coeff data is zero */
@@ -561,15 +563,19 @@
ENTROPY_CONTEXT * L = (ENTROPY_CONTEXT *)x->left_context;
int plane_type;
int b;
-
- stuff2nd_order_b(t,
+ plane_type = 3;
+ if((x->mode_info_context->mbmi.mode != B_PRED
+ && x->mode_info_context->mbmi.mode != SPLITMV))
+ {
+ stuff2nd_order_b(t,
A + vp8_block2above[24], L + vp8_block2left[24], cpi);
- plane_type = 0;
+ plane_type = 0;
+ }
for (b = 0; b < 16; b++)
stuff1st_order_b(t,
A + vp8_block2above[b],
- L + vp8_block2left[b], cpi);
+ L + vp8_block2left[b], plane_type, cpi);
for (b = 16; b < 24; b++)
stuff1st_order_buv(t,
diff --git a/vp8/vp8_common.mk b/vp8/vp8_common.mk
index 683af34..0e56432 100644
--- a/vp8/vp8_common.mk
+++ b/vp8/vp8_common.mk
@@ -9,7 +9,6 @@
##
VP8_COMMON_SRCS-yes += vp8_common.mk
-VP8_COMMON_SRCS-yes += common/type_aliases.h
VP8_COMMON_SRCS-yes += common/pragmas.h
VP8_COMMON_SRCS-yes += common/ppflags.h
VP8_COMMON_SRCS-yes += common/onyx.h
@@ -20,6 +19,8 @@
VP8_COMMON_SRCS-yes += common/coefupdateprobs.h
VP8_COMMON_SRCS-yes += common/debugmodes.c
VP8_COMMON_SRCS-yes += common/default_coef_probs.h
+VP8_COMMON_SRCS-yes += common/dequantize.c
+VP8_COMMON_SRCS-yes += common/dequantize.h
VP8_COMMON_SRCS-yes += common/entropy.c
VP8_COMMON_SRCS-yes += common/entropymode.c
VP8_COMMON_SRCS-yes += common/entropymv.c
@@ -28,17 +29,16 @@
VP8_COMMON_SRCS-yes += common/filter.h
VP8_COMMON_SRCS-yes += common/findnearmv.c
VP8_COMMON_SRCS-yes += common/generic/systemdependent.c
+VP8_COMMON_SRCS-yes += common/idct_blk.c
VP8_COMMON_SRCS-yes += common/idctllm.c
VP8_COMMON_SRCS-yes += common/alloccommon.h
VP8_COMMON_SRCS-yes += common/blockd.h
VP8_COMMON_SRCS-yes += common/common.h
-VP8_COMMON_SRCS-yes += common/common_types.h
VP8_COMMON_SRCS-yes += common/entropy.h
VP8_COMMON_SRCS-yes += common/entropymode.h
VP8_COMMON_SRCS-yes += common/entropymv.h
VP8_COMMON_SRCS-yes += common/extend.h
VP8_COMMON_SRCS-yes += common/findnearmv.h
-VP8_COMMON_SRCS-yes += common/g_common.h
VP8_COMMON_SRCS-yes += common/header.h
VP8_COMMON_SRCS-yes += common/idct.h
VP8_COMMON_SRCS-yes += common/invtrans.h
@@ -57,7 +57,6 @@
VP8_COMMON_SRCS-yes += common/systemdependent.h
VP8_COMMON_SRCS-yes += common/threading.h
VP8_COMMON_SRCS-yes += common/treecoder.h
-VP8_COMMON_SRCS-yes += common/invtrans.c
VP8_COMMON_SRCS-yes += common/loopfilter.c
VP8_COMMON_SRCS-yes += common/loopfilter_filters.c
VP8_COMMON_SRCS-yes += common/mbpitch.c
@@ -69,9 +68,13 @@
VP8_COMMON_SRCS-yes += common/reconintra4x4.c
VP8_COMMON_SRCS-yes += common/setupintrarecon.c
VP8_COMMON_SRCS-yes += common/swapyv12buffer.c
+
+
+
VP8_COMMON_SRCS-$(CONFIG_POSTPROC_VISUALIZER) += common/textblit.c
VP8_COMMON_SRCS-yes += common/treecoder.c
+VP8_COMMON_SRCS-$(ARCH_X86)$(ARCH_X86_64) += common/x86/dequantize_x86.h
VP8_COMMON_SRCS-$(ARCH_X86)$(ARCH_X86_64) += common/x86/filter_x86.c
VP8_COMMON_SRCS-$(ARCH_X86)$(ARCH_X86_64) += common/x86/filter_x86.h
VP8_COMMON_SRCS-$(ARCH_X86)$(ARCH_X86_64) += common/x86/idct_x86.h
@@ -84,11 +87,14 @@
VP8_COMMON_SRCS-$(ARCH_X86)$(ARCH_X86_64) += common/x86/loopfilter_x86.c
VP8_COMMON_SRCS-$(CONFIG_POSTPROC) += common/postproc.h
VP8_COMMON_SRCS-$(CONFIG_POSTPROC) += common/postproc.c
+VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/dequantize_mmx.asm
+VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/idct_blk_mmx.c
VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/idctllm_mmx.asm
VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/iwalsh_mmx.asm
VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/recon_mmx.asm
VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/subpixel_mmx.asm
VP8_COMMON_SRCS-$(HAVE_MMX) += common/x86/loopfilter_mmx.asm
+VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/idct_blk_sse2.c
VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/idctllm_sse2.asm
VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/recon_sse2.asm
VP8_COMMON_SRCS-$(HAVE_SSE2) += common/x86/recon_wrapper_sse2.c
@@ -106,8 +112,6 @@
# common (c)
VP8_COMMON_SRCS-$(ARCH_ARM) += common/arm/arm_systemdependent.c
-VP8_COMMON_SRCS-$(ARCH_ARM) += common/arm/bilinearfilter_arm.c
-VP8_COMMON_SRCS-$(ARCH_ARM) += common/arm/bilinearfilter_arm.h
VP8_COMMON_SRCS-$(ARCH_ARM) += common/arm/filter_arm.c
VP8_COMMON_SRCS-$(ARCH_ARM) += common/arm/idct_arm.h
VP8_COMMON_SRCS-$(ARCH_ARM) += common/arm/loopfilter_arm.c
@@ -115,8 +119,12 @@
VP8_COMMON_SRCS-$(ARCH_ARM) += common/arm/recon_arm.h
VP8_COMMON_SRCS-$(ARCH_ARM) += common/arm/reconintra_arm.c
VP8_COMMON_SRCS-$(ARCH_ARM) += common/arm/subpixel_arm.h
+VP8_COMMON_SRCS-$(ARCH_ARM) += common/arm/dequantize_arm.c
+VP8_COMMON_SRCS-$(ARCH_ARM) += common/arm/dequantize_arm.h
# common (armv6)
+VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/bilinearfilter_arm.c
+VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/bilinearfilter_arm.h
VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/bilinearfilter_v6$(ASM)
VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/copymem8x4_v6$(ASM)
VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/copymem8x8_v6$(ASM)
@@ -129,6 +137,9 @@
VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/simpleloopfilter_v6$(ASM)
VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/sixtappredict8x4_v6$(ASM)
VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/intra4x4_predict_v6$(ASM)
+VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/dequant_idct_v6$(ASM)
+VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/dequantize_v6$(ASM)
+VP8_COMMON_SRCS-$(HAVE_ARMV6) += common/arm/armv6/idct_blk_v6.c
# common (neon)
VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/bilinearpredict4x4_neon$(ASM)
@@ -151,3 +162,8 @@
VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/sixtappredict16x16_neon$(ASM)
VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/buildintrapredictorsmby_neon$(ASM)
VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/save_neon_reg$(ASM)
+VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/dequant_idct_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/idct_dequant_full_2x_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/idct_dequant_0_2x_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/dequantizeb_neon$(ASM)
+VP8_COMMON_SRCS-$(HAVE_ARMV7) += common/arm/neon/idct_blk_neon.c
diff --git a/vp8/vp8_cx_iface.c b/vp8/vp8_cx_iface.c
index 5bb6b40..6181ee8 100644
--- a/vp8/vp8_cx_iface.c
+++ b/vp8/vp8_cx_iface.c
@@ -83,7 +83,7 @@
vpx_codec_enc_cfg_t cfg;
struct vp8_extracfg vp8_cfg;
VP8_CONFIG oxcf;
- VP8_PTR cpi;
+ struct VP8_COMP *cpi;
unsigned char *cx_data;
unsigned int cx_data_sz;
vpx_image_t preview_img;
@@ -137,7 +137,8 @@
static vpx_codec_err_t validate_config(vpx_codec_alg_priv_t *ctx,
const vpx_codec_enc_cfg_t *cfg,
- const struct vp8_extracfg *vp8_cfg)
+ const struct vp8_extracfg *vp8_cfg,
+ int finalize)
{
RANGE_CHECK(cfg, g_w, 1, 16383); /* 14 bits available */
RANGE_CHECK(cfg, g_h, 1, 16383); /* 14 bits available */
@@ -193,6 +194,9 @@
RANGE_CHECK_HI(vp8_cfg, arnr_strength, 6);
RANGE_CHECK(vp8_cfg, arnr_type, 1, 3);
RANGE_CHECK(vp8_cfg, cq_level, 0, 63);
+ if(finalize && cfg->rc_end_usage == VPX_CQ)
+ RANGE_CHECK(vp8_cfg, cq_level,
+ cfg->rc_min_quantizer, cfg->rc_max_quantizer);
#if !(CONFIG_REALTIME_ONLY)
if (cfg->g_pass == VPX_RC_LAST_PASS)
@@ -357,6 +361,10 @@
}
#if CONFIG_MULTI_RES_ENCODING
+ /* When mr_cfg is NULL, oxcf->mr_total_resolutions and oxcf->mr_encoder_id
+ * are both memset to 0, which ensures the correct logic under this
+ * situation.
+ */
if(mr_cfg)
{
oxcf->mr_total_resolutions = mr_cfg->mr_total_resolutions;
@@ -439,7 +447,7 @@
if ((cfg->g_lag_in_frames > ctx->cfg.g_lag_in_frames))
ERROR("Cannot increase lag_in_frames");
- res = validate_config(ctx, cfg, &ctx->vp8_cfg);
+ res = validate_config(ctx, cfg, &ctx->vp8_cfg, 0);
if (!res)
{
@@ -505,7 +513,7 @@
}
- res = validate_config(ctx, &ctx->cfg, &xcfg);
+ res = validate_config(ctx, &ctx->cfg, &xcfg, 0);
if (!res)
{
@@ -548,7 +556,7 @@
vpx_codec_enc_cfg_t *cfg;
unsigned int i;
- VP8_PTR optr;
+ struct VP8_COMP *optr;
if (!ctx->priv)
{
@@ -602,7 +610,7 @@
vp8_initialize();
- res = validate_config(priv, &priv->cfg, &priv->vp8_cfg);
+ res = validate_config(priv, &priv->cfg, &priv->vp8_cfg, 0);
if (!res)
{
@@ -732,6 +740,9 @@
if (img)
res = validate_img(ctx, img);
+ if (!res)
+ res = validate_config(ctx, &ctx->cfg, &ctx->vp8_cfg, 1);
+
pick_quickcompress_mode(ctx, duration, deadline);
vpx_codec_pkt_list_init(&ctx->pkt_list);
diff --git a/vp8/vp8_dx_iface.c b/vp8/vp8_dx_iface.c
index 54bdb85..fbe5817 100644
--- a/vp8/vp8_dx_iface.c
+++ b/vp8/vp8_dx_iface.c
@@ -57,7 +57,7 @@
vp8_stream_info_t si;
int defer_alloc;
int decoder_init;
- VP8D_PTR pbi;
+ struct VP8D_COMP *pbi;
int postproc_cfg_set;
vp8_postproc_cfg_t postproc_cfg;
#if CONFIG_POSTPROC_VISUALIZER
@@ -389,7 +389,7 @@
if (!res)
{
VP8D_CONFIG oxcf;
- VP8D_PTR optr;
+ struct VP8D_COMP* optr;
vp8dx_initialize();
diff --git a/vp8/vp8dx.mk b/vp8/vp8dx.mk
index d88b595..d6dc153 100644
--- a/vp8/vp8dx.mk
+++ b/vp8/vp8dx.mk
@@ -52,7 +52,6 @@
VP8_DX_SRCS-yes += decoder/dboolhuff.c
VP8_DX_SRCS-yes += decoder/decodemv.c
VP8_DX_SRCS-yes += decoder/decodframe.c
-VP8_DX_SRCS-yes += decoder/dequantize.c
VP8_DX_SRCS-yes += decoder/detokenize.c
VP8_DX_SRCS-$(CONFIG_ERROR_CONCEALMENT) += decoder/ec_types.h
VP8_DX_SRCS-$(CONFIG_ERROR_CONCEALMENT) += decoder/error_concealment.h
@@ -61,20 +60,14 @@
VP8_DX_SRCS-yes += decoder/dboolhuff.h
VP8_DX_SRCS-yes += decoder/decodemv.h
VP8_DX_SRCS-yes += decoder/decoderthreading.h
-VP8_DX_SRCS-yes += decoder/dequantize.h
VP8_DX_SRCS-yes += decoder/detokenize.h
VP8_DX_SRCS-yes += decoder/onyxd_int.h
VP8_DX_SRCS-yes += decoder/treereader.h
VP8_DX_SRCS-yes += decoder/onyxd_if.c
VP8_DX_SRCS-$(CONFIG_MULTITHREAD) += decoder/threading.c
-VP8_DX_SRCS-yes += decoder/idct_blk.c
VP8_DX_SRCS-$(CONFIG_MULTITHREAD) += decoder/reconintra_mt.h
VP8_DX_SRCS-$(CONFIG_MULTITHREAD) += decoder/reconintra_mt.c
VP8_DX_SRCS-yes := $(filter-out $(VP8_DX_SRCS_REMOVE-yes),$(VP8_DX_SRCS-yes))
-VP8_DX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += decoder/x86/dequantize_x86.h
VP8_DX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += decoder/x86/x86_dsystemdependent.c
-VP8_DX_SRCS-$(HAVE_MMX) += decoder/x86/dequantize_mmx.asm
-VP8_DX_SRCS-$(HAVE_MMX) += decoder/x86/idct_blk_mmx.c
-VP8_DX_SRCS-$(HAVE_SSE2) += decoder/x86/idct_blk_sse2.c
diff --git a/vp8/vp8dx_arm.mk b/vp8/vp8dx_arm.mk
index b08f946..fa1aaea 100644
--- a/vp8/vp8dx_arm.mk
+++ b/vp8/vp8dx_arm.mk
@@ -12,17 +12,3 @@
#VP8_DX_SRCS list is modified according to different platforms.
VP8_DX_SRCS-$(ARCH_ARM) += decoder/arm/arm_dsystemdependent.c
-VP8_DX_SRCS-$(ARCH_ARM) += decoder/arm/dequantize_arm.c
-VP8_DX_SRCS-$(ARCH_ARM) += decoder/arm/dequantize_arm.h
-
-#File list for armv6
-VP8_DX_SRCS-$(HAVE_ARMV6) += decoder/arm/armv6/dequant_idct_v6$(ASM)
-VP8_DX_SRCS-$(HAVE_ARMV6) += decoder/arm/armv6/dequantize_v6$(ASM)
-VP8_DX_SRCS-$(HAVE_ARMV6) += decoder/arm/armv6/idct_blk_v6.c
-
-#File list for neon
-VP8_DX_SRCS-$(HAVE_ARMV7) += decoder/arm/neon/dequant_idct_neon$(ASM)
-VP8_DX_SRCS-$(HAVE_ARMV7) += decoder/arm/neon/idct_dequant_full_2x_neon$(ASM)
-VP8_DX_SRCS-$(HAVE_ARMV7) += decoder/arm/neon/idct_dequant_0_2x_neon$(ASM)
-VP8_DX_SRCS-$(HAVE_ARMV7) += decoder/arm/neon/dequantizeb_neon$(ASM)
-VP8_DX_SRCS-$(HAVE_ARMV7) += decoder/arm/neon/idct_blk_neon.c
diff --git a/vp8_multi_resolution_encoder.c b/vp8_multi_resolution_encoder.c
index 732f96e..11c33d6 100644
--- a/vp8_multi_resolution_encoder.c
+++ b/vp8_multi_resolution_encoder.c
@@ -78,6 +78,8 @@
exit(EXIT_FAILURE);
}
+int (*read_frame_p)(FILE *f, vpx_image_t *img);
+
static int read_frame(FILE *f, vpx_image_t *img) {
size_t nbytes, to_read;
int res = 1;
@@ -92,6 +94,55 @@
return res;
}
+static int read_frame_by_row(FILE *f, vpx_image_t *img) {
+ size_t nbytes, to_read;
+ int res = 1;
+ int plane;
+
+ for (plane = 0; plane < 3; plane++)
+ {
+ unsigned char *ptr;
+ int w = (plane ? (1 + img->d_w) / 2 : img->d_w);
+ int h = (plane ? (1 + img->d_h) / 2 : img->d_h);
+ int r;
+
+ /* Determine the correct plane based on the image format. The for-loop
+ * always counts in Y,U,V order, but this may not match the order of
+ * the data on disk.
+ */
+ switch (plane)
+ {
+ case 1:
+ ptr = img->planes[img->fmt==VPX_IMG_FMT_YV12? VPX_PLANE_V : VPX_PLANE_U];
+ break;
+ case 2:
+ ptr = img->planes[img->fmt==VPX_IMG_FMT_YV12?VPX_PLANE_U : VPX_PLANE_V];
+ break;
+ default:
+ ptr = img->planes[plane];
+ }
+
+ for (r = 0; r < h; r++)
+ {
+ to_read = w;
+
+ nbytes = fread(ptr, 1, to_read, f);
+ if(nbytes != to_read) {
+ res = 0;
+ if(nbytes > 0)
+ printf("Warning: Read partial frame. Check your width & height!\n");
+ break;
+ }
+
+ ptr += img->stride[plane];
+ }
+ if (!res)
+ break;
+ }
+
+ return res;
+}
+
static void write_ivf_file_header(FILE *outfile,
const vpx_codec_enc_cfg_t *cfg,
int frame_cnt) {
@@ -262,9 +313,14 @@
/* Allocate image for each encoder */
for (i=0; i< NUM_ENCODERS; i++)
- if(!vpx_img_alloc(&raw[i], VPX_IMG_FMT_I420, cfg[i].g_w, cfg[i].g_h, 1))
+ if(!vpx_img_alloc(&raw[i], VPX_IMG_FMT_I420, cfg[i].g_w, cfg[i].g_h, 32))
die("Failed to allocate image", cfg[i].g_w, cfg[i].g_h);
+ if (raw[0].stride[VPX_PLANE_Y] == raw[0].d_w)
+ read_frame_p = read_frame;
+ else
+ read_frame_p = read_frame_by_row;
+
for (i=0; i< NUM_ENCODERS; i++)
write_ivf_file_header(outfile[i], &cfg[i], 0);
@@ -305,35 +361,22 @@
const vpx_codec_cx_pkt_t *pkt[NUM_ENCODERS];
flags = 0;
- frame_avail = read_frame(infile, &raw[0]);
+ frame_avail = read_frame_p(infile, &raw[0]);
- for ( i=1; i<NUM_ENCODERS; i++)
+ if(frame_avail)
{
- if(frame_avail)
+ for ( i=1; i<NUM_ENCODERS; i++)
{
/*Scale the image down a number of times by downsampling factor*/
- int src_uvwidth = (raw[i-1].d_w + 1) >> 1;
- int src_uvheight = (raw[i-1].d_h + 1) >> 1;
- const unsigned char* src_y = raw[i-1].planes[VPX_PLANE_Y];
- const unsigned char* src_u = raw[i-1].planes[VPX_PLANE_Y]
- + raw[i-1].d_w*raw[i-1].d_h;
- const unsigned char* src_v = raw[i-1].planes[VPX_PLANE_Y]
- + raw[i-1].d_w*raw[i-1].d_h
- + src_uvwidth*src_uvheight;
- int dst_uvwidth = (raw[i].d_w + 1) >> 1;
- int dst_uvheight = (raw[i].d_h + 1) >> 1;
- unsigned char* dst_y = raw[i].planes[VPX_PLANE_Y];
- unsigned char* dst_u = raw[i].planes[VPX_PLANE_Y]
- + raw[i].d_w*raw[i].d_h;
- unsigned char* dst_v = raw[i].planes[VPX_PLANE_Y]
- + raw[i].d_w*raw[i].d_h
- + dst_uvwidth*dst_uvheight;
-
/* FilterMode 1 or 2 give better psnr than FilterMode 0. */
- I420Scale(src_y, raw[i-1].d_w, src_u, src_uvwidth, src_v,
- src_uvwidth, raw[i-1].d_w, raw[i-1].d_h,
- dst_y, raw[i].d_w, dst_u, dst_uvwidth,
- dst_v, dst_uvwidth, raw[i].d_w, raw[i].d_h, 1);
+ I420Scale(raw[i-1].planes[VPX_PLANE_Y], raw[i-1].stride[VPX_PLANE_Y],
+ raw[i-1].planes[VPX_PLANE_U], raw[i-1].stride[VPX_PLANE_U],
+ raw[i-1].planes[VPX_PLANE_V], raw[i-1].stride[VPX_PLANE_V],
+ raw[i-1].d_w, raw[i-1].d_h,
+ raw[i].planes[VPX_PLANE_Y], raw[i].stride[VPX_PLANE_Y],
+ raw[i].planes[VPX_PLANE_U], raw[i].stride[VPX_PLANE_U],
+ raw[i].planes[VPX_PLANE_V], raw[i].stride[VPX_PLANE_V],
+ raw[i].d_w, raw[i].d_h, 1);
}
}
diff --git a/vpx/src/vpx_encoder.c b/vpx/src/vpx_encoder.c
index bddad23..03ddc62 100644
--- a/vpx/src/vpx_encoder.c
+++ b/vpx/src/vpx_encoder.c
@@ -243,6 +243,7 @@
ctx--;
if (img) img--;
}
+ ctx++;
}
FLOATING_POINT_RESTORE();
diff --git a/vpx/vpx_integer.h b/vpx/vpx_integer.h
index 608760f..218bca7 100644
--- a/vpx/vpx_integer.h
+++ b/vpx/vpx_integer.h
@@ -29,16 +29,8 @@
typedef unsigned __int64 uint64_t;
#endif
-#ifdef HAVE_ARMV6
-typedef unsigned int int_fast16_t;
-#else
-typedef signed short int_fast16_t;
-#endif
-typedef signed char int_fast8_t;
-typedef unsigned char uint_fast8_t;
-
#ifndef _UINTPTR_T_DEFINED
-typedef unsigned int uintptr_t;
+typedef size_t uintptr_t;
#endif
#else
diff --git a/vpxenc.c b/vpxenc.c
index 2e14f24..3637acb 100644
--- a/vpxenc.c
+++ b/vpxenc.c
@@ -807,7 +807,7 @@
{
EbmlLoc start;
- int i;
+ unsigned int i;
glob->cue_pos = ftello(glob->stream);
Ebml_StartSubElement(glob, &start, Cues);
