generic-library/vpx
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Revert "Remove Wextra warnings from vp9_sad.c"

Browse Source 
This reverts commit 7ab9a9587b

Nightly test http://build.webmproject.org/jenkins/view/libvpx-nightly-tests/job/libvpx%20unit%20tests%20(valgrind-2)/arch=x86_64-linux-gcc,filter=-*VP8*:*Large.*/276/console

Failed 

This patch did not address all the assembly issues 
some of the vp8 assembly counts on 5 arguments being passed in to this function:   

one example : vp8_sad8x16_wmt

Please address or split this into vp9 and vp8 patches.

Change-Id: I78afcc171649894f887bb8ee3c66de24aaddc7ca
This commit is contained in:
Jim Bankoski 2014-05-15 08:31:20 -07:00 committed by Gerrit Code Review
parent 7ab9a9587b
commit a16794dd31
13 changed files with 229 additions and 226 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

19
test/sad_test.cc
View File

@ -32,7 +32,8 @@
typedef unsigned int (*sad_m_by_n_fn_t)(const unsigned char *source_ptr,
int source_stride,
const unsigned char *reference_ptr,
int reference_stride);
int reference_stride,
unsigned int max_sad);
typedef std::tr1::tuple<int, int, sad_m_by_n_fn_t> sad_m_by_n_test_param_t;
typedef void (*sad_n_by_n_by_4_fn_t)(const uint8_t *src_ptr,
@ -86,7 +87,7 @@ class SADTestBase : public ::testing::Test {
// Sum of Absolute Differences. Given two blocks, calculate the absolute
// difference between two pixels in the same relative location; accumulate.
unsigned int ReferenceSAD(int block_idx = 0) {
unsigned int ReferenceSAD(unsigned int max_sad, int block_idx = 0) {
unsigned int sad = 0;
const uint8_t* const reference = GetReference(block_idx);
@ -95,6 +96,9 @@ class SADTestBase : public ::testing::Test {
sad += abs(source_data_[h * source_stride_ + w]
- reference[h * reference_stride_ + w]);
}
if (sad > max_sad) {
break;
}
}
return sad;
}
@ -130,20 +134,21 @@ class SADTest : public SADTestBase,
SADTest() : SADTestBase(GET_PARAM(0), GET_PARAM(1)) {}
protected:
unsigned int SAD(int block_idx = 0) {
unsigned int SAD(unsigned int max_sad, int block_idx = 0) {
unsigned int ret;
const uint8_t* const reference = GetReference(block_idx);
REGISTER_STATE_CHECK(ret = GET_PARAM(2)(source_data_, source_stride_,
reference, reference_stride_));
reference, reference_stride_,
max_sad));
return ret;
}
void CheckSad(unsigned int max_sad) {
unsigned int reference_sad, exp_sad;
reference_sad = ReferenceSAD();
exp_sad = SAD();
reference_sad = ReferenceSAD(max_sad);
exp_sad = SAD(max_sad);
if (reference_sad <= max_sad) {
ASSERT_EQ(exp_sad, reference_sad);
@ -174,7 +179,7 @@ class SADx4Test : public SADTestBase,
SADs(exp_sad);
for (int block = 0; block < 4; block++) {
reference_sad = ReferenceSAD(block);
reference_sad = ReferenceSAD(UINT_MAX, block);
EXPECT_EQ(exp_sad[block], reference_sad) << "block " << block;
}

10
vp8/common/rtcd_defs.pl
View File

@ -316,23 +316,23 @@ $vp8_variance_halfpixvar16x16_hv_neon_asm=vp8_variance_halfpixvar16x16_hv_neon;
#
# Single block SAD
#
add_proto qw/unsigned int vp8_sad4x4/, "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride";
add_proto qw/unsigned int vp8_sad4x4/, "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad";
specialize qw/vp8_sad4x4 mmx sse2 neon/;
$vp8_sad4x4_sse2=vp8_sad4x4_wmt;
add_proto qw/unsigned int vp8_sad8x8/, "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride";
add_proto qw/unsigned int vp8_sad8x8/, "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad";
specialize qw/vp8_sad8x8 mmx sse2 neon/;
$vp8_sad8x8_sse2=vp8_sad8x8_wmt;
add_proto qw/unsigned int vp8_sad8x16/, "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride";
add_proto qw/unsigned int vp8_sad8x16/, "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad";
specialize qw/vp8_sad8x16 mmx sse2 neon/;
$vp8_sad8x16_sse2=vp8_sad8x16_wmt;
add_proto qw/unsigned int vp8_sad16x8/, "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride";
add_proto qw/unsigned int vp8_sad16x8/, "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad";
specialize qw/vp8_sad16x8 mmx sse2 neon/;
$vp8_sad16x8_sse2=vp8_sad16x8_wmt;
add_proto qw/unsigned int vp8_sad16x16/, "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride";
add_proto qw/unsigned int vp8_sad16x16/, "const unsigned char *src_ptr, int src_stride, const unsigned char *ref_ptr, int ref_stride, unsigned int max_sad";
specialize qw/vp8_sad16x16 mmx sse2 sse3 media neon/;
$vp8_sad16x16_sse2=vp8_sad16x16_wmt;
$vp8_sad16x16_media=vp8_sad16x16_armv6;

180
vp8/common/sad_c.c
View File

@ -16,7 +16,7 @@
static unsigned int sad_mx_n_c(const unsigned char *src_ptr, int src_stride,
const unsigned char *ref_ptr, int ref_stride,
int m, int n)
unsigned int max_sad, int m, int n)
{
int r, c;
unsigned int sad = 0;
@ -28,6 +28,9 @@ static unsigned int sad_mx_n_c(const unsigned char *src_ptr, int src_stride,
sad += abs(src_ptr[c] - ref_ptr[c]);
}
if (sad > max_sad)
break;
src_ptr += src_stride;
ref_ptr += ref_stride;
}
@ -40,199 +43,204 @@ static unsigned int sad_mx_n_c(const unsigned char *src_ptr, int src_stride,
*/
unsigned int vp8_sad16x16_c(const unsigned char *src_ptr, int src_stride,
const unsigned char *ref_ptr, int ref_stride)
const unsigned char *ref_ptr, int ref_stride,
unsigned int max_sad)
{
return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, 16, 16);
return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, max_sad, 16, 16);
}
unsigned int vp8_sad8x8_c(const unsigned char *src_ptr, int src_stride,
const unsigned char *ref_ptr, int ref_stride)
const unsigned char *ref_ptr, int ref_stride,
unsigned int max_sad)
{
return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, 8, 8);
return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, max_sad, 8, 8);
}
unsigned int vp8_sad16x8_c(const unsigned char *src_ptr, int src_stride,
const unsigned char *ref_ptr, int ref_stride)
const unsigned char *ref_ptr, int ref_stride,
unsigned int max_sad)
{
return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, 16, 8);
return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, max_sad, 16, 8);
}
unsigned int vp8_sad8x16_c(const unsigned char *src_ptr, int src_stride,
const unsigned char *ref_ptr, int ref_stride)
const unsigned char *ref_ptr, int ref_stride,
unsigned int max_sad)
{
return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, 8, 16);
return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, max_sad, 8, 16);
}
unsigned int vp8_sad4x4_c(const unsigned char *src_ptr, int src_stride,
const unsigned char *ref_ptr, int ref_stride)
const unsigned char *ref_ptr, int ref_stride,
unsigned int max_sad)
{
return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, 4, 4);
return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, max_sad, 4, 4);
}
void vp8_sad16x16x3_c(const unsigned char *src_ptr, int src_stride,
const unsigned char *ref_ptr, int ref_stride,
unsigned int *sad_array)
{
sad_array[0] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 0, ref_stride);
sad_array[1] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 1, ref_stride);
sad_array[2] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 2, ref_stride);
sad_array[0] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 0, ref_stride, UINT_MAX);
sad_array[1] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, UINT_MAX);
sad_array[2] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, UINT_MAX);
}
void vp8_sad16x16x8_c(const unsigned char *src_ptr, int src_stride,
const unsigned char *ref_ptr, int ref_stride,
unsigned short *sad_array)
{
sad_array[0] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 0, ref_stride);
sad_array[1] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 1, ref_stride);
sad_array[2] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 2, ref_stride);
sad_array[3] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 3, ref_stride);
sad_array[4] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 4, ref_stride);
sad_array[5] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 5, ref_stride);
sad_array[6] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 6, ref_stride);
sad_array[7] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 7, ref_stride);
sad_array[0] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 0, ref_stride, UINT_MAX);
sad_array[1] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, UINT_MAX);
sad_array[2] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, UINT_MAX);
sad_array[3] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 3, ref_stride, UINT_MAX);
sad_array[4] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 4, ref_stride, UINT_MAX);
sad_array[5] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 5, ref_stride, UINT_MAX);
sad_array[6] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 6, ref_stride, UINT_MAX);
sad_array[7] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 7, ref_stride, UINT_MAX);
}
void vp8_sad16x8x3_c(const unsigned char *src_ptr, int src_stride,
const unsigned char *ref_ptr, int ref_stride,
unsigned int *sad_array)
{
sad_array[0] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 0, ref_stride);
sad_array[1] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 1, ref_stride);
sad_array[2] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 2, ref_stride);
sad_array[0] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 0, ref_stride, UINT_MAX);
sad_array[1] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, UINT_MAX);
sad_array[2] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, UINT_MAX);
}
void vp8_sad16x8x8_c(const unsigned char *src_ptr, int src_stride,
const unsigned char *ref_ptr, int ref_stride,
unsigned short *sad_array)
{
sad_array[0] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 0, ref_stride);
sad_array[1] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 1, ref_stride);
sad_array[2] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 2, ref_stride);
sad_array[3] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 3, ref_stride);
sad_array[4] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 4, ref_stride);
sad_array[5] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 5, ref_stride);
sad_array[6] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 6, ref_stride);
sad_array[7] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 7, ref_stride);
sad_array[0] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 0, ref_stride, UINT_MAX);
sad_array[1] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, UINT_MAX);
sad_array[2] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, UINT_MAX);
sad_array[3] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 3, ref_stride, UINT_MAX);
sad_array[4] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 4, ref_stride, UINT_MAX);
sad_array[5] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 5, ref_stride, UINT_MAX);
sad_array[6] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 6, ref_stride, UINT_MAX);
sad_array[7] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 7, ref_stride, UINT_MAX);
}
void vp8_sad8x8x3_c(const unsigned char *src_ptr, int src_stride,
const unsigned char *ref_ptr, int ref_stride,
unsigned int *sad_array)
{
sad_array[0] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 0, ref_stride);
sad_array[1] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 1, ref_stride);
sad_array[2] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 2, ref_stride);
sad_array[0] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 0, ref_stride, UINT_MAX);
sad_array[1] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, UINT_MAX);
sad_array[2] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, UINT_MAX);
}
void vp8_sad8x8x8_c(const unsigned char *src_ptr, int src_stride,
const unsigned char *ref_ptr, int ref_stride,
unsigned short *sad_array)
{
sad_array[0] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 0, ref_stride);
sad_array[1] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 1, ref_stride);
sad_array[2] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 2, ref_stride);
sad_array[3] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 3, ref_stride);
sad_array[4] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 4, ref_stride);
sad_array[5] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 5, ref_stride);
sad_array[6] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 6, ref_stride);
sad_array[7] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 7, ref_stride);
sad_array[0] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 0, ref_stride, UINT_MAX);
sad_array[1] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, UINT_MAX);
sad_array[2] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, UINT_MAX);
sad_array[3] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 3, ref_stride, UINT_MAX);
sad_array[4] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 4, ref_stride, UINT_MAX);
sad_array[5] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 5, ref_stride, UINT_MAX);
sad_array[6] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 6, ref_stride, UINT_MAX);
sad_array[7] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 7, ref_stride, UINT_MAX);
}
void vp8_sad8x16x3_c(const unsigned char *src_ptr, int src_stride,
const unsigned char *ref_ptr, int ref_stride,
unsigned int *sad_array)
{
sad_array[0] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 0, ref_stride);
sad_array[1] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 1, ref_stride);
sad_array[2] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 2, ref_stride);
sad_array[0] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 0, ref_stride, UINT_MAX);
sad_array[1] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, UINT_MAX);
sad_array[2] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, UINT_MAX);
}
void vp8_sad8x16x8_c(const unsigned char *src_ptr, int src_stride,
const unsigned char *ref_ptr, int ref_stride,
unsigned short *sad_array)
{
sad_array[0] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 0, ref_stride);
sad_array[1] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 1, ref_stride);
sad_array[2] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 2, ref_stride);
sad_array[3] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 3, ref_stride);
sad_array[4] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 4, ref_stride);
sad_array[5] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 5, ref_stride);
sad_array[6] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 6, ref_stride);
sad_array[7] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 7, ref_stride);
sad_array[0] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 0, ref_stride, UINT_MAX);
sad_array[1] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, UINT_MAX);
sad_array[2] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, UINT_MAX);
sad_array[3] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 3, ref_stride, UINT_MAX);
sad_array[4] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 4, ref_stride, UINT_MAX);
sad_array[5] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 5, ref_stride, UINT_MAX);
sad_array[6] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 6, ref_stride, UINT_MAX);
sad_array[7] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 7, ref_stride, UINT_MAX);
}
void vp8_sad4x4x3_c(const unsigned char *src_ptr, int src_stride,
const unsigned char *ref_ptr, int ref_stride,
unsigned int *sad_array)
{
sad_array[0] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 0, ref_stride);
sad_array[1] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 1, ref_stride);
sad_array[2] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 2, ref_stride);
sad_array[0] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 0, ref_stride, UINT_MAX);
sad_array[1] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, UINT_MAX);
sad_array[2] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, UINT_MAX);
}
void vp8_sad4x4x8_c(const unsigned char *src_ptr, int src_stride,
const unsigned char *ref_ptr, int ref_stride,
unsigned short *sad_array)
{
sad_array[0] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 0, ref_stride);
sad_array[1] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 1, ref_stride);
sad_array[2] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 2, ref_stride);
sad_array[3] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 3, ref_stride);
sad_array[4] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 4, ref_stride);
sad_array[5] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 5, ref_stride);
sad_array[6] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 6, ref_stride);
sad_array[7] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 7, ref_stride);
sad_array[0] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 0, ref_stride, UINT_MAX);
sad_array[1] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, UINT_MAX);
sad_array[2] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, UINT_MAX);
sad_array[3] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 3, ref_stride, UINT_MAX);
sad_array[4] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 4, ref_stride, UINT_MAX);
sad_array[5] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 5, ref_stride, UINT_MAX);
sad_array[6] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 6, ref_stride, UINT_MAX);
sad_array[7] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 7, ref_stride, UINT_MAX);
}
void vp8_sad16x16x4d_c(const unsigned char *src_ptr, int src_stride,
const unsigned char * const ref_ptr[], int ref_stride,
unsigned int *sad_array)
{
sad_array[0] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr[0], ref_stride);
sad_array[1] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr[1], ref_stride);
sad_array[2] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr[2], ref_stride);
sad_array[3] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr[3], ref_stride);
sad_array[0] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr[0], ref_stride, UINT_MAX);
sad_array[1] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr[1], ref_stride, UINT_MAX);
sad_array[2] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr[2], ref_stride, UINT_MAX);
sad_array[3] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr[3], ref_stride, UINT_MAX);
}
void vp8_sad16x8x4d_c(const unsigned char *src_ptr, int src_stride,
const unsigned char * const ref_ptr[], int ref_stride,
unsigned int *sad_array)
{
sad_array[0] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr[0], ref_stride);
sad_array[1] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr[1], ref_stride);
sad_array[2] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr[2], ref_stride);
sad_array[3] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr[3], ref_stride);
sad_array[0] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr[0], ref_stride, UINT_MAX);
sad_array[1] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr[1], ref_stride, UINT_MAX);
sad_array[2] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr[2], ref_stride, UINT_MAX);
sad_array[3] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr[3], ref_stride, UINT_MAX);
}
void vp8_sad8x8x4d_c(const unsigned char *src_ptr, int src_stride,
const unsigned char * const ref_ptr[], int ref_stride,
unsigned int *sad_array)
{
sad_array[0] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr[0], ref_stride);
sad_array[1] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr[1], ref_stride);
sad_array[2] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr[2], ref_stride);
sad_array[3] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr[3], ref_stride);
sad_array[0] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr[0], ref_stride, UINT_MAX);
sad_array[1] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr[1], ref_stride, UINT_MAX);
sad_array[2] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr[2], ref_stride, UINT_MAX);
sad_array[3] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr[3], ref_stride, UINT_MAX);
}
void vp8_sad8x16x4d_c(const unsigned char *src_ptr, int src_stride,
const unsigned char * const ref_ptr[], int ref_stride,
unsigned int *sad_array)
{
sad_array[0] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr[0], ref_stride);
sad_array[1] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr[1], ref_stride);
sad_array[2] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr[2], ref_stride);
sad_array[3] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr[3], ref_stride);
sad_array[0] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr[0], ref_stride, UINT_MAX);
sad_array[1] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr[1], ref_stride, UINT_MAX);
sad_array[2] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr[2], ref_stride, UINT_MAX);
sad_array[3] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr[3], ref_stride, UINT_MAX);
}
void vp8_sad4x4x4d_c(const unsigned char *src_ptr, int src_stride,
const unsigned char * const ref_ptr[], int ref_stride,
unsigned int *sad_array)
{
sad_array[0] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr[0], ref_stride);
sad_array[1] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr[1], ref_stride);
sad_array[2] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr[2], ref_stride);
sad_array[3] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr[3], ref_stride);
sad_array[0] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr[0], ref_stride, UINT_MAX);
sad_array[1] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr[1], ref_stride, UINT_MAX);
sad_array[2] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr[2], ref_stride, UINT_MAX);
sad_array[3] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr[3], ref_stride, UINT_MAX);
}
/* Copy 2 macroblocks to a buffer */

3
vp8/common/variance.h
View File

@ -22,7 +22,8 @@ typedef unsigned int(*vp8_sad_fn_t)(
const unsigned char *src_ptr,
int source_stride,
const unsigned char *ref_ptr,
int ref_stride);
int ref_stride,
unsigned int max_sad);
typedef void (*vp8_copy32xn_fn_t)(
const unsigned char *src_ptr,

53
vp8/encoder/mcomp.c
View File

@ -898,7 +898,7 @@ int vp8_hex_search
this_offset = base_offset + (br * (pre_stride)) + bc;
this_mv.as_mv.row = br;
this_mv.as_mv.col = bc;
bestsad = vfp->sdf(what, what_stride, this_offset, in_what_stride)
bestsad = vfp->sdf(what, what_stride, this_offset, in_what_stride, UINT_MAX)
+ mvsad_err_cost(&this_mv, &fcenter_mv, mvsadcost, sad_per_bit);
#if CONFIG_MULTI_RES_ENCODING
@ -923,7 +923,7 @@ int vp8_hex_search
this_mv.as_mv.row = br + hex[i].row;
this_mv.as_mv.col = bc + hex[i].col;
this_offset = base_offset + (this_mv.as_mv.row * in_what_stride) + this_mv.as_mv.col;
thissad = vfp->sdf(what, what_stride, this_offset, in_what_stride);
thissad = vfp->sdf(what, what_stride, this_offset, in_what_stride, bestsad);
CHECK_BETTER
}
}else
@ -934,7 +934,7 @@ int vp8_hex_search
this_mv.as_mv.col = bc + hex[i].col;
CHECK_POINT
this_offset = base_offset + (this_mv.as_mv.row * in_what_stride) + this_mv.as_mv.col;
thissad = vfp->sdf(what, what_stride, this_offset, in_what_stride);
thissad = vfp->sdf(what, what_stride, this_offset, in_what_stride, bestsad);
CHECK_BETTER
}
}
@ -960,8 +960,7 @@ int vp8_hex_search
this_mv.as_mv.row = br + next_chkpts[k][i].row;
this_mv.as_mv.col = bc + next_chkpts[k][i].col;
this_offset = base_offset + (this_mv.as_mv.row * (in_what_stride)) + this_mv.as_mv.col;
thissad = vfp->sdf(what, what_stride, this_offset,
in_what_stride);
thissad = vfp->sdf(what, what_stride, this_offset, in_what_stride, bestsad);
CHECK_BETTER
}
}else
@ -972,8 +971,7 @@ int vp8_hex_search
this_mv.as_mv.col = bc + next_chkpts[k][i].col;
CHECK_POINT
this_offset = base_offset + (this_mv.as_mv.row * (in_what_stride)) + this_mv.as_mv.col;
thissad = vfp->sdf(what, what_stride, this_offset,
in_what_stride);
thissad = vfp->sdf(what, what_stride, this_offset, in_what_stride, bestsad);
CHECK_BETTER
}
}
@ -1004,8 +1002,7 @@ cal_neighbors:
this_mv.as_mv.row = br + neighbors[i].row;
this_mv.as_mv.col = bc + neighbors[i].col;
this_offset = base_offset + (this_mv.as_mv.row * (in_what_stride)) + this_mv.as_mv.col;
thissad = vfp->sdf(what, what_stride, this_offset,
in_what_stride);
thissad = vfp->sdf(what, what_stride, this_offset, in_what_stride, bestsad);
CHECK_BETTER
}
}else
@ -1016,8 +1013,7 @@ cal_neighbors:
this_mv.as_mv.col = bc + neighbors[i].col;
CHECK_POINT
this_offset = base_offset + (this_mv.as_mv.row * (in_what_stride)) + this_mv.as_mv.col;
thissad = vfp->sdf(what, what_stride, this_offset,
in_what_stride);
thissad = vfp->sdf(what, what_stride, this_offset, in_what_stride, bestsad);
CHECK_BETTER
}
}
@ -1101,7 +1097,7 @@ int vp8_diamond_search_sad_c
best_address = in_what;
/* Check the starting position */
bestsad = fn_ptr->sdf(what, what_stride, in_what, in_what_stride)
bestsad = fn_ptr->sdf(what, what_stride, in_what, in_what_stride, UINT_MAX)
+ mvsad_err_cost(best_mv, &fcenter_mv, mvsadcost, sad_per_bit);
/* search_param determines the length of the initial step and hence
@ -1126,8 +1122,7 @@ int vp8_diamond_search_sad_c
{
check_here = ss[i].offset + best_address;
thissad = fn_ptr->sdf(what, what_stride, check_here,
in_what_stride);
thissad = fn_ptr->sdf(what, what_stride, check_here, in_what_stride, bestsad);
if (thissad < bestsad)
{
@ -1226,7 +1221,7 @@ int vp8_diamond_search_sadx4
best_address = in_what;
/* Check the starting position */
bestsad = fn_ptr->sdf(what, what_stride, in_what, in_what_stride)
bestsad = fn_ptr->sdf(what, what_stride, in_what, in_what_stride, UINT_MAX)
+ mvsad_err_cost(best_mv, &fcenter_mv, mvsadcost, sad_per_bit);
/* search_param determines the length of the initial step and hence the
@ -1294,8 +1289,7 @@ int vp8_diamond_search_sadx4
(this_row_offset > x->mv_row_min) && (this_row_offset < x->mv_row_max))
{
check_here = ss[i].offset + best_address;
thissad = fn_ptr->sdf(what, what_stride, check_here,
in_what_stride);
thissad = fn_ptr->sdf(what, what_stride, check_here, in_what_stride, bestsad);
if (thissad < bestsad)
{
@ -1379,7 +1373,7 @@ int vp8_full_search_sad_c(MACROBLOCK *x, BLOCK *b, BLOCKD *d, int_mv *ref_mv,
/* Baseline value at the centre */
bestsad = fn_ptr->sdf(what, what_stride, bestaddress,
in_what_stride)
in_what_stride, UINT_MAX)
+ mvsad_err_cost(best_mv, &fcenter_mv, mvsadcost, sad_per_bit);
/* Apply further limits to prevent us looking using vectors that
@ -1404,8 +1398,7 @@ int vp8_full_search_sad_c(MACROBLOCK *x, BLOCK *b, BLOCKD *d, int_mv *ref_mv,
for (c = col_min; c < col_max; c++)
{
thissad = fn_ptr->sdf(what, what_stride, check_here,
in_what_stride);
thissad = fn_ptr->sdf(what, what_stride, check_here, in_what_stride, bestsad);
this_mv.as_mv.col = c;
thissad += mvsad_err_cost(&this_mv, &fcenter_mv,
@ -1478,7 +1471,7 @@ int vp8_full_search_sadx3(MACROBLOCK *x, BLOCK *b, BLOCKD *d, int_mv *ref_mv,
/* Baseline value at the centre */
bestsad = fn_ptr->sdf(what, what_stride, bestaddress,
in_what_stride)
in_what_stride, UINT_MAX)
+ mvsad_err_cost(best_mv, &fcenter_mv, mvsadcost, sad_per_bit);
/* Apply further limits to prevent us looking using vectors that stretch
@ -1534,8 +1527,7 @@ int vp8_full_search_sadx3(MACROBLOCK *x, BLOCK *b, BLOCKD *d, int_mv *ref_mv,
while (c < col_max)
{
thissad = fn_ptr->sdf(what, what_stride, check_here,
in_what_stride);
thissad = fn_ptr->sdf(what, what_stride, check_here, in_what_stride, bestsad);
if (thissad < bestsad)
{
@ -1614,7 +1606,7 @@ int vp8_full_search_sadx8(MACROBLOCK *x, BLOCK *b, BLOCKD *d, int_mv *ref_mv,
/* Baseline value at the centre */
bestsad = fn_ptr->sdf(what, what_stride,
bestaddress, in_what_stride)
bestaddress, in_what_stride, UINT_MAX)
+ mvsad_err_cost(best_mv, &fcenter_mv, mvsadcost, sad_per_bit);
/* Apply further limits to prevent us looking using vectors that stretch
@ -1700,8 +1692,7 @@ int vp8_full_search_sadx8(MACROBLOCK *x, BLOCK *b, BLOCKD *d, int_mv *ref_mv,
while (c < col_max)
{
thissad = fn_ptr->sdf(what, what_stride, check_here,
in_what_stride);
thissad = fn_ptr->sdf(what, what_stride, check_here , in_what_stride, bestsad);
if (thissad < bestsad)
{
@ -1760,7 +1751,7 @@ int vp8_refining_search_sad_c(MACROBLOCK *x, BLOCK *b, BLOCKD *d, int_mv *ref_mv
fcenter_mv.as_mv.col = center_mv->as_mv.col >> 3;
bestsad = fn_ptr->sdf(what, what_stride, best_address,
in_what_stride)
in_what_stride, UINT_MAX)
+ mvsad_err_cost(ref_mv, &fcenter_mv, mvsadcost, error_per_bit);
for (i=0; i<search_range; i++)
@ -1776,8 +1767,7 @@ int vp8_refining_search_sad_c(MACROBLOCK *x, BLOCK *b, BLOCKD *d, int_mv *ref_mv
(this_row_offset > x->mv_row_min) && (this_row_offset < x->mv_row_max))
{
check_here = (neighbors[j].row)*in_what_stride + neighbors[j].col + best_address;
thissad = fn_ptr->sdf(what, what_stride, check_here,
in_what_stride);
thissad = fn_ptr->sdf(what, what_stride, check_here , in_what_stride, bestsad);
if (thissad < bestsad)
{
@ -1841,7 +1831,7 @@ int vp8_refining_search_sadx4(MACROBLOCK *x, BLOCK *b, BLOCKD *d,
fcenter_mv.as_mv.col = center_mv->as_mv.col >> 3;
bestsad = fn_ptr->sdf(what, what_stride, best_address,
in_what_stride)
in_what_stride, UINT_MAX)
+ mvsad_err_cost(ref_mv, &fcenter_mv, mvsadcost, error_per_bit);
for (i=0; i<search_range; i++)
@ -1892,8 +1882,7 @@ int vp8_refining_search_sadx4(MACROBLOCK *x, BLOCK *b, BLOCKD *d,
(this_row_offset > x->mv_row_min) && (this_row_offset < x->mv_row_max))
{
check_here = (neighbors[j].row)*in_what_stride + neighbors[j].col + best_address;
thissad = fn_ptr->sdf(what, what_stride, check_here,
in_what_stride);
thissad = fn_ptr->sdf(what, what_stride, check_here , in_what_stride, bestsad);
if (thissad < bestsad)
{

36
vp8/encoder/rdopt.c
View File

@ -1686,25 +1686,16 @@ void vp8_cal_sad(VP8_COMP *cpi, MACROBLOCKD *xd, MACROBLOCK *x, int recon_yoffse
}else if(xd->mb_to_top_edge==0)
{ /* only has left MB for sad calculation. */
near_sad[0] = near_sad[2] = INT_MAX;
near_sad[1] = cpi->fn_ptr[BLOCK_16X16].sdf(
src_y_ptr, b->src_stride, xd->dst.y_buffer - 16, xd->dst.y_stride);
near_sad[1] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, xd->dst.y_buffer - 16,xd->dst.y_stride, UINT_MAX);
}else if(xd->mb_to_left_edge ==0)
{ /* only has left MB for sad calculation. */
near_sad[1] = near_sad[2] = INT_MAX;
near_sad[0] = cpi->fn_ptr[BLOCK_16X16].sdf(
src_y_ptr, b->src_stride, xd->dst.y_buffer - xd->dst.y_stride *16,
xd->dst.y_stride);
near_sad[0] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, xd->dst.y_buffer - xd->dst.y_stride *16,xd->dst.y_stride, UINT_MAX);
}else
{
near_sad[0] = cpi->fn_ptr[BLOCK_16X16].sdf(
src_y_ptr, b->src_stride, xd->dst.y_buffer - xd->dst.y_stride *16,
xd->dst.y_stride);
near_sad[1] = cpi->fn_ptr[BLOCK_16X16].sdf(
src_y_ptr, b->src_stride, xd->dst.y_buffer - 16, xd->dst.y_stride);
near_sad[2] = cpi->fn_ptr[BLOCK_16X16].sdf(
src_y_ptr,
b->src_stride, xd->dst.y_buffer - xd->dst.y_stride *16 -16,
xd->dst.y_stride);
near_sad[0] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, xd->dst.y_buffer - xd->dst.y_stride *16,xd->dst.y_stride, UINT_MAX);
near_sad[1] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, xd->dst.y_buffer - 16,xd->dst.y_stride, UINT_MAX);
near_sad[2] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, xd->dst.y_buffer - xd->dst.y_stride *16 -16,xd->dst.y_stride, UINT_MAX);
}
if(cpi->common.last_frame_type != KEY_FRAME)
@ -1719,21 +1710,14 @@ void vp8_cal_sad(VP8_COMP *cpi, MACROBLOCKD *xd, MACROBLOCK *x, int recon_yoffse
if(xd->mb_to_bottom_edge==0) near_sad[7] = INT_MAX;
if(near_sad[4] != INT_MAX)
near_sad[4] = cpi->fn_ptr[BLOCK_16X16].sdf(
src_y_ptr, b->src_stride, pre_y_buffer - pre_y_stride *16,
pre_y_stride);
near_sad[4] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, pre_y_buffer - pre_y_stride *16, pre_y_stride, UINT_MAX);
if(near_sad[5] != INT_MAX)
near_sad[5] = cpi->fn_ptr[BLOCK_16X16].sdf(
src_y_ptr, b->src_stride, pre_y_buffer - 16, pre_y_stride);
near_sad[3] = cpi->fn_ptr[BLOCK_16X16].sdf(
src_y_ptr, b->src_stride, pre_y_buffer, pre_y_stride);
near_sad[5] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, pre_y_buffer - 16, pre_y_stride, UINT_MAX);
near_sad[3] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, pre_y_buffer, pre_y_stride, UINT_MAX);
if(near_sad[6] != INT_MAX)
near_sad[6] = cpi->fn_ptr[BLOCK_16X16].sdf(
src_y_ptr, b->src_stride, pre_y_buffer + 16, pre_y_stride);
near_sad[6] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, pre_y_buffer + 16, pre_y_stride, UINT_MAX);
if(near_sad[7] != INT_MAX)
near_sad[7] = cpi->fn_ptr[BLOCK_16X16].sdf(
src_y_ptr, b->src_stride, pre_y_buffer + pre_y_stride *16,
pre_y_stride);
near_sad[7] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, pre_y_buffer + pre_y_stride *16, pre_y_stride, UINT_MAX);
}
if(cpi->common.last_frame_type != KEY_FRAME)

52
vp9/common/vp9_rtcd_defs.pl
View File

@ -520,82 +520,82 @@ specialize qw/vp9_sub_pixel_variance4x4/, "$sse_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sub_pixel_avg_variance4x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred";
specialize qw/vp9_sub_pixel_avg_variance4x4/, "$sse_x86inc", "$ssse3_x86inc";
add_proto qw/unsigned int vp9_sad64x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride";
add_proto qw/unsigned int vp9_sad64x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad";
specialize qw/vp9_sad64x64/, "$sse2_x86inc";
add_proto qw/unsigned int vp9_sad32x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride";
add_proto qw/unsigned int vp9_sad32x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad";
specialize qw/vp9_sad32x64/, "$sse2_x86inc";
add_proto qw/unsigned int vp9_sad64x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride";
add_proto qw/unsigned int vp9_sad64x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad";
specialize qw/vp9_sad64x32/, "$sse2_x86inc";
add_proto qw/unsigned int vp9_sad32x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride";
add_proto qw/unsigned int vp9_sad32x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad";
specialize qw/vp9_sad32x16/, "$sse2_x86inc";
add_proto qw/unsigned int vp9_sad16x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride";
add_proto qw/unsigned int vp9_sad16x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad";
specialize qw/vp9_sad16x32/, "$sse2_x86inc";
add_proto qw/unsigned int vp9_sad32x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride";
add_proto qw/unsigned int vp9_sad32x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad";
specialize qw/vp9_sad32x32/, "$sse2_x86inc";
add_proto qw/unsigned int vp9_sad16x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride";
add_proto qw/unsigned int vp9_sad16x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad";
specialize qw/vp9_sad16x16 mmx/, "$sse2_x86inc";
add_proto qw/unsigned int vp9_sad16x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride";
add_proto qw/unsigned int vp9_sad16x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad";
specialize qw/vp9_sad16x8 mmx/, "$sse2_x86inc";
add_proto qw/unsigned int vp9_sad8x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride";
add_proto qw/unsigned int vp9_sad8x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad";
specialize qw/vp9_sad8x16 mmx/, "$sse2_x86inc";
add_proto qw/unsigned int vp9_sad8x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride";
add_proto qw/unsigned int vp9_sad8x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad";
specialize qw/vp9_sad8x8 mmx/, "$sse2_x86inc";
add_proto qw/unsigned int vp9_sad8x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride";
add_proto qw/unsigned int vp9_sad8x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad";
specialize qw/vp9_sad8x4/, "$sse2_x86inc";
add_proto qw/unsigned int vp9_sad4x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride";
add_proto qw/unsigned int vp9_sad4x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad";
specialize qw/vp9_sad4x8/, "$sse_x86inc";
add_proto qw/unsigned int vp9_sad4x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride";
add_proto qw/unsigned int vp9_sad4x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad";
specialize qw/vp9_sad4x4 mmx/, "$sse_x86inc";
add_proto qw/unsigned int vp9_sad64x64_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
add_proto qw/unsigned int vp9_sad64x64_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad";
specialize qw/vp9_sad64x64_avg/, "$sse2_x86inc";
add_proto qw/unsigned int vp9_sad32x64_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
add_proto qw/unsigned int vp9_sad32x64_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad";
specialize qw/vp9_sad32x64_avg/, "$sse2_x86inc";
add_proto qw/unsigned int vp9_sad64x32_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
add_proto qw/unsigned int vp9_sad64x32_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad";
specialize qw/vp9_sad64x32_avg/, "$sse2_x86inc";
add_proto qw/unsigned int vp9_sad32x16_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
add_proto qw/unsigned int vp9_sad32x16_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad";
specialize qw/vp9_sad32x16_avg/, "$sse2_x86inc";
add_proto qw/unsigned int vp9_sad16x32_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
add_proto qw/unsigned int vp9_sad16x32_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad";
specialize qw/vp9_sad16x32_avg/, "$sse2_x86inc";
add_proto qw/unsigned int vp9_sad32x32_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
add_proto qw/unsigned int vp9_sad32x32_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad";
specialize qw/vp9_sad32x32_avg/, "$sse2_x86inc";
add_proto qw/unsigned int vp9_sad16x16_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
add_proto qw/unsigned int vp9_sad16x16_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad";
specialize qw/vp9_sad16x16_avg/, "$sse2_x86inc";
add_proto qw/unsigned int vp9_sad16x8_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
add_proto qw/unsigned int vp9_sad16x8_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad";
specialize qw/vp9_sad16x8_avg/, "$sse2_x86inc";
add_proto qw/unsigned int vp9_sad8x16_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
add_proto qw/unsigned int vp9_sad8x16_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad";
specialize qw/vp9_sad8x16_avg/, "$sse2_x86inc";
add_proto qw/unsigned int vp9_sad8x8_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
add_proto qw/unsigned int vp9_sad8x8_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad";
specialize qw/vp9_sad8x8_avg/, "$sse2_x86inc";
add_proto qw/unsigned int vp9_sad8x4_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
add_proto qw/unsigned int vp9_sad8x4_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad";
specialize qw/vp9_sad8x4_avg/, "$sse2_x86inc";
add_proto qw/unsigned int vp9_sad4x8_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
add_proto qw/unsigned int vp9_sad4x8_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad";
specialize qw/vp9_sad4x8_avg/, "$sse_x86inc";
add_proto qw/unsigned int vp9_sad4x4_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
add_proto qw/unsigned int vp9_sad4x4_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad";
specialize qw/vp9_sad4x4_avg/, "$sse_x86inc";
add_proto qw/void vp9_sad64x64x3/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sad_array";

6
vp9/encoder/vp9_encodeframe.c
View File

@ -1297,14 +1297,14 @@ static int is_background(VP9_COMP *cpi, const TileInfo *const tile,
if (row8x8_remaining >= MI_BLOCK_SIZE &&
col8x8_remaining >= MI_BLOCK_SIZE) {
this_sad = cpi->fn_ptr[BLOCK_64X64].sdf(src, src_stride,
pre, pre_stride);
pre, pre_stride, 0x7fffffff);
threshold = (1 << 12);
} else {
int r, c;
for (r = 0; r < row8x8_remaining; r += 2)
for (c = 0; c < col8x8_remaining; c += 2)
this_sad += cpi->fn_ptr[BLOCK_16X16].sdf(src, src_stride,
pre, pre_stride);
this_sad += cpi->fn_ptr[BLOCK_16X16].sdf(src, src_stride, pre,
pre_stride, 0x7fffffff);
threshold = (row8x8_remaining * col8x8_remaining) << 6;
}

11
vp9/encoder/vp9_mbgraph.c
View File

@ -73,7 +73,8 @@ static unsigned int do_16x16_motion_iteration(VP9_COMP *cpi,
x->mv_row_max = tmp_row_max;
return vp9_sad16x16(x->plane[0].src.buf, x->plane[0].src.stride,
xd->plane[0].dst.buf, xd->plane[0].dst.stride);
xd->plane[0].dst.buf, xd->plane[0].dst.stride,
INT_MAX);
}
static int do_16x16_motion_search(VP9_COMP *cpi, const MV *ref_mv,
@ -86,7 +87,8 @@ static int do_16x16_motion_search(VP9_COMP *cpi, const MV *ref_mv,
// Try zero MV first
// FIXME should really use something like near/nearest MV and/or MV prediction
err = vp9_sad16x16(x->plane[0].src.buf, x->plane[0].src.stride,
xd->plane[0].pre[0].buf, xd->plane[0].pre[0].stride);
xd->plane[0].pre[0].buf, xd->plane[0].pre[0].stride,
INT_MAX);
dst_mv->as_int = 0;
// Test last reference frame using the previous best mv as the
@ -122,7 +124,8 @@ static int do_16x16_zerozero_search(VP9_COMP *cpi, int_mv *dst_mv) {
// Try zero MV first
// FIXME should really use something like near/nearest MV and/or MV prediction
err = vp9_sad16x16(x->plane[0].src.buf, x->plane[0].src.stride,
xd->plane[0].pre[0].buf, xd->plane[0].pre[0].stride);
xd->plane[0].pre[0].buf, xd->plane[0].pre[0].stride,
INT_MAX);
dst_mv->as_int = 0;
@ -145,7 +148,7 @@ static int find_best_16x16_intra(VP9_COMP *cpi, PREDICTION_MODE *pbest_mode) {
xd->plane[0].dst.buf, xd->plane[0].dst.stride,
0, 0, 0);
err = vp9_sad16x16(x->plane[0].src.buf, x->plane[0].src.stride,
xd->plane[0].dst.buf, xd->plane[0].dst.stride);
xd->plane[0].dst.buf, xd->plane[0].dst.stride, best_err);
// find best
if (err < best_err) {

64
vp9/encoder/vp9_mcomp.c
View File

@ -524,7 +524,9 @@ static int vp9_pattern_search(const MACROBLOCK *x,
// Work out the start point for the search
bestsad = vfp->sdf(what->buf, what->stride,
get_buf_from_mv(in_what, ref_mv), in_what->stride);
get_buf_from_mv(in_what, ref_mv), in_what->stride,
0x7fffffff) + mvsad_err_cost(x, ref_mv, &fcenter_mv,
sad_per_bit);
// Search all possible scales upto the search param around the center point
// pick the scale of the point that is best as the starting scale of
@ -540,7 +542,7 @@ static int vp9_pattern_search(const MACROBLOCK *x,
bc + candidates[t][i].col};
thissad = vfp->sdf(what->buf, what->stride,
get_buf_from_mv(in_what, &this_mv),
in_what->stride);
in_what->stride, bestsad);
CHECK_BETTER
}
} else {
@ -551,7 +553,7 @@ static int vp9_pattern_search(const MACROBLOCK *x,
continue;
thissad = vfp->sdf(what->buf, what->stride,
get_buf_from_mv(in_what, &this_mv),
in_what->stride);
in_what->stride, bestsad);
CHECK_BETTER
}
}
@ -583,7 +585,7 @@ static int vp9_pattern_search(const MACROBLOCK *x,
bc + candidates[s][i].col};
thissad = vfp->sdf(what->buf, what->stride,
get_buf_from_mv(in_what, &this_mv),
in_what->stride);
in_what->stride, bestsad);
CHECK_BETTER
}
} else {
@ -594,7 +596,7 @@ static int vp9_pattern_search(const MACROBLOCK *x,
continue;
thissad = vfp->sdf(what->buf, what->stride,
get_buf_from_mv(in_what, &this_mv),
in_what->stride);
in_what->stride, bestsad);
CHECK_BETTER
}
}
@ -621,7 +623,7 @@ static int vp9_pattern_search(const MACROBLOCK *x,
bc + candidates[s][next_chkpts_indices[i]].col};
thissad = vfp->sdf(what->buf, what->stride,
get_buf_from_mv(in_what, &this_mv),
in_what->stride);
in_what->stride, bestsad);
CHECK_BETTER
}
} else {
@ -632,7 +634,7 @@ static int vp9_pattern_search(const MACROBLOCK *x,
continue;
thissad = vfp->sdf(what->buf, what->stride,
get_buf_from_mv(in_what, &this_mv),
in_what->stride);
in_what->stride, bestsad);
CHECK_BETTER
}
}
@ -659,7 +661,7 @@ static int vp9_pattern_search(const MACROBLOCK *x,
bc + neighbors[i].col};
thissad = vfp->sdf(what->buf, what->stride,
get_buf_from_mv(in_what, &this_mv),
in_what->stride);
in_what->stride, bestsad);
CHECK_BETTER
}
} else {
@ -670,7 +672,7 @@ static int vp9_pattern_search(const MACROBLOCK *x,
continue;
thissad = vfp->sdf(what->buf, what->stride,
get_buf_from_mv(in_what, &this_mv),
in_what->stride);
in_what->stride, bestsad);
CHECK_BETTER
}
}
@ -888,7 +890,8 @@ int vp9_full_range_search_c(const MACROBLOCK *x,
*best_mv = *ref_mv;
*num00 = 11;
best_sad = fn_ptr->sdf(what->buf, what->stride,
get_buf_from_mv(in_what, ref_mv), in_what->stride) +
get_buf_from_mv(in_what, ref_mv), in_what->stride,
0x7fffffff) +
mvsad_err_cost(x, ref_mv, &fcenter_mv, sad_per_bit);
start_row = MAX(-range, x->mv_row_min - ref_mv->row);
start_col = MAX(-range, x->mv_col_min - ref_mv->col);
@ -922,7 +925,7 @@ int vp9_full_range_search_c(const MACROBLOCK *x,
for (i = 0; i < end_col - c; ++i) {
const MV mv = {ref_mv->row + r, ref_mv->col + c + i};
unsigned int sad = fn_ptr->sdf(what->buf, what->stride,
get_buf_from_mv(in_what, &mv), in_what->stride);
get_buf_from_mv(in_what, &mv), in_what->stride, best_sad);
if (sad < best_sad) {
sad += mvsad_err_cost(x, &mv, &fcenter_mv, sad_per_bit);
if (sad < best_sad) {
@ -968,7 +971,7 @@ int vp9_diamond_search_sad_c(const MACROBLOCK *x,
// Check the starting position
best_sad = fn_ptr->sdf(what->buf, what->stride,
best_address, in_what->stride) +
best_address, in_what->stride, 0x7fffffff) +
mvsad_err_cost(x, best_mv, &fcenter_mv, sad_per_bit);
i = 1;
@ -979,7 +982,8 @@ int vp9_diamond_search_sad_c(const MACROBLOCK *x,
best_mv->col + ss[i].mv.col};
if (is_mv_in(x, &mv)) {
int sad = fn_ptr->sdf(what->buf, what->stride,
best_address + ss[i].offset, in_what->stride);
best_address + ss[i].offset, in_what->stride,
best_sad);
if (sad < best_sad) {
sad += mvsad_err_cost(x, &mv, &fcenter_mv, sad_per_bit);
if (sad < best_sad) {
@ -1004,7 +1008,7 @@ int vp9_diamond_search_sad_c(const MACROBLOCK *x,
if (is_mv_in(x, &this_mv)) {
int sad = fn_ptr->sdf(what->buf, what->stride,
best_address + ss[best_site].offset,
in_what->stride);
in_what->stride, best_sad);
if (sad < best_sad) {
sad += mvsad_err_cost(x, &this_mv, &fcenter_mv, sad_per_bit);
if (sad < best_sad) {
@ -1069,7 +1073,7 @@ int vp9_diamond_search_sadx4(const MACROBLOCK *x,
best_address = in_what;
// Check the starting position
bestsad = fn_ptr->sdf(what, what_stride, in_what, in_what_stride)
bestsad = fn_ptr->sdf(what, what_stride, in_what, in_what_stride, 0x7fffffff)
+ mvsad_err_cost(x, best_mv, &fcenter_mv, sad_per_bit);
i = 1;
@ -1121,7 +1125,7 @@ int vp9_diamond_search_sadx4(const MACROBLOCK *x,
if (is_mv_in(x, &this_mv)) {
const uint8_t *const check_here = ss[i].offset + best_address;
unsigned int thissad = fn_ptr->sdf(what, what_stride, check_here,
in_what_stride);
in_what_stride, bestsad);
if (thissad < bestsad) {
thissad += mvsad_err_cost(x, &this_mv, &fcenter_mv, sad_per_bit);
@ -1146,7 +1150,7 @@ int vp9_diamond_search_sadx4(const MACROBLOCK *x,
if (is_mv_in(x, &this_mv)) {
const uint8_t *const check_here = ss[best_site].offset + best_address;
unsigned int thissad = fn_ptr->sdf(what, what_stride, check_here,
in_what_stride);
in_what_stride, bestsad);
if (thissad < bestsad) {
thissad += mvsad_err_cost(&this_mv, &fcenter_mv,
mvjsadcost, mvsadcost, sad_per_bit);
@ -1245,7 +1249,7 @@ int vp9_full_search_sad_c(const MACROBLOCK *x, const MV *ref_mv,
const int col_max = MIN(ref_mv->col + distance, x->mv_col_max);
const MV fcenter_mv = {center_mv->row >> 3, center_mv->col >> 3};
int best_sad = fn_ptr->sdf(what->buf, what->stride,
get_buf_from_mv(in_what, ref_mv), in_what->stride) +
get_buf_from_mv(in_what, ref_mv), in_what->stride, 0x7fffffff) +
mvsad_err_cost(x, ref_mv, &fcenter_mv, sad_per_bit);
*best_mv = *ref_mv;
@ -1253,7 +1257,7 @@ int vp9_full_search_sad_c(const MACROBLOCK *x, const MV *ref_mv,
for (c = col_min; c < col_max; ++c) {
const MV mv = {r, c};
const int sad = fn_ptr->sdf(what->buf, what->stride,
get_buf_from_mv(in_what, &mv), in_what->stride) +
get_buf_from_mv(in_what, &mv), in_what->stride, best_sad) +
mvsad_err_cost(x, &mv, &fcenter_mv, sad_per_bit);
if (sad < best_sad) {
best_sad = sad;
@ -1278,7 +1282,7 @@ int vp9_full_search_sadx3(const MACROBLOCK *x, const MV *ref_mv,
const int col_max = MIN(ref_mv->col + distance, x->mv_col_max);
const MV fcenter_mv = {center_mv->row >> 3, center_mv->col >> 3};
unsigned int best_sad = fn_ptr->sdf(what->buf, what->stride,
get_buf_from_mv(in_what, ref_mv), in_what->stride) +
get_buf_from_mv(in_what, ref_mv), in_what->stride, 0x7fffffff) +
mvsad_err_cost(x, ref_mv, &fcenter_mv, sad_per_bit);
*best_mv = *ref_mv;
@ -1312,7 +1316,7 @@ int vp9_full_search_sadx3(const MACROBLOCK *x, const MV *ref_mv,
while (c < col_max) {
unsigned int sad = fn_ptr->sdf(what->buf, what->stride,
check_here, in_what->stride);
check_here, in_what->stride, best_sad);
if (sad < best_sad) {
const MV mv = {r, c};
sad += mvsad_err_cost(x, &mv, &fcenter_mv, sad_per_bit);
@ -1343,7 +1347,7 @@ int vp9_full_search_sadx8(const MACROBLOCK *x, const MV *ref_mv,
const int col_max = MIN(ref_mv->col + distance, x->mv_col_max);
const MV fcenter_mv = {center_mv->row >> 3, center_mv->col >> 3};
unsigned int best_sad = fn_ptr->sdf(what->buf, what->stride,
get_buf_from_mv(in_what, ref_mv), in_what->stride) +
get_buf_from_mv(in_what, ref_mv), in_what->stride, 0x7fffffff) +
mvsad_err_cost(x, ref_mv, &fcenter_mv, sad_per_bit);
*best_mv = *ref_mv;
@ -1401,7 +1405,7 @@ int vp9_full_search_sadx8(const MACROBLOCK *x, const MV *ref_mv,
while (c < col_max) {
unsigned int sad = fn_ptr->sdf(what->buf, what->stride,
check_here, in_what->stride);
check_here, in_what->stride, best_sad);
if (sad < best_sad) {
const MV mv = {r, c};
sad += mvsad_err_cost(x, &mv, &fcenter_mv, sad_per_bit);
@ -1430,7 +1434,7 @@ int vp9_refining_search_sad_c(const MACROBLOCK *x,
const MV fcenter_mv = {center_mv->row >> 3, center_mv->col >> 3};
unsigned int best_sad = fn_ptr->sdf(what->buf, what->stride,
get_buf_from_mv(in_what, ref_mv),
in_what->stride) +
in_what->stride, 0x7fffffff) +
mvsad_err_cost(x, ref_mv, &fcenter_mv, error_per_bit);
int i, j;
@ -1442,7 +1446,7 @@ int vp9_refining_search_sad_c(const MACROBLOCK *x,
ref_mv->col + neighbors[j].col};
if (is_mv_in(x, &mv)) {
unsigned int sad = fn_ptr->sdf(what->buf, what->stride,
get_buf_from_mv(in_what, &mv), in_what->stride);
get_buf_from_mv(in_what, &mv), in_what->stride, best_sad);
if (sad < best_sad) {
sad += mvsad_err_cost(x, &mv, &fcenter_mv, error_per_bit);
if (sad < best_sad) {
@ -1475,7 +1479,7 @@ int vp9_refining_search_sadx4(const MACROBLOCK *x,
const MV fcenter_mv = {center_mv->row >> 3, center_mv->col >> 3};
const uint8_t *best_address = get_buf_from_mv(in_what, ref_mv);
unsigned int best_sad = fn_ptr->sdf(what->buf, what->stride, best_address,
in_what->stride) +
in_what->stride, 0x7fffffff) +
mvsad_err_cost(x, ref_mv, &fcenter_mv, error_per_bit);
int i, j;
@ -1516,7 +1520,7 @@ int vp9_refining_search_sadx4(const MACROBLOCK *x,
if (is_mv_in(x, &mv)) {
unsigned int sad = fn_ptr->sdf(what->buf, what->stride,
get_buf_from_mv(in_what, &mv),
in_what->stride);
in_what->stride, best_sad);
if (sad < best_sad) {
sad += mvsad_err_cost(x, &mv, &fcenter_mv, error_per_bit);
if (sad < best_sad) {
@ -1555,7 +1559,8 @@ int vp9_refining_search_8p_c(const MACROBLOCK *x,
const struct buf_2d *const in_what = &xd->plane[0].pre[0];
const MV fcenter_mv = {center_mv->row >> 3, center_mv->col >> 3};
unsigned int best_sad = fn_ptr->sdaf(what->buf, what->stride,
get_buf_from_mv(in_what, ref_mv), in_what->stride, second_pred) +
get_buf_from_mv(in_what, ref_mv), in_what->stride,
second_pred, 0x7fffffff) +
mvsad_err_cost(x, ref_mv, &fcenter_mv, error_per_bit);
int i, j;
@ -1568,7 +1573,8 @@ int vp9_refining_search_8p_c(const MACROBLOCK *x,
if (is_mv_in(x, &mv)) {
unsigned int sad = fn_ptr->sdaf(what->buf, what->stride,
get_buf_from_mv(in_what, &mv), in_what->stride, second_pred);
get_buf_from_mv(in_what, &mv), in_what->stride,
second_pred, best_sad);
if (sad < best_sad) {
sad += mvsad_err_cost(x, &mv, &fcenter_mv, error_per_bit);
if (sad < best_sad) {

3
vp9/encoder/vp9_rdopt.c
View File

@ -2124,7 +2124,8 @@ static void mv_pred(VP9_COMP *cpi, MACROBLOCK *x,
// Find sad for current vector.
this_sad = cpi->fn_ptr[block_size].sdf(src_y_ptr, x->plane[0].src.stride,
ref_y_ptr, ref_y_stride);
ref_y_ptr, ref_y_stride,
0x7fffffff);
// Note if it is the best so far.
if (this_sad < best_sad) {

12
vp9/encoder/vp9_sad.c
View File

@ -35,12 +35,14 @@ static INLINE unsigned int sad(const uint8_t *a, int a_stride,
#define sadMxN(m, n) \
unsigned int vp9_sad##m##x##n##_c(const uint8_t *src, int src_stride, \
const uint8_t *ref, int ref_stride) { \
const uint8_t *ref, int ref_stride, \
unsigned int max_sad) { \
return sad(src, src_stride, ref, ref_stride, m, n); \
} \
unsigned int vp9_sad##m##x##n##_avg_c(const uint8_t *src, int src_stride, \
const uint8_t *ref, int ref_stride, \
const uint8_t *second_pred) { \
const uint8_t *second_pred, \
unsigned int max_sad) { \
uint8_t comp_pred[m * n]; \
vp9_comp_avg_pred(comp_pred, second_pred, m, n, ref, ref_stride); \
return sad(src, src_stride, comp_pred, m, m, n); \
@ -52,7 +54,8 @@ void vp9_sad##m##x##n##x##k##_c(const uint8_t *src, int src_stride, \
unsigned int *sads) { \
int i; \
for (i = 0; i < k; ++i) \
sads[i] = vp9_sad##m##x##n##_c(src, src_stride, &ref[i], ref_stride); \
sads[i] = vp9_sad##m##x##n##_c(src, src_stride, &ref[i], ref_stride, \
0x7fffffff); \
}
#define sadMxNx4D(m, n) \
@ -61,7 +64,8 @@ void vp9_sad##m##x##n##x4d_c(const uint8_t *src, int src_stride, \
unsigned int *sads) { \
int i; \
for (i = 0; i < 4; ++i) \
sads[i] = vp9_sad##m##x##n##_c(src, src_stride, refs[i], ref_stride); \
sads[i] = vp9_sad##m##x##n##_c(src, src_stride, refs[i], ref_stride, \
0x7fffffff); \
}
// 64x64

6
vp9/encoder/vp9_variance.h
View File

@ -25,13 +25,15 @@ void variance(const uint8_t *a, int a_stride,
typedef unsigned int(*vp9_sad_fn_t)(const uint8_t *src_ptr,
int source_stride,
const uint8_t *ref_ptr,
int ref_stride);
int ref_stride,
unsigned int max_sad);
typedef unsigned int(*vp9_sad_avg_fn_t)(const uint8_t *src_ptr,
int source_stride,
const uint8_t *ref_ptr,
int ref_stride,
const uint8_t *second_pred);
const uint8_t *second_pred,
unsigned int max_sad);
typedef void (*vp9_sad_multi_fn_t)(const uint8_t *src_ptr,
int source_stride,