generic-library/vpx
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
b5e6d6cccf
vpx/vp9/encoder/vp9_sad_c.c

616 lines
27 KiB
C
Raw Normal View History

Initial WebM release
2010-05-18 17:58:33 +02:00
/*
Use WebM in copyright notice for consistency Changes 'The VP8 project' to 'The WebM project', for consistency with other webmproject.org repositories. Fixes issue #97. Change-Id: I37c13ed5fbdb9d334ceef71c6350e9febed9bbba
2010-09-09 14:16:39 +02:00
* Copyright (c) 2010 The WebM project authors. All Rights Reserved.
Initial WebM release
2010-05-18 17:58:33 +02:00
*
cosmetics: trim trailing whitespace When the license headers were updated, they accidentally contained trailing whitespace, so unfortunately we have to touch all the files again. Change-Id: I236c05fade06589e417179c0444cb39b09e4200d
2010-06-18 18:39:21 +02:00
* Use of this source code is governed by a BSD-style license
LICENSE: update with latest text Change-Id: Ieebea089095d9073b3a94932791099f614ce120c
2010-06-04 22:19:40 +02:00
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
cosmetics: trim trailing whitespace When the license headers were updated, they accidentally contained trailing whitespace, so unfortunately we have to touch all the files again. Change-Id: I236c05fade06589e417179c0444cb39b09e4200d
2010-06-18 18:39:21 +02:00
* in the file PATENTS. All contributing project authors may
LICENSE: update with latest text Change-Id: Ieebea089095d9073b3a94932791099f614ce120c
2010-06-04 22:19:40 +02:00
* be found in the AUTHORS file in the root of the source tree.
Initial WebM release
2010-05-18 17:58:33 +02:00
*/
#include <stdlib.h>
Move vp9_sadmxn.h from common to encoder Change-Id: I6f6ba91b1b8b280902b171472314d665aa0baf0b
2013-11-19 21:46:08 +01:00
#include "./vp9_rtcd.h"
Build fixes to merge vp9-preview into master Various fixups to resolve issues when building vp9-preview under the more stringent checks placed on the experimental branch. Change-Id: I21749de83552e1e75c799003f849e6a0f1a35b07
2012-12-23 16:20:10 +01:00
#include "./vpx_config.h"
Move vp9_sadmxn.h from common to encoder Change-Id: I6f6ba91b1b8b280902b171472314d665aa0baf0b
2013-11-19 21:46:08 +01:00
#include "vp9/encoder/vp9_sadmxn.h"
#include "vp9/encoder/vp9_variance.h"
Preload reference area to an intermediate buffer in sub-pixel motion search In sub-pixel motion search, the search range is small(+/- 3 pixels). Preload whole search area from reference buffer into a 32-byte aligned buffer. Then in search, load reference data from this buffer instead. This keeps data in cache, and reduces the crossing cache- line penalty. For tulip clip, tests on Intel Core2 Quad machine(linux) showed encoder speed improvement: 3.4% at --rt --cpu-used =-4 2.8% at --rt --cpu-used =-3 2.3% at --rt --cpu-used =-2 2.2% at --rt --cpu-used =-1 Test on Atom notebook showed only 1.1% speed improvement(speed=-4). Test on Xeon machine also showed less improvement, since unaligned data access latency is greatly reduced in newer cores. Next, I will apply similar idea to other 2 sub-pixel search functions for encoding speed > 4. Make this change exclusively for x86 platforms. Change-Id: Ia7bb9f56169eac0f01009fe2b2f2ab5b61d2eb2f
2011-06-28 15:14:13 +02:00
#include "vpx/vpx_integer.h"
Initial WebM release
2010-05-18 17:58:33 +02:00
Add averaging-SAD functions for 8-point comp-inter motion search. Makes first 50 frames of bus @ 1500kbps encode from 3min22.7 to 3min18.2, i.e. 2.3% faster. In addition, use the sub_pixel_avg functions to calc the variance of the averaging predictor. This is slightly suboptimal because the function is subpixel-position-aware, but it will (at least for the SSE2 version) not actually use a bilinear filter for a full-pixel position, thus leading to approximately the same performance compared to if we implemented an actual average-aware full-pixel variance function. That gains another 0.3 seconds (i.e. encode time goes to 3min17.4), thus leading to a total gain of 2.7%. Change-Id: I3f059d2b04243921868cfed2568d4fa65d7b5acd
2013-06-25 20:26:49 +02:00
#define sad_mxn_func(m, n) \
unsigned int vp9_sad##m##x##n##_c(const uint8_t *src_ptr, \
int src_stride, \
const uint8_t *ref_ptr, \
int ref_stride, \
unsigned int max_sad) { \
return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, m, n); \
} \
unsigned int vp9_sad##m##x##n##_avg_c(const uint8_t *src_ptr, \
int src_stride, \
const uint8_t *ref_ptr, \
int ref_stride, \
const uint8_t *second_pred, \
unsigned int max_sad) { \
uint8_t comp_pred[m * n]; \
comp_avg_pred(comp_pred, second_pred, m, n, ref_ptr, ref_stride); \
return sad_mx_n_c(src_ptr, src_stride, comp_pred, m, m, n); \
}
sad_mxn_func(64, 64)
sad_mxn_func(64, 32)
sad_mxn_func(32, 64)
sad_mxn_func(32, 32)
sad_mxn_func(32, 16)
sad_mxn_func(16, 32)
sad_mxn_func(16, 16)
sad_mxn_func(16, 8)
sad_mxn_func(8, 16)
sad_mxn_func(8, 8)
sad_mxn_func(8, 4)
sad_mxn_func(4, 8)
sad_mxn_func(4, 4)
Add rectangular block size variance/sad functions. With this, the RD loop properly supports rectangular blocks. Change-Id: Iece79048fb4e84741ee1ada982da129a7bf00470
2013-04-15 19:00:34 +02:00
void vp9_sad64x32x4d_c(const uint8_t *src_ptr,
int src_stride,
const uint8_t* const ref_ptr[],
int ref_stride,
unsigned int *sad_array) {
sad_array[0] = vp9_sad64x32(src_ptr, src_stride,
ref_ptr[0], ref_stride, 0x7fffffff);
sad_array[1] = vp9_sad64x32(src_ptr, src_stride,
ref_ptr[1], ref_stride, 0x7fffffff);
sad_array[2] = vp9_sad64x32(src_ptr, src_stride,
ref_ptr[2], ref_stride, 0x7fffffff);
sad_array[3] = vp9_sad64x32(src_ptr, src_stride,
ref_ptr[3], ref_stride, 0x7fffffff);
}
void vp9_sad32x64x4d_c(const uint8_t *src_ptr,
int src_stride,
const uint8_t* const ref_ptr[],
int ref_stride,
unsigned int *sad_array) {
sad_array[0] = vp9_sad32x64(src_ptr, src_stride,
ref_ptr[0], ref_stride, 0x7fffffff);
sad_array[1] = vp9_sad32x64(src_ptr, src_stride,
ref_ptr[1], ref_stride, 0x7fffffff);
sad_array[2] = vp9_sad32x64(src_ptr, src_stride,
ref_ptr[2], ref_stride, 0x7fffffff);
sad_array[3] = vp9_sad32x64(src_ptr, src_stride,
ref_ptr[3], ref_stride, 0x7fffffff);
}
void vp9_sad32x16x4d_c(const uint8_t *src_ptr,
int src_stride,
const uint8_t* const ref_ptr[],
int ref_stride,
unsigned int *sad_array) {
sad_array[0] = vp9_sad32x16(src_ptr, src_stride,
ref_ptr[0], ref_stride, 0x7fffffff);
sad_array[1] = vp9_sad32x16(src_ptr, src_stride,
ref_ptr[1], ref_stride, 0x7fffffff);
sad_array[2] = vp9_sad32x16(src_ptr, src_stride,
ref_ptr[2], ref_stride, 0x7fffffff);
sad_array[3] = vp9_sad32x16(src_ptr, src_stride,
ref_ptr[3], ref_stride, 0x7fffffff);
}
void vp9_sad16x32x4d_c(const uint8_t *src_ptr,
int src_stride,
const uint8_t* const ref_ptr[],
int ref_stride,
unsigned int *sad_array) {
sad_array[0] = vp9_sad16x32(src_ptr, src_stride,
ref_ptr[0], ref_stride, 0x7fffffff);
sad_array[1] = vp9_sad16x32(src_ptr, src_stride,
ref_ptr[1], ref_stride, 0x7fffffff);
sad_array[2] = vp9_sad16x32(src_ptr, src_stride,
ref_ptr[2], ref_stride, 0x7fffffff);
sad_array[3] = vp9_sad16x32(src_ptr, src_stride,
ref_ptr[3], ref_stride, 0x7fffffff);
}
64x64 blocksize support. 3.2% gains on std/hd, 1.0% gains on hd. Change-Id: I481d5df23d8a4fc650a5bcba956554490b2bd200
2013-01-06 03:20:25 +01:00
void vp9_sad64x64x3_c(const uint8_t *src_ptr,
int src_stride,
const uint8_t *ref_ptr,
int ref_stride,
unsigned int *sad_array) {
fix to parameters to match rtcd Change-Id: I919e2dd72292fe44f2e53ada56bd42287d50cdeb Signed-off-by: Jim Bankoski <jimbankoski@google.com>
2013-02-28 17:10:08 +01:00
sad_array[0] = vp9_sad64x64(src_ptr, src_stride, ref_ptr, ref_stride,
0x7fffffff);
sad_array[1] = vp9_sad64x64(src_ptr, src_stride, ref_ptr + 1, ref_stride,
0x7fffffff);
sad_array[2] = vp9_sad64x64(src_ptr, src_stride, ref_ptr + 2, ref_stride,
0x7fffffff);
64x64 blocksize support. 3.2% gains on std/hd, 1.0% gains on hd. Change-Id: I481d5df23d8a4fc650a5bcba956554490b2bd200
2013-01-06 03:20:25 +01:00
}
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
void vp9_sad32x32x3_c(const uint8_t *src_ptr,
Superblock coding. This commit adds a pick_sb_mode() function which selects the best 32x32 superblock coding mode. Then it selects the best per-MB modes, compares the two and encodes that in the bitstream. The bitstream coding is rather simplistic right now. At the SB level, we code a bit to indicate whether this block uses SB-coding (32x32 prediction) or MB-coding (anything else), and then we follow with the actual modes. This could and should be modified in the future, but is omitted from this commit because it will likely involve reorganizing much more code rather than just adding SB coding, so it's better to let that be judged on its own merits. Gains on derf: about even, YT/HD: +0.75%, STD/HD: +1.5%. Change-Id: Iae313a7cbd8f75b3c66d04a68b991cb096eaaba6
2012-08-20 23:43:34 +02:00
int src_stride,
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
const uint8_t *ref_ptr,
Superblock coding. This commit adds a pick_sb_mode() function which selects the best 32x32 superblock coding mode. Then it selects the best per-MB modes, compares the two and encodes that in the bitstream. The bitstream coding is rather simplistic right now. At the SB level, we code a bit to indicate whether this block uses SB-coding (32x32 prediction) or MB-coding (anything else), and then we follow with the actual modes. This could and should be modified in the future, but is omitted from this commit because it will likely involve reorganizing much more code rather than just adding SB coding, so it's better to let that be judged on its own merits. Gains on derf: about even, YT/HD: +0.75%, STD/HD: +1.5%. Change-Id: Iae313a7cbd8f75b3c66d04a68b991cb096eaaba6
2012-08-20 23:43:34 +02:00
int ref_stride,
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
unsigned int *sad_array) {
fix to parameters to match rtcd Change-Id: I919e2dd72292fe44f2e53ada56bd42287d50cdeb Signed-off-by: Jim Bankoski <jimbankoski@google.com>
2013-02-28 17:10:08 +01:00
sad_array[0] = vp9_sad32x32(src_ptr, src_stride,
ref_ptr, ref_stride, 0x7fffffff);
sad_array[1] = vp9_sad32x32(src_ptr, src_stride,
ref_ptr + 1, ref_stride, 0x7fffffff);
sad_array[2] = vp9_sad32x32(src_ptr, src_stride,
ref_ptr + 2, ref_stride, 0x7fffffff);
Superblock coding. This commit adds a pick_sb_mode() function which selects the best 32x32 superblock coding mode. Then it selects the best per-MB modes, compares the two and encodes that in the bitstream. The bitstream coding is rather simplistic right now. At the SB level, we code a bit to indicate whether this block uses SB-coding (32x32 prediction) or MB-coding (anything else), and then we follow with the actual modes. This could and should be modified in the future, but is omitted from this commit because it will likely involve reorganizing much more code rather than just adding SB coding, so it's better to let that be judged on its own merits. Gains on derf: about even, YT/HD: +0.75%, STD/HD: +1.5%. Change-Id: Iae313a7cbd8f75b3c66d04a68b991cb096eaaba6
2012-08-20 23:43:34 +02:00
}
64x64 blocksize support. 3.2% gains on std/hd, 1.0% gains on hd. Change-Id: I481d5df23d8a4fc650a5bcba956554490b2bd200
2013-01-06 03:20:25 +01:00
void vp9_sad64x64x8_c(const uint8_t *src_ptr,
int src_stride,
const uint8_t *ref_ptr,
int ref_stride,
this commit converts all sad ptrs to uint32 sse4_1 code used uint16_t for returning sad, but that won't work for 32x32 or 64x64. This code fixes the assembly for those and also reenables sse4_1 on linux Change-Id: I5ce7288d581db870a148e5f7c5092826f59edd81
2013-02-28 17:32:14 +01:00
unsigned int *sad_array) {
sad_array[0] = vp9_sad64x64(src_ptr, src_stride,
ref_ptr, ref_stride,
0x7fffffff);
sad_array[1] = vp9_sad64x64(src_ptr, src_stride,
ref_ptr + 1, ref_stride,
0x7fffffff);
sad_array[2] = vp9_sad64x64(src_ptr, src_stride,
ref_ptr + 2, ref_stride,
0x7fffffff);
sad_array[3] = vp9_sad64x64(src_ptr, src_stride,
ref_ptr + 3, ref_stride,
0x7fffffff);
sad_array[4] = vp9_sad64x64(src_ptr, src_stride,
ref_ptr + 4, ref_stride,
0x7fffffff);
sad_array[5] = vp9_sad64x64(src_ptr, src_stride,
ref_ptr + 5, ref_stride,
0x7fffffff);
sad_array[6] = vp9_sad64x64(src_ptr, src_stride,
ref_ptr + 6, ref_stride,
0x7fffffff);
sad_array[7] = vp9_sad64x64(src_ptr, src_stride,
ref_ptr + 7, ref_stride,
0x7fffffff);
64x64 blocksize support. 3.2% gains on std/hd, 1.0% gains on hd. Change-Id: I481d5df23d8a4fc650a5bcba956554490b2bd200
2013-01-06 03:20:25 +01:00
}
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
void vp9_sad32x32x8_c(const uint8_t *src_ptr,
Superblock coding. This commit adds a pick_sb_mode() function which selects the best 32x32 superblock coding mode. Then it selects the best per-MB modes, compares the two and encodes that in the bitstream. The bitstream coding is rather simplistic right now. At the SB level, we code a bit to indicate whether this block uses SB-coding (32x32 prediction) or MB-coding (anything else), and then we follow with the actual modes. This could and should be modified in the future, but is omitted from this commit because it will likely involve reorganizing much more code rather than just adding SB coding, so it's better to let that be judged on its own merits. Gains on derf: about even, YT/HD: +0.75%, STD/HD: +1.5%. Change-Id: Iae313a7cbd8f75b3c66d04a68b991cb096eaaba6
2012-08-20 23:43:34 +02:00
int src_stride,
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
const uint8_t *ref_ptr,
Superblock coding. This commit adds a pick_sb_mode() function which selects the best 32x32 superblock coding mode. Then it selects the best per-MB modes, compares the two and encodes that in the bitstream. The bitstream coding is rather simplistic right now. At the SB level, we code a bit to indicate whether this block uses SB-coding (32x32 prediction) or MB-coding (anything else), and then we follow with the actual modes. This could and should be modified in the future, but is omitted from this commit because it will likely involve reorganizing much more code rather than just adding SB coding, so it's better to let that be judged on its own merits. Gains on derf: about even, YT/HD: +0.75%, STD/HD: +1.5%. Change-Id: Iae313a7cbd8f75b3c66d04a68b991cb096eaaba6
2012-08-20 23:43:34 +02:00
int ref_stride,
this commit converts all sad ptrs to uint32 sse4_1 code used uint16_t for returning sad, but that won't work for 32x32 or 64x64. This code fixes the assembly for those and also reenables sse4_1 on linux Change-Id: I5ce7288d581db870a148e5f7c5092826f59edd81
2013-02-28 17:32:14 +01:00
unsigned int *sad_array) {
sad_array[0] = vp9_sad32x32(src_ptr, src_stride,
ref_ptr, ref_stride,
0x7fffffff);
sad_array[1] = vp9_sad32x32(src_ptr, src_stride,
ref_ptr + 1, ref_stride,
0x7fffffff);
sad_array[2] = vp9_sad32x32(src_ptr, src_stride,
ref_ptr + 2, ref_stride,
0x7fffffff);
sad_array[3] = vp9_sad32x32(src_ptr, src_stride,
ref_ptr + 3, ref_stride,
0x7fffffff);
sad_array[4] = vp9_sad32x32(src_ptr, src_stride,
ref_ptr + 4, ref_stride,
0x7fffffff);
sad_array[5] = vp9_sad32x32(src_ptr, src_stride,
ref_ptr + 5, ref_stride,
0x7fffffff);
sad_array[6] = vp9_sad32x32(src_ptr, src_stride,
ref_ptr + 6, ref_stride,
0x7fffffff);
sad_array[7] = vp9_sad32x32(src_ptr, src_stride,
ref_ptr + 7, ref_stride,
0x7fffffff);
Superblock coding. This commit adds a pick_sb_mode() function which selects the best 32x32 superblock coding mode. Then it selects the best per-MB modes, compares the two and encodes that in the bitstream. The bitstream coding is rather simplistic right now. At the SB level, we code a bit to indicate whether this block uses SB-coding (32x32 prediction) or MB-coding (anything else), and then we follow with the actual modes. This could and should be modified in the future, but is omitted from this commit because it will likely involve reorganizing much more code rather than just adding SB coding, so it's better to let that be judged on its own merits. Gains on derf: about even, YT/HD: +0.75%, STD/HD: +1.5%. Change-Id: Iae313a7cbd8f75b3c66d04a68b991cb096eaaba6
2012-08-20 23:43:34 +02:00
}
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
void vp9_sad16x16x3_c(const uint8_t *src_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int src_stride,
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
const uint8_t *ref_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int ref_stride,
unsigned int *sad_array) {
fix to parameters to match rtcd Change-Id: I919e2dd72292fe44f2e53ada56bd42287d50cdeb Signed-off-by: Jim Bankoski <jimbankoski@google.com>
2013-02-28 17:10:08 +01:00
sad_array[0] = vp9_sad16x16(src_ptr, src_stride,
ref_ptr, ref_stride, 0x7fffffff);
sad_array[1] = vp9_sad16x16(src_ptr, src_stride,
ref_ptr + 1, ref_stride, 0x7fffffff);
sad_array[2] = vp9_sad16x16(src_ptr, src_stride,
ref_ptr + 2, ref_stride, 0x7fffffff);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
void vp9_sad16x16x8_c(const uint8_t *src_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int src_stride,
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
const uint8_t *ref_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int ref_stride,
this commit converts all sad ptrs to uint32 sse4_1 code used uint16_t for returning sad, but that won't work for 32x32 or 64x64. This code fixes the assembly for those and also reenables sse4_1 on linux Change-Id: I5ce7288d581db870a148e5f7c5092826f59edd81
2013-02-28 17:32:14 +01:00
uint32_t *sad_array) {
sad_array[0] = vp9_sad16x16(src_ptr, src_stride,
ref_ptr, ref_stride,
0x7fffffff);
sad_array[1] = vp9_sad16x16(src_ptr, src_stride,
ref_ptr + 1, ref_stride,
0x7fffffff);
sad_array[2] = vp9_sad16x16(src_ptr, src_stride,
ref_ptr + 2, ref_stride,
0x7fffffff);
sad_array[3] = vp9_sad16x16(src_ptr, src_stride,
ref_ptr + 3, ref_stride,
0x7fffffff);
sad_array[4] = vp9_sad16x16(src_ptr, src_stride,
ref_ptr + 4, ref_stride,
0x7fffffff);
sad_array[5] = vp9_sad16x16(src_ptr, src_stride,
ref_ptr + 5, ref_stride,
0x7fffffff);
sad_array[6] = vp9_sad16x16(src_ptr, src_stride,
ref_ptr + 6, ref_stride,
0x7fffffff);
sad_array[7] = vp9_sad16x16(src_ptr, src_stride,
ref_ptr + 7, ref_stride,
0x7fffffff);
Full search SAD function optimization in SSE4.1 Use mpsadbw, and calculate 8 sad at once. Function list: vp8_sad16x16x8_sse4 vp8_sad16x8x8_sse4 vp8_sad8x16x8_sse4 vp8_sad8x8x8_sse4 vp8_sad4x4x8_sse4 (test clip: tulip) For best quality mode, this gave encoder a 5% performance boost. For good quality mode with speed=1, this gave encoder a 3% performance boost. Change-Id: I083b5a39d39144f88dcbccbef95da6498e490134
2010-10-27 14:45:24 +02:00
}
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
void vp9_sad16x8x3_c(const uint8_t *src_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int src_stride,
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
const uint8_t *ref_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int ref_stride,
unsigned int *sad_array) {
fix to parameters to match rtcd Change-Id: I919e2dd72292fe44f2e53ada56bd42287d50cdeb Signed-off-by: Jim Bankoski <jimbankoski@google.com>
2013-02-28 17:10:08 +01:00
sad_array[0] = vp9_sad16x8(src_ptr, src_stride,
ref_ptr, ref_stride, 0x7fffffff);
sad_array[1] = vp9_sad16x8(src_ptr, src_stride,
ref_ptr + 1, ref_stride, 0x7fffffff);
sad_array[2] = vp9_sad16x8(src_ptr, src_stride,
ref_ptr + 2, ref_stride, 0x7fffffff);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
void vp9_sad16x8x8_c(const uint8_t *src_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int src_stride,
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
const uint8_t *ref_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int ref_stride,
this commit converts all sad ptrs to uint32 sse4_1 code used uint16_t for returning sad, but that won't work for 32x32 or 64x64. This code fixes the assembly for those and also reenables sse4_1 on linux Change-Id: I5ce7288d581db870a148e5f7c5092826f59edd81
2013-02-28 17:32:14 +01:00
uint32_t *sad_array) {
sad_array[0] = vp9_sad16x8(src_ptr, src_stride,
ref_ptr, ref_stride,
0x7fffffff);
sad_array[1] = vp9_sad16x8(src_ptr, src_stride,
ref_ptr + 1, ref_stride,
0x7fffffff);
sad_array[2] = vp9_sad16x8(src_ptr, src_stride,
ref_ptr + 2, ref_stride,
0x7fffffff);
sad_array[3] = vp9_sad16x8(src_ptr, src_stride,
ref_ptr + 3, ref_stride,
0x7fffffff);
sad_array[4] = vp9_sad16x8(src_ptr, src_stride,
ref_ptr + 4, ref_stride,
0x7fffffff);
sad_array[5] = vp9_sad16x8(src_ptr, src_stride,
ref_ptr + 5, ref_stride,
0x7fffffff);
sad_array[6] = vp9_sad16x8(src_ptr, src_stride,
ref_ptr + 6, ref_stride,
0x7fffffff);
sad_array[7] = vp9_sad16x8(src_ptr, src_stride,
ref_ptr + 7, ref_stride,
0x7fffffff);
Full search SAD function optimization in SSE4.1 Use mpsadbw, and calculate 8 sad at once. Function list: vp8_sad16x16x8_sse4 vp8_sad16x8x8_sse4 vp8_sad8x16x8_sse4 vp8_sad8x8x8_sse4 vp8_sad4x4x8_sse4 (test clip: tulip) For best quality mode, this gave encoder a 5% performance boost. For good quality mode with speed=1, this gave encoder a 3% performance boost. Change-Id: I083b5a39d39144f88dcbccbef95da6498e490134
2010-10-27 14:45:24 +02:00
}
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
void vp9_sad8x8x3_c(const uint8_t *src_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int src_stride,
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
const uint8_t *ref_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int ref_stride,
unsigned int *sad_array) {
fix to parameters to match rtcd Change-Id: I919e2dd72292fe44f2e53ada56bd42287d50cdeb Signed-off-by: Jim Bankoski <jimbankoski@google.com>
2013-02-28 17:10:08 +01:00
sad_array[0] = vp9_sad8x8(src_ptr, src_stride,
ref_ptr, ref_stride, 0x7fffffff);
sad_array[1] = vp9_sad8x8(src_ptr, src_stride,
ref_ptr + 1, ref_stride, 0x7fffffff);
sad_array[2] = vp9_sad8x8(src_ptr, src_stride,
ref_ptr + 2, ref_stride, 0x7fffffff);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
void vp9_sad8x8x8_c(const uint8_t *src_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int src_stride,
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
const uint8_t *ref_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int ref_stride,
this commit converts all sad ptrs to uint32 sse4_1 code used uint16_t for returning sad, but that won't work for 32x32 or 64x64. This code fixes the assembly for those and also reenables sse4_1 on linux Change-Id: I5ce7288d581db870a148e5f7c5092826f59edd81
2013-02-28 17:32:14 +01:00
uint32_t *sad_array) {
sad_array[0] = vp9_sad8x8(src_ptr, src_stride,
ref_ptr, ref_stride,
0x7fffffff);
sad_array[1] = vp9_sad8x8(src_ptr, src_stride,
ref_ptr + 1, ref_stride,
0x7fffffff);
sad_array[2] = vp9_sad8x8(src_ptr, src_stride,
ref_ptr + 2, ref_stride,
0x7fffffff);
sad_array[3] = vp9_sad8x8(src_ptr, src_stride,
ref_ptr + 3, ref_stride,
0x7fffffff);
sad_array[4] = vp9_sad8x8(src_ptr, src_stride,
ref_ptr + 4, ref_stride,
0x7fffffff);
sad_array[5] = vp9_sad8x8(src_ptr, src_stride,
ref_ptr + 5, ref_stride,
0x7fffffff);
sad_array[6] = vp9_sad8x8(src_ptr, src_stride,
ref_ptr + 6, ref_stride,
0x7fffffff);
sad_array[7] = vp9_sad8x8(src_ptr, src_stride,
ref_ptr + 7, ref_stride,
0x7fffffff);
Full search SAD function optimization in SSE4.1 Use mpsadbw, and calculate 8 sad at once. Function list: vp8_sad16x16x8_sse4 vp8_sad16x8x8_sse4 vp8_sad8x16x8_sse4 vp8_sad8x8x8_sse4 vp8_sad4x4x8_sse4 (test clip: tulip) For best quality mode, this gave encoder a 5% performance boost. For good quality mode with speed=1, this gave encoder a 3% performance boost. Change-Id: I083b5a39d39144f88dcbccbef95da6498e490134
2010-10-27 14:45:24 +02:00
}
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
void vp9_sad8x16x3_c(const uint8_t *src_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int src_stride,
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
const uint8_t *ref_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int ref_stride,
unsigned int *sad_array) {
fix to parameters to match rtcd Change-Id: I919e2dd72292fe44f2e53ada56bd42287d50cdeb Signed-off-by: Jim Bankoski <jimbankoski@google.com>
2013-02-28 17:10:08 +01:00
sad_array[0] = vp9_sad8x16(src_ptr, src_stride,
ref_ptr, ref_stride, 0x7fffffff);
sad_array[1] = vp9_sad8x16(src_ptr, src_stride,
ref_ptr + 1, ref_stride, 0x7fffffff);
sad_array[2] = vp9_sad8x16(src_ptr, src_stride,
ref_ptr + 2, ref_stride, 0x7fffffff);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
void vp9_sad8x16x8_c(const uint8_t *src_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int src_stride,
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
const uint8_t *ref_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int ref_stride,
this commit converts all sad ptrs to uint32 sse4_1 code used uint16_t for returning sad, but that won't work for 32x32 or 64x64. This code fixes the assembly for those and also reenables sse4_1 on linux Change-Id: I5ce7288d581db870a148e5f7c5092826f59edd81
2013-02-28 17:32:14 +01:00
uint32_t *sad_array) {
sad_array[0] = vp9_sad8x16(src_ptr, src_stride,
ref_ptr, ref_stride,
0x7fffffff);
sad_array[1] = vp9_sad8x16(src_ptr, src_stride,
ref_ptr + 1, ref_stride,
0x7fffffff);
sad_array[2] = vp9_sad8x16(src_ptr, src_stride,
ref_ptr + 2, ref_stride,
0x7fffffff);
sad_array[3] = vp9_sad8x16(src_ptr, src_stride,
ref_ptr + 3, ref_stride,
0x7fffffff);
sad_array[4] = vp9_sad8x16(src_ptr, src_stride,
ref_ptr + 4, ref_stride,
0x7fffffff);
sad_array[5] = vp9_sad8x16(src_ptr, src_stride,
ref_ptr + 5, ref_stride,
0x7fffffff);
sad_array[6] = vp9_sad8x16(src_ptr, src_stride,
ref_ptr + 6, ref_stride,
0x7fffffff);
sad_array[7] = vp9_sad8x16(src_ptr, src_stride,
ref_ptr + 7, ref_stride,
0x7fffffff);
Full search SAD function optimization in SSE4.1 Use mpsadbw, and calculate 8 sad at once. Function list: vp8_sad16x16x8_sse4 vp8_sad16x8x8_sse4 vp8_sad8x16x8_sse4 vp8_sad8x8x8_sse4 vp8_sad4x4x8_sse4 (test clip: tulip) For best quality mode, this gave encoder a 5% performance boost. For good quality mode with speed=1, this gave encoder a 3% performance boost. Change-Id: I083b5a39d39144f88dcbccbef95da6498e490134
2010-10-27 14:45:24 +02:00
}
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
void vp9_sad4x4x3_c(const uint8_t *src_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int src_stride,
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
const uint8_t *ref_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int ref_stride,
unsigned int *sad_array) {
fix to parameters to match rtcd Change-Id: I919e2dd72292fe44f2e53ada56bd42287d50cdeb Signed-off-by: Jim Bankoski <jimbankoski@google.com>
2013-02-28 17:10:08 +01:00
sad_array[0] = vp9_sad4x4(src_ptr, src_stride,
ref_ptr, ref_stride, 0x7fffffff);
sad_array[1] = vp9_sad4x4(src_ptr, src_stride,
ref_ptr + 1, ref_stride, 0x7fffffff);
sad_array[2] = vp9_sad4x4(src_ptr, src_stride,
ref_ptr + 2, ref_stride, 0x7fffffff);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
void vp9_sad4x4x8_c(const uint8_t *src_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int src_stride,
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
const uint8_t *ref_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int ref_stride,
this commit converts all sad ptrs to uint32 sse4_1 code used uint16_t for returning sad, but that won't work for 32x32 or 64x64. This code fixes the assembly for those and also reenables sse4_1 on linux Change-Id: I5ce7288d581db870a148e5f7c5092826f59edd81
2013-02-28 17:32:14 +01:00
uint32_t *sad_array) {
sad_array[0] = vp9_sad4x4(src_ptr, src_stride,
ref_ptr, ref_stride,
0x7fffffff);
sad_array[1] = vp9_sad4x4(src_ptr, src_stride,
ref_ptr + 1, ref_stride,
0x7fffffff);
sad_array[2] = vp9_sad4x4(src_ptr, src_stride,
ref_ptr + 2, ref_stride,
0x7fffffff);
sad_array[3] = vp9_sad4x4(src_ptr, src_stride,
ref_ptr + 3, ref_stride,
0x7fffffff);
sad_array[4] = vp9_sad4x4(src_ptr, src_stride,
ref_ptr + 4, ref_stride,
0x7fffffff);
sad_array[5] = vp9_sad4x4(src_ptr, src_stride,
ref_ptr + 5, ref_stride,
0x7fffffff);
sad_array[6] = vp9_sad4x4(src_ptr, src_stride,
ref_ptr + 6, ref_stride,
0x7fffffff);
sad_array[7] = vp9_sad4x4(src_ptr, src_stride,
ref_ptr + 7, ref_stride,
0x7fffffff);
Full search SAD function optimization in SSE4.1 Use mpsadbw, and calculate 8 sad at once. Function list: vp8_sad16x16x8_sse4 vp8_sad16x8x8_sse4 vp8_sad8x16x8_sse4 vp8_sad8x8x8_sse4 vp8_sad4x4x8_sse4 (test clip: tulip) For best quality mode, this gave encoder a 5% performance boost. For good quality mode with speed=1, this gave encoder a 3% performance boost. Change-Id: I083b5a39d39144f88dcbccbef95da6498e490134
2010-10-27 14:45:24 +02:00
}
64x64 blocksize support. 3.2% gains on std/hd, 1.0% gains on hd. Change-Id: I481d5df23d8a4fc650a5bcba956554490b2bd200
2013-01-06 03:20:25 +01:00
void vp9_sad64x64x4d_c(const uint8_t *src_ptr,
int src_stride,
Add unit test for x4 multi-SAD functions Update the function prototypes to match between VP9 and VP8. Change-Id: If58965073989e87df3b62b67a030ec6ce23ca04f
2013-03-01 21:43:41 +01:00
const uint8_t* const ref_ptr[],
64x64 blocksize support. 3.2% gains on std/hd, 1.0% gains on hd. Change-Id: I481d5df23d8a4fc650a5bcba956554490b2bd200
2013-01-06 03:20:25 +01:00
int ref_stride,
unsigned int *sad_array) {
fix to parameters to match rtcd Change-Id: I919e2dd72292fe44f2e53ada56bd42287d50cdeb Signed-off-by: Jim Bankoski <jimbankoski@google.com>
2013-02-28 17:10:08 +01:00
sad_array[0] = vp9_sad64x64(src_ptr, src_stride,
ref_ptr[0], ref_stride, 0x7fffffff);
sad_array[1] = vp9_sad64x64(src_ptr, src_stride,
ref_ptr[1], ref_stride, 0x7fffffff);
sad_array[2] = vp9_sad64x64(src_ptr, src_stride,
ref_ptr[2], ref_stride, 0x7fffffff);
sad_array[3] = vp9_sad64x64(src_ptr, src_stride,
ref_ptr[3], ref_stride, 0x7fffffff);
64x64 blocksize support. 3.2% gains on std/hd, 1.0% gains on hd. Change-Id: I481d5df23d8a4fc650a5bcba956554490b2bd200
2013-01-06 03:20:25 +01:00
}
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
void vp9_sad32x32x4d_c(const uint8_t *src_ptr,
Superblock coding. This commit adds a pick_sb_mode() function which selects the best 32x32 superblock coding mode. Then it selects the best per-MB modes, compares the two and encodes that in the bitstream. The bitstream coding is rather simplistic right now. At the SB level, we code a bit to indicate whether this block uses SB-coding (32x32 prediction) or MB-coding (anything else), and then we follow with the actual modes. This could and should be modified in the future, but is omitted from this commit because it will likely involve reorganizing much more code rather than just adding SB coding, so it's better to let that be judged on its own merits. Gains on derf: about even, YT/HD: +0.75%, STD/HD: +1.5%. Change-Id: Iae313a7cbd8f75b3c66d04a68b991cb096eaaba6
2012-08-20 23:43:34 +02:00
int src_stride,
Add unit test for x4 multi-SAD functions Update the function prototypes to match between VP9 and VP8. Change-Id: If58965073989e87df3b62b67a030ec6ce23ca04f
2013-03-01 21:43:41 +01:00
const uint8_t* const ref_ptr[],
Superblock coding. This commit adds a pick_sb_mode() function which selects the best 32x32 superblock coding mode. Then it selects the best per-MB modes, compares the two and encodes that in the bitstream. The bitstream coding is rather simplistic right now. At the SB level, we code a bit to indicate whether this block uses SB-coding (32x32 prediction) or MB-coding (anything else), and then we follow with the actual modes. This could and should be modified in the future, but is omitted from this commit because it will likely involve reorganizing much more code rather than just adding SB coding, so it's better to let that be judged on its own merits. Gains on derf: about even, YT/HD: +0.75%, STD/HD: +1.5%. Change-Id: Iae313a7cbd8f75b3c66d04a68b991cb096eaaba6
2012-08-20 23:43:34 +02:00
int ref_stride,
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
unsigned int *sad_array) {
fix to parameters to match rtcd Change-Id: I919e2dd72292fe44f2e53ada56bd42287d50cdeb Signed-off-by: Jim Bankoski <jimbankoski@google.com>
2013-02-28 17:10:08 +01:00
sad_array[0] = vp9_sad32x32(src_ptr, src_stride,
ref_ptr[0], ref_stride, 0x7fffffff);
sad_array[1] = vp9_sad32x32(src_ptr, src_stride,
ref_ptr[1], ref_stride, 0x7fffffff);
sad_array[2] = vp9_sad32x32(src_ptr, src_stride,
ref_ptr[2], ref_stride, 0x7fffffff);
sad_array[3] = vp9_sad32x32(src_ptr, src_stride,
ref_ptr[3], ref_stride, 0x7fffffff);
Superblock coding. This commit adds a pick_sb_mode() function which selects the best 32x32 superblock coding mode. Then it selects the best per-MB modes, compares the two and encodes that in the bitstream. The bitstream coding is rather simplistic right now. At the SB level, we code a bit to indicate whether this block uses SB-coding (32x32 prediction) or MB-coding (anything else), and then we follow with the actual modes. This could and should be modified in the future, but is omitted from this commit because it will likely involve reorganizing much more code rather than just adding SB coding, so it's better to let that be judged on its own merits. Gains on derf: about even, YT/HD: +0.75%, STD/HD: +1.5%. Change-Id: Iae313a7cbd8f75b3c66d04a68b991cb096eaaba6
2012-08-20 23:43:34 +02:00
}
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
void vp9_sad16x16x4d_c(const uint8_t *src_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int src_stride,
Add unit test for x4 multi-SAD functions Update the function prototypes to match between VP9 and VP8. Change-Id: If58965073989e87df3b62b67a030ec6ce23ca04f
2013-03-01 21:43:41 +01:00
const uint8_t* const ref_ptr[],
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int ref_stride,
unsigned int *sad_array) {
fix to parameters to match rtcd Change-Id: I919e2dd72292fe44f2e53ada56bd42287d50cdeb Signed-off-by: Jim Bankoski <jimbankoski@google.com>
2013-02-28 17:10:08 +01:00
sad_array[0] = vp9_sad16x16(src_ptr, src_stride,
ref_ptr[0], ref_stride, 0x7fffffff);
sad_array[1] = vp9_sad16x16(src_ptr, src_stride,
ref_ptr[1], ref_stride, 0x7fffffff);
sad_array[2] = vp9_sad16x16(src_ptr, src_stride,
ref_ptr[2], ref_stride, 0x7fffffff);
sad_array[3] = vp9_sad16x16(src_ptr, src_stride,
ref_ptr[3], ref_stride, 0x7fffffff);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
void vp9_sad16x8x4d_c(const uint8_t *src_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int src_stride,
Add unit test for x4 multi-SAD functions Update the function prototypes to match between VP9 and VP8. Change-Id: If58965073989e87df3b62b67a030ec6ce23ca04f
2013-03-01 21:43:41 +01:00
const uint8_t* const ref_ptr[],
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int ref_stride,
unsigned int *sad_array) {
fix to parameters to match rtcd Change-Id: I919e2dd72292fe44f2e53ada56bd42287d50cdeb Signed-off-by: Jim Bankoski <jimbankoski@google.com>
2013-02-28 17:10:08 +01:00
sad_array[0] = vp9_sad16x8(src_ptr, src_stride,
ref_ptr[0], ref_stride, 0x7fffffff);
sad_array[1] = vp9_sad16x8(src_ptr, src_stride,
ref_ptr[1], ref_stride, 0x7fffffff);
sad_array[2] = vp9_sad16x8(src_ptr, src_stride,
ref_ptr[2], ref_stride, 0x7fffffff);
sad_array[3] = vp9_sad16x8(src_ptr, src_stride,
ref_ptr[3], ref_stride, 0x7fffffff);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
void vp9_sad8x8x4d_c(const uint8_t *src_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int src_stride,
Add unit test for x4 multi-SAD functions Update the function prototypes to match between VP9 and VP8. Change-Id: If58965073989e87df3b62b67a030ec6ce23ca04f
2013-03-01 21:43:41 +01:00
const uint8_t* const ref_ptr[],
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int ref_stride,
unsigned int *sad_array) {
fix to parameters to match rtcd Change-Id: I919e2dd72292fe44f2e53ada56bd42287d50cdeb Signed-off-by: Jim Bankoski <jimbankoski@google.com>
2013-02-28 17:10:08 +01:00
sad_array[0] = vp9_sad8x8(src_ptr, src_stride,
ref_ptr[0], ref_stride, 0x7fffffff);
sad_array[1] = vp9_sad8x8(src_ptr, src_stride,
ref_ptr[1], ref_stride, 0x7fffffff);
sad_array[2] = vp9_sad8x8(src_ptr, src_stride,
ref_ptr[2], ref_stride, 0x7fffffff);
sad_array[3] = vp9_sad8x8(src_ptr, src_stride,
ref_ptr[3], ref_stride, 0x7fffffff);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
void vp9_sad8x16x4d_c(const uint8_t *src_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int src_stride,
Add unit test for x4 multi-SAD functions Update the function prototypes to match between VP9 and VP8. Change-Id: If58965073989e87df3b62b67a030ec6ce23ca04f
2013-03-01 21:43:41 +01:00
const uint8_t* const ref_ptr[],
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int ref_stride,
unsigned int *sad_array) {
fix to parameters to match rtcd Change-Id: I919e2dd72292fe44f2e53ada56bd42287d50cdeb Signed-off-by: Jim Bankoski <jimbankoski@google.com>
2013-02-28 17:10:08 +01:00
sad_array[0] = vp9_sad8x16(src_ptr, src_stride,
ref_ptr[0], ref_stride, 0x7fffffff);
sad_array[1] = vp9_sad8x16(src_ptr, src_stride,
ref_ptr[1], ref_stride, 0x7fffffff);
sad_array[2] = vp9_sad8x16(src_ptr, src_stride,
ref_ptr[2], ref_stride, 0x7fffffff);
sad_array[3] = vp9_sad8x16(src_ptr, src_stride,
ref_ptr[3], ref_stride, 0x7fffffff);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
Add building blocks for 4x8/8x4 rd search These building blocks enable rate-distortion optimization search over block sizes of 8x4 and 4x8. Need to convert them into mmx/sse forms. Change-Id: I570ea2d22d14ceec3fe3575128d7dfa172a577de
2013-05-15 21:19:59 +02:00
void vp9_sad8x4x4d_c(const uint8_t *src_ptr,
int src_stride,
const uint8_t* const ref_ptr[],
int ref_stride,
unsigned int *sad_array) {
sad_array[0] = vp9_sad8x4(src_ptr, src_stride,
ref_ptr[0], ref_stride, 0x7fffffff);
sad_array[1] = vp9_sad8x4(src_ptr, src_stride,
ref_ptr[1], ref_stride, 0x7fffffff);
sad_array[2] = vp9_sad8x4(src_ptr, src_stride,
ref_ptr[2], ref_stride, 0x7fffffff);
sad_array[3] = vp9_sad8x4(src_ptr, src_stride,
ref_ptr[3], ref_stride, 0x7fffffff);
}
Merge 4x4 block level partition into codebase Move 4x4/4x8/8x4 partition coding out of experimental list. This commit fixed the unit test failure issues. It also resolved the merge conflicts between 4x4 block level partition and iterative motion search for comp_inter_inter. Change-Id: I898671f0631f5ddc4f5cc68d4c62ead7de9c5a58
2013-05-22 06:28:42 +02:00
void vp9_sad8x4x8_c(const uint8_t *src_ptr,
int src_stride,
const uint8_t *ref_ptr,
int ref_stride,
uint32_t *sad_array) {
sad_array[0] = vp9_sad8x4(src_ptr, src_stride,
ref_ptr, ref_stride,
0x7fffffff);
sad_array[1] = vp9_sad8x4(src_ptr, src_stride,
ref_ptr + 1, ref_stride,
0x7fffffff);
sad_array[2] = vp9_sad8x4(src_ptr, src_stride,
ref_ptr + 2, ref_stride,
0x7fffffff);
sad_array[3] = vp9_sad8x4(src_ptr, src_stride,
ref_ptr + 3, ref_stride,
0x7fffffff);
sad_array[4] = vp9_sad8x4(src_ptr, src_stride,
ref_ptr + 4, ref_stride,
0x7fffffff);
sad_array[5] = vp9_sad8x4(src_ptr, src_stride,
ref_ptr + 5, ref_stride,
0x7fffffff);
sad_array[6] = vp9_sad8x4(src_ptr, src_stride,
ref_ptr + 6, ref_stride,
0x7fffffff);
sad_array[7] = vp9_sad8x4(src_ptr, src_stride,
ref_ptr + 7, ref_stride,
0x7fffffff);
}
Add building blocks for 4x8/8x4 rd search These building blocks enable rate-distortion optimization search over block sizes of 8x4 and 4x8. Need to convert them into mmx/sse forms. Change-Id: I570ea2d22d14ceec3fe3575128d7dfa172a577de
2013-05-15 21:19:59 +02:00
void vp9_sad4x8x4d_c(const uint8_t *src_ptr,
int src_stride,
const uint8_t* const ref_ptr[],
int ref_stride,
unsigned int *sad_array) {
sad_array[0] = vp9_sad4x8(src_ptr, src_stride,
ref_ptr[0], ref_stride, 0x7fffffff);
sad_array[1] = vp9_sad4x8(src_ptr, src_stride,
ref_ptr[1], ref_stride, 0x7fffffff);
sad_array[2] = vp9_sad4x8(src_ptr, src_stride,
ref_ptr[2], ref_stride, 0x7fffffff);
sad_array[3] = vp9_sad4x8(src_ptr, src_stride,
ref_ptr[3], ref_stride, 0x7fffffff);
}
Merge 4x4 block level partition into codebase Move 4x4/4x8/8x4 partition coding out of experimental list. This commit fixed the unit test failure issues. It also resolved the merge conflicts between 4x4 block level partition and iterative motion search for comp_inter_inter. Change-Id: I898671f0631f5ddc4f5cc68d4c62ead7de9c5a58
2013-05-22 06:28:42 +02:00
void vp9_sad4x8x8_c(const uint8_t *src_ptr,
int src_stride,
const uint8_t *ref_ptr,
int ref_stride,
uint32_t *sad_array) {
sad_array[0] = vp9_sad4x8(src_ptr, src_stride,
ref_ptr, ref_stride,
0x7fffffff);
sad_array[1] = vp9_sad4x8(src_ptr, src_stride,
ref_ptr + 1, ref_stride,
0x7fffffff);
sad_array[2] = vp9_sad4x8(src_ptr, src_stride,
ref_ptr + 2, ref_stride,
0x7fffffff);
sad_array[3] = vp9_sad4x8(src_ptr, src_stride,
ref_ptr + 3, ref_stride,
0x7fffffff);
sad_array[4] = vp9_sad4x8(src_ptr, src_stride,
ref_ptr + 4, ref_stride,
0x7fffffff);
sad_array[5] = vp9_sad4x8(src_ptr, src_stride,
ref_ptr + 5, ref_stride,
0x7fffffff);
sad_array[6] = vp9_sad4x8(src_ptr, src_stride,
ref_ptr + 6, ref_stride,
0x7fffffff);
sad_array[7] = vp9_sad4x8(src_ptr, src_stride,
ref_ptr + 7, ref_stride,
0x7fffffff);
}
Use standard integer types for pixel values and coefficients. For coefficients, use int16_t (instead of short); for pixel values in 16-bit intermediates, use uint16_t (instead of unsigned short); for all others, use uint8_t (instead of unsigned char). Change-Id: I3619cd9abf106c3742eccc2e2f5e89a62774f7da
2012-12-19 00:31:19 +01:00
void vp9_sad4x4x4d_c(const uint8_t *src_ptr,
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int src_stride,
Add unit test for x4 multi-SAD functions Update the function prototypes to match between VP9 and VP8. Change-Id: If58965073989e87df3b62b67a030ec6ce23ca04f
2013-03-01 21:43:41 +01:00
const uint8_t* const ref_ptr[],
Adjust style to match Google Coding Style a little more closely. Most of these were picked up in the previous commit (prefix change from vp8_ to vp9_), but I'm pushing this separately so that it's easier to review. Change-Id: I91e959895778b8632d7d33375523df8a7568a490
2012-10-30 22:25:33 +01:00
int ref_stride,
unsigned int *sad_array) {
fix to parameters to match rtcd Change-Id: I919e2dd72292fe44f2e53ada56bd42287d50cdeb Signed-off-by: Jim Bankoski <jimbankoski@google.com>
2013-02-28 17:10:08 +01:00
sad_array[0] = vp9_sad4x4(src_ptr, src_stride,
ref_ptr[0], ref_stride, 0x7fffffff);
sad_array[1] = vp9_sad4x4(src_ptr, src_stride,
ref_ptr[1], ref_stride, 0x7fffffff);
sad_array[2] = vp9_sad4x4(src_ptr, src_stride,
ref_ptr[2], ref_stride, 0x7fffffff);
sad_array[3] = vp9_sad4x4(src_ptr, src_stride,
ref_ptr[3], ref_stride, 0x7fffffff);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
Reference in New Issue Copy Permalink