generic-library/vpx
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
6d8538e508
vpx/vp8/encoder/sad_c.c

426 lines
16 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>
WebM Experimental Codec Branch Snapshot This is a code snapshot of experimental work currently ongoing for a next-generation codec. The codebase has been cut down considerably from the libvpx baseline. For example, we are currently only supporting VBR 2-pass rate control and have removed most of the code relating to coding speed, threading, error resilience, partitions and various other features. This is in part to make the codebase easier to work on and experiment with, but also because we want to have an open discussion about how the bitstream will be structured and partitioned and not have that conversation constrained by past work. Our basic working pattern has been to initially encapsulate experiments using configure options linked to #IF CONFIG_XXX statements in the code. Once experiments have matured and we are reasonably happy that they give benefit and can be merged without breaking other experiments, we remove the conditional compile statements and merge them in. Current changes include: * Temporal coding experiment for segments (though still only 4 max, it will likely be increased). * Segment feature experiment - to allow various bits of information to be coded at the segment level. Features tested so far include mode and reference frame information, limiting end of block offset and transform size, alongside Q and loop filter parameters, but this set is very fluid. * Support for 8x8 transform - 8x8 dct with 2nd order 2x2 haar is used in MBs using 16x16 prediction modes within inter frames. * Compound prediction (combination of signals from existing predictors to create a new predictor). * 8 tap interpolation filters and 1/8th pel motion vectors. * Loop filter modifications. * Various entropy modifications and changes to how entropy contexts and updates are handled. * Extended quantizer range matched to transform precision improvements. There are also ongoing further experiments that we hope to merge in the near future: For example, coding of motion and other aspects of the prediction signal to better support larger image formats, use of larger block sizes (e.g. 32x32 and up) and lossless non-transform based coding options (especially for key frames). It is our hope that we will be able to make regular updates and we will warmly welcome community contributions. Please be warned that, at this stage, the codebase is currently slower than VP8 stable branch as most new code has not been optimized, and even the 'C' has been deliberately written to be simple and obvious, not fast. The following graphs have the initial test results, numbers in the tables measure the compression improvement in terms of percentage. The build has the following optional experiments configured: --enable-experimental --enable-enhanced_interp --enable-uvintra --enable-high_precision_mv --enable-sixteenth_subpel_uv CIF Size clips: http://getwebm.org/tmp/cif/ HD size clips: http://getwebm.org/tmp/hd/ (stable_20120309 represents encoding results of WebM master branch build as of commit#7a15907) They were encoded using the following encode parameters: --good --cpu-used=0 -t 0 --lag-in-frames=25 --min-q=0 --max-q=63 --end-usage=0 --auto-alt-ref=1 -p 2 --pass=2 --kf-max-dist=9999 --kf-min-dist=0 --drop-frame=0 --static-thresh=0 --bias-pct=50 --minsection-pct=0 --maxsection-pct=800 --sharpness=0 --arnr-maxframes=7 --arnr-strength=3(for HD,6 for CIF) --arnr-type=3 Change-Id: I5c62ed09cfff5815a2bb34e7820d6a810c23183c
2012-03-10 02:32:50 +01:00
#include "vpx_ports/config.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
static __inline
unsigned int sad_mx_n_c(
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
const unsigned char *src_ptr,
int src_stride,
const unsigned char *ref_ptr,
int ref_stride,
int m,
int n) {
int r, c;
unsigned int sad = 0;
for (r = 0; r < n; r++) {
for (c = 0; c < m; c++) {
sad += abs(src_ptr[c] - ref_ptr[c]);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
src_ptr += src_stride;
ref_ptr += ref_stride;
}
return sad;
Initial WebM release
2010-05-18 17:58:33 +02:00
}
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
unsigned int vp8_sad32x32_c(const unsigned char *src_ptr,
int src_stride,
const unsigned char *ref_ptr,
int ref_stride,
int max_sad) {
return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, 32, 32);
}
unsigned int vp8_sad16x16_c(const unsigned char *src_ptr,
int src_stride,
const unsigned char *ref_ptr,
int ref_stride,
int max_sad) {
return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, 16, 16);
}
Initial WebM release
2010-05-18 17:58:33 +02:00
unsigned int vp8_sad8x8_c(
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
const unsigned char *src_ptr,
int src_stride,
const unsigned char *ref_ptr,
int ref_stride,
int max_sad) {
return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, 8, 8);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
unsigned int vp8_sad16x8_c(
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
const unsigned char *src_ptr,
int src_stride,
const unsigned char *ref_ptr,
int ref_stride,
int max_sad) {
Initial WebM release
2010-05-18 17:58:33 +02:00
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, 16, 8);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
unsigned int vp8_sad8x16_c(
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
const unsigned char *src_ptr,
int src_stride,
const unsigned char *ref_ptr,
int ref_stride,
int max_sad) {
return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, 8, 16);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
unsigned int vp8_sad4x4_c(
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
const unsigned char *src_ptr,
int src_stride,
const unsigned char *ref_ptr,
int ref_stride,
int max_sad) {
return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, 4, 4);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
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
a new way of determining reference motion vector Using surrounding reconstructed pixels from left and above to select best matching mv to use as reference motion vector for mv encoding. Test results: AVGPSNR GLBPSNR VPXSSIM Derf: 1.107% 1.062% 0.992% Std-hd:1.209% 1.176% 1.029% Change-Id: I8f10e09ee6538c05df2fb9f069abcaf1edb3fca6
2012-08-06 19:51:20 +02:00
#if CONFIG_NEWBESTREFMV
Use 3 rows and cols of pixels for ref_mv scoring The commit changed to use 3 rows above and 3 cols from left for SAD scoring for selecting the best reference motion vector. The change helped std-hd set by >.2% on psnr/ssim metrics. Change-Id: Ifad3b528d0b4b6e3c22518af789d76eff23c1520
2012-09-14 17:58:02 +02:00
unsigned int vp8_sad3x16_c(
a new way of determining reference motion vector Using surrounding reconstructed pixels from left and above to select best matching mv to use as reference motion vector for mv encoding. Test results: AVGPSNR GLBPSNR VPXSSIM Derf: 1.107% 1.062% 0.992% Std-hd:1.209% 1.176% 1.029% Change-Id: I8f10e09ee6538c05df2fb9f069abcaf1edb3fca6
2012-08-06 19:51:20 +02:00
const unsigned char *src_ptr,
int src_stride,
const unsigned char *ref_ptr,
int ref_stride,
int max_sad){
Use 3 rows and cols of pixels for ref_mv scoring The commit changed to use 3 rows above and 3 cols from left for SAD scoring for selecting the best reference motion vector. The change helped std-hd set by >.2% on psnr/ssim metrics. Change-Id: Ifad3b528d0b4b6e3c22518af789d76eff23c1520
2012-09-14 17:58:02 +02:00
return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, 3, 16);
a new way of determining reference motion vector Using surrounding reconstructed pixels from left and above to select best matching mv to use as reference motion vector for mv encoding. Test results: AVGPSNR GLBPSNR VPXSSIM Derf: 1.107% 1.062% 0.992% Std-hd:1.209% 1.176% 1.029% Change-Id: I8f10e09ee6538c05df2fb9f069abcaf1edb3fca6
2012-08-06 19:51:20 +02:00
}
Use 3 rows and cols of pixels for ref_mv scoring The commit changed to use 3 rows above and 3 cols from left for SAD scoring for selecting the best reference motion vector. The change helped std-hd set by >.2% on psnr/ssim metrics. Change-Id: Ifad3b528d0b4b6e3c22518af789d76eff23c1520
2012-09-14 17:58:02 +02:00
unsigned int vp8_sad16x3_c(
a new way of determining reference motion vector Using surrounding reconstructed pixels from left and above to select best matching mv to use as reference motion vector for mv encoding. Test results: AVGPSNR GLBPSNR VPXSSIM Derf: 1.107% 1.062% 0.992% Std-hd:1.209% 1.176% 1.029% Change-Id: I8f10e09ee6538c05df2fb9f069abcaf1edb3fca6
2012-08-06 19:51:20 +02:00
const unsigned char *src_ptr,
int src_stride,
const unsigned char *ref_ptr,
int ref_stride,
int max_sad){
Use 3 rows and cols of pixels for ref_mv scoring The commit changed to use 3 rows above and 3 cols from left for SAD scoring for selecting the best reference motion vector. The change helped std-hd set by >.2% on psnr/ssim metrics. Change-Id: Ifad3b528d0b4b6e3c22518af789d76eff23c1520
2012-09-14 17:58:02 +02:00
return sad_mx_n_c(src_ptr, src_stride, ref_ptr, ref_stride, 16, 3);
a new way of determining reference motion vector Using surrounding reconstructed pixels from left and above to select best matching mv to use as reference motion vector for mv encoding. Test results: AVGPSNR GLBPSNR VPXSSIM Derf: 1.107% 1.062% 0.992% Std-hd:1.209% 1.176% 1.029% Change-Id: I8f10e09ee6538c05df2fb9f069abcaf1edb3fca6
2012-08-06 19:51:20 +02:00
}
#endif
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
void vp8_sad32x32x3_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_sad32x32_c(src_ptr, src_stride, ref_ptr, ref_stride, 0x7fffffff);
sad_array[1] = vp8_sad32x32_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, 0x7fffffff);
sad_array[2] = vp8_sad32x32_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, 0x7fffffff);
}
void vp8_sad32x32x8_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_sad32x32_c(src_ptr, src_stride, ref_ptr, ref_stride, 0x7fffffff);
sad_array[1] = (unsigned short)vp8_sad32x32_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, 0x7fffffff);
sad_array[2] = (unsigned short)vp8_sad32x32_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, 0x7fffffff);
sad_array[3] = (unsigned short)vp8_sad32x32_c(src_ptr, src_stride, ref_ptr + 3, ref_stride, 0x7fffffff);
sad_array[4] = (unsigned short)vp8_sad32x32_c(src_ptr, src_stride, ref_ptr + 4, ref_stride, 0x7fffffff);
sad_array[5] = (unsigned short)vp8_sad32x32_c(src_ptr, src_stride, ref_ptr + 5, ref_stride, 0x7fffffff);
sad_array[6] = (unsigned short)vp8_sad32x32_c(src_ptr, src_stride, ref_ptr + 6, ref_stride, 0x7fffffff);
sad_array[7] = (unsigned short)vp8_sad32x32_c(src_ptr, src_stride, ref_ptr + 7, ref_stride, 0x7fffffff);
}
Initial WebM release
2010-05-18 17:58:33 +02:00
void vp8_sad16x16x3_c(
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
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, ref_stride, 0x7fffffff);
sad_array[1] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, 0x7fffffff);
sad_array[2] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, 0x7fffffff);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
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
void vp8_sad16x16x8_c(
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
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, ref_stride, 0x7fffffff);
sad_array[1] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, 0x7fffffff);
sad_array[2] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, 0x7fffffff);
sad_array[3] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 3, ref_stride, 0x7fffffff);
sad_array[4] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 4, ref_stride, 0x7fffffff);
sad_array[5] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 5, ref_stride, 0x7fffffff);
sad_array[6] = (unsigned short)vp8_sad16x16_c(src_ptr, src_stride, ref_ptr + 6, ref_stride, 0x7fffffff);
sad_array[7] = (unsigned short)vp8_sad16x16_c(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
}
Initial WebM release
2010-05-18 17:58:33 +02:00
void vp8_sad16x8x3_c(
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
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, ref_stride, 0x7fffffff);
sad_array[1] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, 0x7fffffff);
sad_array[2] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, 0x7fffffff);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
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
void vp8_sad16x8x8_c(
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
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, ref_stride, 0x7fffffff);
sad_array[1] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, 0x7fffffff);
sad_array[2] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, 0x7fffffff);
sad_array[3] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 3, ref_stride, 0x7fffffff);
sad_array[4] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 4, ref_stride, 0x7fffffff);
sad_array[5] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 5, ref_stride, 0x7fffffff);
sad_array[6] = (unsigned short)vp8_sad16x8_c(src_ptr, src_stride, ref_ptr + 6, ref_stride, 0x7fffffff);
sad_array[7] = (unsigned short)vp8_sad16x8_c(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
}
Initial WebM release
2010-05-18 17:58:33 +02:00
void vp8_sad8x8x3_c(
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
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, ref_stride, 0x7fffffff);
sad_array[1] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, 0x7fffffff);
sad_array[2] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, 0x7fffffff);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
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
void vp8_sad8x8x8_c(
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
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, ref_stride, 0x7fffffff);
sad_array[1] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, 0x7fffffff);
sad_array[2] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, 0x7fffffff);
sad_array[3] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 3, ref_stride, 0x7fffffff);
sad_array[4] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 4, ref_stride, 0x7fffffff);
sad_array[5] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 5, ref_stride, 0x7fffffff);
sad_array[6] = (unsigned short)vp8_sad8x8_c(src_ptr, src_stride, ref_ptr + 6, ref_stride, 0x7fffffff);
sad_array[7] = (unsigned short)vp8_sad8x8_c(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
}
Initial WebM release
2010-05-18 17:58:33 +02:00
void vp8_sad8x16x3_c(
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
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, ref_stride, 0x7fffffff);
sad_array[1] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, 0x7fffffff);
sad_array[2] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, 0x7fffffff);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
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
void vp8_sad8x16x8_c(
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
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, ref_stride, 0x7fffffff);
sad_array[1] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, 0x7fffffff);
sad_array[2] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, 0x7fffffff);
sad_array[3] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 3, ref_stride, 0x7fffffff);
sad_array[4] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 4, ref_stride, 0x7fffffff);
sad_array[5] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 5, ref_stride, 0x7fffffff);
sad_array[6] = (unsigned short)vp8_sad8x16_c(src_ptr, src_stride, ref_ptr + 6, ref_stride, 0x7fffffff);
sad_array[7] = (unsigned short)vp8_sad8x16_c(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
}
Initial WebM release
2010-05-18 17:58:33 +02:00
void vp8_sad4x4x3_c(
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
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, ref_stride, 0x7fffffff);
sad_array[1] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, 0x7fffffff);
sad_array[2] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, 0x7fffffff);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
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
void vp8_sad4x4x8_c(
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
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, ref_stride, 0x7fffffff);
sad_array[1] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 1, ref_stride, 0x7fffffff);
sad_array[2] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 2, ref_stride, 0x7fffffff);
sad_array[3] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 3, ref_stride, 0x7fffffff);
sad_array[4] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 4, ref_stride, 0x7fffffff);
sad_array[5] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 5, ref_stride, 0x7fffffff);
sad_array[6] = (unsigned short)vp8_sad4x4_c(src_ptr, src_stride, ref_ptr + 6, ref_stride, 0x7fffffff);
sad_array[7] = (unsigned short)vp8_sad4x4_c(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
}
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
void vp8_sad32x32x4d_c(const unsigned char *src_ptr,
int src_stride,
unsigned char *ref_ptr[],
int ref_stride,
unsigned int *sad_array
) {
sad_array[0] = vp8_sad32x32_c(src_ptr, src_stride, ref_ptr[0], ref_stride, 0x7fffffff);
sad_array[1] = vp8_sad32x32_c(src_ptr, src_stride, ref_ptr[1], ref_stride, 0x7fffffff);
sad_array[2] = vp8_sad32x32_c(src_ptr, src_stride, ref_ptr[2], ref_stride, 0x7fffffff);
sad_array[3] = vp8_sad32x32_c(src_ptr, src_stride, ref_ptr[3], ref_stride, 0x7fffffff);
}
Initial WebM release
2010-05-18 17:58:33 +02:00
void vp8_sad16x16x4d_c(
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
const unsigned char *src_ptr,
int src_stride,
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, 0x7fffffff);
sad_array[1] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr[1], ref_stride, 0x7fffffff);
sad_array[2] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr[2], ref_stride, 0x7fffffff);
sad_array[3] = vp8_sad16x16_c(src_ptr, src_stride, ref_ptr[3], ref_stride, 0x7fffffff);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
void vp8_sad16x8x4d_c(
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
const unsigned char *src_ptr,
int src_stride,
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, 0x7fffffff);
sad_array[1] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr[1], ref_stride, 0x7fffffff);
sad_array[2] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr[2], ref_stride, 0x7fffffff);
sad_array[3] = vp8_sad16x8_c(src_ptr, src_stride, ref_ptr[3], ref_stride, 0x7fffffff);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
void vp8_sad8x8x4d_c(
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
const unsigned char *src_ptr,
int src_stride,
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, 0x7fffffff);
sad_array[1] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr[1], ref_stride, 0x7fffffff);
sad_array[2] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr[2], ref_stride, 0x7fffffff);
sad_array[3] = vp8_sad8x8_c(src_ptr, src_stride, ref_ptr[3], ref_stride, 0x7fffffff);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
void vp8_sad8x16x4d_c(
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
const unsigned char *src_ptr,
int src_stride,
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, 0x7fffffff);
sad_array[1] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr[1], ref_stride, 0x7fffffff);
sad_array[2] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr[2], ref_stride, 0x7fffffff);
sad_array[3] = vp8_sad8x16_c(src_ptr, src_stride, ref_ptr[3], ref_stride, 0x7fffffff);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
void vp8_sad4x4x4d_c(
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
const unsigned char *src_ptr,
int src_stride,
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, 0x7fffffff);
sad_array[1] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr[1], ref_stride, 0x7fffffff);
sad_array[2] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr[2], ref_stride, 0x7fffffff);
sad_array[3] = vp8_sad4x4_c(src_ptr, src_stride, ref_ptr[3], ref_stride, 0x7fffffff);
Initial WebM release
2010-05-18 17:58:33 +02:00
}
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
/* Copy 2 macroblocks to a buffer */
void vp8_copy32xn_c(
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
unsigned char *src_ptr,
int src_stride,
unsigned char *dst_ptr,
int dst_stride,
int height) {
int r;
for (r = 0; r < height; r++) {
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
#if !(CONFIG_FAST_UNALIGNED)
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
dst_ptr[0] = src_ptr[0];
dst_ptr[1] = src_ptr[1];
dst_ptr[2] = src_ptr[2];
dst_ptr[3] = src_ptr[3];
dst_ptr[4] = src_ptr[4];
dst_ptr[5] = src_ptr[5];
dst_ptr[6] = src_ptr[6];
dst_ptr[7] = src_ptr[7];
dst_ptr[8] = src_ptr[8];
dst_ptr[9] = src_ptr[9];
dst_ptr[10] = src_ptr[10];
dst_ptr[11] = src_ptr[11];
dst_ptr[12] = src_ptr[12];
dst_ptr[13] = src_ptr[13];
dst_ptr[14] = src_ptr[14];
dst_ptr[15] = src_ptr[15];
dst_ptr[16] = src_ptr[16];
dst_ptr[17] = src_ptr[17];
dst_ptr[18] = src_ptr[18];
dst_ptr[19] = src_ptr[19];
dst_ptr[20] = src_ptr[20];
dst_ptr[21] = src_ptr[21];
dst_ptr[22] = src_ptr[22];
dst_ptr[23] = src_ptr[23];
dst_ptr[24] = src_ptr[24];
dst_ptr[25] = src_ptr[25];
dst_ptr[26] = src_ptr[26];
dst_ptr[27] = src_ptr[27];
dst_ptr[28] = src_ptr[28];
dst_ptr[29] = src_ptr[29];
dst_ptr[30] = src_ptr[30];
dst_ptr[31] = src_ptr[31];
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
#else
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
((uint32_t *)dst_ptr)[0] = ((uint32_t *)src_ptr)[0];
((uint32_t *)dst_ptr)[1] = ((uint32_t *)src_ptr)[1];
((uint32_t *)dst_ptr)[2] = ((uint32_t *)src_ptr)[2];
((uint32_t *)dst_ptr)[3] = ((uint32_t *)src_ptr)[3];
((uint32_t *)dst_ptr)[4] = ((uint32_t *)src_ptr)[4];
((uint32_t *)dst_ptr)[5] = ((uint32_t *)src_ptr)[5];
((uint32_t *)dst_ptr)[6] = ((uint32_t *)src_ptr)[6];
((uint32_t *)dst_ptr)[7] = ((uint32_t *)src_ptr)[7];
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
#endif
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
src_ptr += src_stride;
dst_ptr += dst_stride;
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
Restyle code Approximate the Google style guide[1] so that that there's a written document to follow and tools to check compliance[2]. [1]: http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml [2]: http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py Change-Id: Idf40e3d8dddcc72150f6af127b13e5dab838685f
2012-07-14 00:21:29 +02:00
}
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
}
Reference in New Issue Copy Permalink