generic-library/vpx
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
d75eb73653
vpx/vp8/encoder/x86/variance_x86.h

336 lines
11 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
*/
#ifndef VARIANCE_X86_H
#define VARIANCE_X86_H
/* Note:
*
* This platform is commonly built for runtime CPU detection. If you modify
* any of the function mappings present in this file, be sure to also update
* them in the function pointer initialization code
*/
#if HAVE_MMX
extern prototype_sad(vp8_sad4x4_mmx);
extern prototype_sad(vp8_sad8x8_mmx);
extern prototype_sad(vp8_sad8x16_mmx);
extern prototype_sad(vp8_sad16x8_mmx);
extern prototype_sad(vp8_sad16x16_mmx);
extern prototype_variance(vp8_variance4x4_mmx);
extern prototype_variance(vp8_variance8x8_mmx);
extern prototype_variance(vp8_variance8x16_mmx);
extern prototype_variance(vp8_variance16x8_mmx);
extern prototype_variance(vp8_variance16x16_mmx);
extern prototype_subpixvariance(vp8_sub_pixel_variance4x4_mmx);
extern prototype_subpixvariance(vp8_sub_pixel_variance8x8_mmx);
extern prototype_subpixvariance(vp8_sub_pixel_variance8x16_mmx);
extern prototype_subpixvariance(vp8_sub_pixel_variance16x8_mmx);
extern prototype_subpixvariance(vp8_sub_pixel_variance16x16_mmx);
Fix half-pixel variance RTCD functions This patch fixes the system dependent entries for the half-pixel variance functions in both the RTCD and non-RTCD cases: - The generic C versions of these functions are now correct. Before all three cases called the hv code. - Wire up the ARM functions in RTCD mode - Created stubs for x86 to call the optimized subpixel functions with the correct parameters, rather than falling back to C code. Change-Id: I1d937d074d929e0eb93aacb1232cc5e0ad1c6184
2010-10-27 17:28:43 +02:00
extern prototype_variance(vp8_variance_halfpixvar16x16_h_mmx);
extern prototype_variance(vp8_variance_halfpixvar16x16_v_mmx);
extern prototype_variance(vp8_variance_halfpixvar16x16_hv_mmx);
Initial WebM release
2010-05-18 17:58:33 +02:00
extern prototype_subpixvariance(vp8_sub_pixel_mse16x16_mmx);
extern prototype_getmbss(vp8_get_mb_ss_mmx);
extern prototype_variance(vp8_mse16x16_mmx);
extern prototype_sad(vp8_get16x16pred_error_mmx);
extern prototype_variance2(vp8_get8x8var_mmx);
extern prototype_variance2(vp8_get16x16var_mmx);
extern prototype_sad(vp8_get4x4sse_cs_mmx);
#if !CONFIG_RUNTIME_CPU_DETECT
#undef vp8_variance_sad4x4
#define vp8_variance_sad4x4 vp8_sad4x4_mmx
#undef vp8_variance_sad8x8
#define vp8_variance_sad8x8 vp8_sad8x8_mmx
#undef vp8_variance_sad8x16
#define vp8_variance_sad8x16 vp8_sad8x16_mmx
#undef vp8_variance_sad16x8
#define vp8_variance_sad16x8 vp8_sad16x8_mmx
#undef vp8_variance_sad16x16
#define vp8_variance_sad16x16 vp8_sad16x16_mmx
#undef vp8_variance_var4x4
#define vp8_variance_var4x4 vp8_variance4x4_mmx
#undef vp8_variance_var8x8
#define vp8_variance_var8x8 vp8_variance8x8_mmx
#undef vp8_variance_var8x16
#define vp8_variance_var8x16 vp8_variance8x16_mmx
#undef vp8_variance_var16x8
#define vp8_variance_var16x8 vp8_variance16x8_mmx
#undef vp8_variance_var16x16
#define vp8_variance_var16x16 vp8_variance16x16_mmx
#undef vp8_variance_subpixvar4x4
#define vp8_variance_subpixvar4x4 vp8_sub_pixel_variance4x4_mmx
#undef vp8_variance_subpixvar8x8
#define vp8_variance_subpixvar8x8 vp8_sub_pixel_variance8x8_mmx
#undef vp8_variance_subpixvar8x16
#define vp8_variance_subpixvar8x16 vp8_sub_pixel_variance8x16_mmx
#undef vp8_variance_subpixvar16x8
#define vp8_variance_subpixvar16x8 vp8_sub_pixel_variance16x8_mmx
#undef vp8_variance_subpixvar16x16
#define vp8_variance_subpixvar16x16 vp8_sub_pixel_variance16x16_mmx
Fix half-pixel variance RTCD functions This patch fixes the system dependent entries for the half-pixel variance functions in both the RTCD and non-RTCD cases: - The generic C versions of these functions are now correct. Before all three cases called the hv code. - Wire up the ARM functions in RTCD mode - Created stubs for x86 to call the optimized subpixel functions with the correct parameters, rather than falling back to C code. Change-Id: I1d937d074d929e0eb93aacb1232cc5e0ad1c6184
2010-10-27 17:28:43 +02:00
#undef vp8_variance_halfpixvar16x16_h
#define vp8_variance_halfpixvar16x16_h vp8_variance_halfpixvar16x16_h_mmx
#undef vp8_variance_halfpixvar16x16_v
#define vp8_variance_halfpixvar16x16_v vp8_variance_halfpixvar16x16_v_mmx
#undef vp8_variance_halfpixvar16x16_hv
#define vp8_variance_halfpixvar16x16_hv vp8_variance_halfpixvar16x16_hv_mmx
Initial WebM release
2010-05-18 17:58:33 +02:00
#undef vp8_variance_subpixmse16x16
#define vp8_variance_subpixmse16x16 vp8_sub_pixel_mse16x16_mmx
#undef vp8_variance_getmbss
#define vp8_variance_getmbss vp8_get_mb_ss_mmx
#undef vp8_variance_mse16x16
#define vp8_variance_mse16x16 vp8_mse16x16_mmx
#undef vp8_variance_get16x16prederror
#define vp8_variance_get16x16prederror vp8_get16x16pred_error_mmx
#undef vp8_variance_get8x8var
#define vp8_variance_get8x8var vp8_get8x8var_mmx
#undef vp8_variance_get16x16var
#define vp8_variance_get16x16var vp8_get16x16var_mmx
#undef vp8_variance_get4x4sse_cs
#define vp8_variance_get4x4sse_cs vp8_get4x4sse_cs_mmx
#endif
#endif
#if HAVE_SSE2
extern prototype_sad(vp8_sad4x4_wmt);
extern prototype_sad(vp8_sad8x8_wmt);
extern prototype_sad(vp8_sad8x16_wmt);
extern prototype_sad(vp8_sad16x8_wmt);
extern prototype_sad(vp8_sad16x16_wmt);
extern prototype_variance(vp8_variance4x4_wmt);
extern prototype_variance(vp8_variance8x8_wmt);
extern prototype_variance(vp8_variance8x16_wmt);
extern prototype_variance(vp8_variance16x8_wmt);
extern prototype_variance(vp8_variance16x16_wmt);
extern prototype_subpixvariance(vp8_sub_pixel_variance4x4_wmt);
extern prototype_subpixvariance(vp8_sub_pixel_variance8x8_wmt);
extern prototype_subpixvariance(vp8_sub_pixel_variance8x16_wmt);
extern prototype_subpixvariance(vp8_sub_pixel_variance16x8_wmt);
extern prototype_subpixvariance(vp8_sub_pixel_variance16x16_wmt);
Fix half-pixel variance RTCD functions This patch fixes the system dependent entries for the half-pixel variance functions in both the RTCD and non-RTCD cases: - The generic C versions of these functions are now correct. Before all three cases called the hv code. - Wire up the ARM functions in RTCD mode - Created stubs for x86 to call the optimized subpixel functions with the correct parameters, rather than falling back to C code. Change-Id: I1d937d074d929e0eb93aacb1232cc5e0ad1c6184
2010-10-27 17:28:43 +02:00
extern prototype_variance(vp8_variance_halfpixvar16x16_h_wmt);
extern prototype_variance(vp8_variance_halfpixvar16x16_v_wmt);
extern prototype_variance(vp8_variance_halfpixvar16x16_hv_wmt);
Initial WebM release
2010-05-18 17:58:33 +02:00
extern prototype_subpixvariance(vp8_sub_pixel_mse16x16_wmt);
extern prototype_getmbss(vp8_get_mb_ss_sse2);
extern prototype_variance(vp8_mse16x16_wmt);
extern prototype_sad(vp8_get16x16pred_error_sse2);
extern prototype_variance2(vp8_get8x8var_sse2);
extern prototype_variance2(vp8_get16x16var_sse2);
#if !CONFIG_RUNTIME_CPU_DETECT
#undef vp8_variance_sad4x4
#define vp8_variance_sad4x4 vp8_sad4x4_wmt
#undef vp8_variance_sad8x8
#define vp8_variance_sad8x8 vp8_sad8x8_wmt
#undef vp8_variance_sad8x16
#define vp8_variance_sad8x16 vp8_sad8x16_wmt
#undef vp8_variance_sad16x8
#define vp8_variance_sad16x8 vp8_sad16x8_wmt
#undef vp8_variance_sad16x16
#define vp8_variance_sad16x16 vp8_sad16x16_wmt
#undef vp8_variance_var4x4
#define vp8_variance_var4x4 vp8_variance4x4_wmt
#undef vp8_variance_var8x8
#define vp8_variance_var8x8 vp8_variance8x8_wmt
#undef vp8_variance_var8x16
#define vp8_variance_var8x16 vp8_variance8x16_wmt
#undef vp8_variance_var16x8
#define vp8_variance_var16x8 vp8_variance16x8_wmt
#undef vp8_variance_var16x16
#define vp8_variance_var16x16 vp8_variance16x16_wmt
#undef vp8_variance_subpixvar4x4
#define vp8_variance_subpixvar4x4 vp8_sub_pixel_variance4x4_wmt
#undef vp8_variance_subpixvar8x8
#define vp8_variance_subpixvar8x8 vp8_sub_pixel_variance8x8_wmt
#undef vp8_variance_subpixvar8x16
#define vp8_variance_subpixvar8x16 vp8_sub_pixel_variance8x16_wmt
#undef vp8_variance_subpixvar16x8
#define vp8_variance_subpixvar16x8 vp8_sub_pixel_variance16x8_wmt
#undef vp8_variance_subpixvar16x16
#define vp8_variance_subpixvar16x16 vp8_sub_pixel_variance16x16_wmt
Fix half-pixel variance RTCD functions This patch fixes the system dependent entries for the half-pixel variance functions in both the RTCD and non-RTCD cases: - The generic C versions of these functions are now correct. Before all three cases called the hv code. - Wire up the ARM functions in RTCD mode - Created stubs for x86 to call the optimized subpixel functions with the correct parameters, rather than falling back to C code. Change-Id: I1d937d074d929e0eb93aacb1232cc5e0ad1c6184
2010-10-27 17:28:43 +02:00
#undef vp8_variance_halfpixvar16x16_h
#define vp8_variance_halfpixvar16x16_h vp8_variance_halfpixvar16x16_h_wmt
#undef vp8_variance_halfpixvar16x16_v
#define vp8_variance_halfpixvar16x16_v vp8_variance_halfpixvar16x16_v_wmt
#undef vp8_variance_halfpixvar16x16_hv
#define vp8_variance_halfpixvar16x16_hv vp8_variance_halfpixvar16x16_hv_wmt
Initial WebM release
2010-05-18 17:58:33 +02:00
#undef vp8_variance_subpixmse16x16
#define vp8_variance_subpixmse16x16 vp8_sub_pixel_mse16x16_wmt
#undef vp8_variance_getmbss
#define vp8_variance_getmbss vp8_get_mb_ss_sse2
#undef vp8_variance_mse16x16
#define vp8_variance_mse16x16 vp8_mse16x16_wmt
#undef vp8_variance_get16x16prederror
#define vp8_variance_get16x16prederror vp8_get16x16pred_error_sse2
#undef vp8_variance_get8x8var
#define vp8_variance_get8x8var vp8_get8x8var_sse2
#undef vp8_variance_get16x16var
#define vp8_variance_get16x16var vp8_get16x16var_sse2
#endif
#endif
#if HAVE_SSE3
extern prototype_sad(vp8_sad16x16_sse3);
extern prototype_sad(vp8_sad16x8_sse3);
extern prototype_sad_multi_same_address(vp8_sad16x16x3_sse3);
extern prototype_sad_multi_same_address(vp8_sad16x8x3_sse3);
extern prototype_sad_multi_same_address(vp8_sad8x16x3_sse3);
extern prototype_sad_multi_same_address(vp8_sad8x8x3_sse3);
extern prototype_sad_multi_same_address(vp8_sad4x4x3_sse3);
extern prototype_sad_multi_dif_address(vp8_sad16x16x4d_sse3);
extern prototype_sad_multi_dif_address(vp8_sad16x8x4d_sse3);
extern prototype_sad_multi_dif_address(vp8_sad8x16x4d_sse3);
extern prototype_sad_multi_dif_address(vp8_sad8x8x4d_sse3);
extern prototype_sad_multi_dif_address(vp8_sad4x4x4d_sse3);
#if !CONFIG_RUNTIME_CPU_DETECT
#undef vp8_variance_sad16x16
#define vp8_variance_sad16x16 vp8_sad16x16_sse3
#undef vp8_variance_sad16x16x3
#define vp8_variance_sad16x16x3 vp8_sad16x16x3_sse3
#undef vp8_variance_sad16x8x3
#define vp8_variance_sad16x8x3 vp8_sad16x8x3_sse3
#undef vp8_variance_sad8x16x3
#define vp8_variance_sad8x16x3 vp8_sad8x16x3_sse3
#undef vp8_variance_sad8x8x3
#define vp8_variance_sad8x8x3 vp8_sad8x8x3_sse3
#undef vp8_variance_sad4x4x3
#define vp8_variance_sad4x4x3 vp8_sad4x4x3_sse3
#undef vp8_variance_sad16x16x4d
Enable vp8_sad16x16x4d_sse3 in non-RTCD case Typo caused C version of 16x16x4 SAD to be called when built with --disable-runtime-cpu-detect. Change-Id: I0fe6fa67280b3a5f13acb3c8ed914f039aaaf316
2010-06-11 19:15:30 +02:00
#define vp8_variance_sad16x16x4d vp8_sad16x16x4d_sse3
Initial WebM release
2010-05-18 17:58:33 +02:00
#undef vp8_variance_sad16x8x4d
#define vp8_variance_sad16x8x4d vp8_sad16x8x4d_sse3
#undef vp8_variance_sad8x16x4d
#define vp8_variance_sad8x16x4d vp8_sad8x16x4d_sse3
#undef vp8_variance_sad8x8x4d
#define vp8_variance_sad8x8x4d vp8_sad8x8x4d_sse3
#undef vp8_variance_sad4x4x4d
#define vp8_variance_sad4x4x4d vp8_sad4x4x4d_sse3
#endif
#endif
#if HAVE_SSSE3
extern prototype_sad_multi_same_address(vp8_sad16x16x3_ssse3);
extern prototype_sad_multi_same_address(vp8_sad16x8x3_ssse3);
Add vp8_sub_pixel_variance16x8_ssse3 function Added SSSE3 function Change-Id: I8c304c92458618d93fda3a2f62bd09ccb63e75ad
2011-03-09 17:16:30 +01:00
extern prototype_subpixvariance(vp8_sub_pixel_variance16x8_ssse3);
Write SSSE3 sub-pixel filter function 1. Process 16 pixels at one time instead of 8. 2. Add check for both xoffset =0 and yoffset=0, which happens during motion search. This change gave encoder 1%~3% performance gain. Change-Id: Idaa39506b48f4f8b2fbbeb45aae8226fa32afb3e
2011-03-04 01:02:45 +01:00
extern prototype_subpixvariance(vp8_sub_pixel_variance16x16_ssse3);
Initial WebM release
2010-05-18 17:58:33 +02:00
#if !CONFIG_RUNTIME_CPU_DETECT
#undef vp8_variance_sad16x16x3
#define vp8_variance_sad16x16x3 vp8_sad16x16x3_ssse3
#undef vp8_variance_sad16x8x3
#define vp8_variance_sad16x8x3 vp8_sad16x8x3_ssse3
Add vp8_sub_pixel_variance16x8_ssse3 function Added SSSE3 function Change-Id: I8c304c92458618d93fda3a2f62bd09ccb63e75ad
2011-03-09 17:16:30 +01:00
#undef vp8_variance_subpixvar16x8
#define vp8_variance_subpixvar16x8 vp8_sub_pixel_variance16x8_ssse3
Write SSSE3 sub-pixel filter function 1. Process 16 pixels at one time instead of 8. 2. Add check for both xoffset =0 and yoffset=0, which happens during motion search. This change gave encoder 1%~3% performance gain. Change-Id: Idaa39506b48f4f8b2fbbeb45aae8226fa32afb3e
2011-03-04 01:02:45 +01:00
#undef vp8_variance_subpixvar16x16
#define vp8_variance_subpixvar16x16 vp8_sub_pixel_variance16x16_ssse3
Initial WebM release
2010-05-18 17:58:33 +02:00
#endif
#endif
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
#if HAVE_SSE4_1
extern prototype_sad_multi_same_address_1(vp8_sad16x16x8_sse4);
extern prototype_sad_multi_same_address_1(vp8_sad16x8x8_sse4);
extern prototype_sad_multi_same_address_1(vp8_sad8x16x8_sse4);
extern prototype_sad_multi_same_address_1(vp8_sad8x8x8_sse4);
extern prototype_sad_multi_same_address_1(vp8_sad4x4x8_sse4);
#if !CONFIG_RUNTIME_CPU_DETECT
#undef vp8_variance_sad16x16x8
#define vp8_variance_sad16x16x8 vp8_sad16x16x8_sse4
#undef vp8_variance_sad16x8x8
#define vp8_variance_sad16x8x8 vp8_sad16x8x8_sse4
#undef vp8_variance_sad8x16x8
#define vp8_variance_sad8x16x8 vp8_sad8x16x8_sse4
#undef vp8_variance_sad8x8x8
#define vp8_variance_sad8x8x8 vp8_sad8x8x8_sse4
#undef vp8_variance_sad4x4x8
#define vp8_variance_sad4x4x8 vp8_sad4x4x8_sse4
#endif
#endif
Initial WebM release
2010-05-18 17:58:33 +02:00
#endif
Reference in New Issue Copy Permalink