| .. |
|
compare_common.cc
|
compare_common visual c port
|
2014-01-09 19:11:09 +00:00 |
|
compare_gcc.cc
|
rename source files from row_posix.cc etc to row_gcc.cc to avoid gyp build filtering out source files from build when on windows with clang. The source code contained in row_gcc.cc is gcc syntax inline assembly available for any platform that supports gcc or clang for intel cpus.
|
2015-06-09 17:27:52 +00:00 |
|
compare_neon64.cc
|
fix for warning on ios 64 bit that int width doesnt match %2 size which is 64 bit by default. change size to explicitely 32 bit with %w2.
|
2015-05-04 21:16:03 +00:00 |
|
compare_neon.cc
|
change vector range notation to a list of registers for clang compatibility. break compare into 2 neon files for consistency with other neon64 files.
|
2014-09-15 23:39:43 +00:00 |
|
compare_win.cc
|
rename source files from row_posix.cc etc to row_gcc.cc to avoid gyp build filtering out source files from build when on windows with clang. The source code contained in row_gcc.cc is gcc syntax inline assembly available for any platform that supports gcc or clang for intel cpus.
|
2015-06-09 17:27:52 +00:00 |
|
compare.cc
|
Enable assembly for clangcl build on Windows. Previously assembly was disabled so clangcl would work, but only with C code. As clangcl mimics both Visual C and GCC, ifdefs need to pick one or the other or often you'll end up with both. In this CL we disable most Visual C code and use the GCC versions which allow assembly for both 32 and 64 bit intel.
|
2015-03-19 20:36:31 +00:00 |
|
convert_argb.cc
|
ARGB4444ToARGB and ARGB1555ToARGB ported to AVX2.
|
2015-04-07 23:52:57 +00:00 |
|
convert_from_argb.cc
|
ARGBToYJRow_AVX2 hooked up for ARGBToJ422
|
2015-04-07 00:39:25 +00:00 |
|
convert_from.cc
|
Remove unused variable 'I422ToRGB565Row' that breaks osx builds.
|
2015-04-14 02:50:35 +00:00 |
|
convert_jpeg.cc
|
libyuv::MJPGToI420() and libyuv::MJPGToARGB() return failure if callback to JPeg fails.
|
2014-01-28 03:08:59 +00:00 |
|
convert_to_argb.cc
|
Remove Q420 fourcc support.
|
2015-02-11 18:20:54 +00:00 |
|
convert_to_i420.cc
|
Remove Q420 fourcc support.
|
2015-02-11 18:20:54 +00:00 |
|
convert.cc
|
disable faulty avx2 in argb conversions and box filter. and extend temporary buffer to 128 for an avx2 any function.
|
2015-07-07 15:40:24 -07:00 |
|
cpu_id.cc
|
Enable assembly for clangcl build on Windows. Previously assembly was disabled so clangcl would work, but only with C code. As clangcl mimics both Visual C and GCC, ifdefs need to pick one or the other or often you'll end up with both. In this CL we disable most Visual C code and use the GCC versions which allow assembly for both 32 and 64 bit intel.
|
2015-03-19 20:36:31 +00:00 |
|
mjpeg_decoder.cc
|
disable stucture padded warnings on win64 builds.
|
2015-04-28 23:18:07 +00:00 |
|
mjpeg_validate.cc
|
Remove code alignment declspec from Visual C versions for vs2014 compatibility.
|
2015-04-12 23:54:26 +00:00 |
|
planar_functions.cc
|
odd width support for ARGBSobel functions. Improves performance for images that are not a multiple of 8 pixels.
|
2015-05-28 22:22:28 +00:00 |
|
rotate_argb.cc
|
Fix the bug (0 extension from int to int64) in ScaleARGBRowDownEven_NEON.
|
2015-02-10 06:04:12 +00:00 |
|
rotate_mips.cc
|
only enable mips assembly for old 32 bit abi. new 32 bit abi and 64 bit bit able remove t4 to t7 and add a4 to a7
|
2014-06-24 23:40:52 +00:00 |
|
rotate_neon64.cc
|
Copy width to 64 bit register to work around clang 3.4 warning
|
2014-10-13 23:26:17 +00:00 |
|
rotate_neon.cc
|
Change ifdefs for arm 32 and 64 bit so there will only be 32 bit in legacy mode.
|
2014-09-15 22:05:01 +00:00 |
|
rotate.cc
|
check height is non-zero
|
2015-06-11 16:35:20 -07:00 |
|
row_any.cc
|
disable faulty avx2 in argb conversions and box filter. and extend temporary buffer to 128 for an avx2 any function.
|
2015-07-07 15:40:24 -07:00 |
|
row_common.cc
|
simd for 1 to 1
|
2015-06-17 18:22:11 -07:00 |
|
row_gcc.cc
|
blend remove alignment 1 pixel loop for less overhead.
|
2015-06-24 11:34:12 -07:00 |
|
row_mips.cc
|
build fixe for InterpolateRow_MIPS_DSPR2
|
2015-02-07 01:01:30 +00:00 |
|
row_neon64.cc
|
Change Sobel to use JPeg Luma calculation instead of extracting G channel. Using luma produces a better sobel that respects all 3 channels of RGB. Historically the G channel was used to improve performance, and because the luma of I420 is a constrained range, hurting quality. Using the JPeg variation of YUV, the luma is more accurate, including cross platform, better optimized for AVX2 and odd widths, and full range.
|
2015-05-27 22:32:26 +00:00 |
|
row_neon.cc
|
Change Sobel to use JPeg Luma calculation instead of extracting G channel. Using luma produces a better sobel that respects all 3 channels of RGB. Historically the G channel was used to improve performance, and because the luma of I420 is a constrained range, hurting quality. Using the JPeg variation of YUV, the luma is more accurate, including cross platform, better optimized for AVX2 and odd widths, and full range.
|
2015-05-27 22:32:26 +00:00 |
|
row_win.cc
|
disable faulty avx2 in argb conversions and box filter. and extend temporary buffer to 128 for an avx2 any function.
|
2015-07-07 15:40:24 -07:00 |
|
scale_any.cc
|
Box filter for YUV use rows with accumulation buffer for better memory behavior. The old code would do columns accumulated into registers, and then store the result once. This was slow from a memory point of view. The new code does a row of source at a time, updating an accumulation buffer every row. The accumulation buffer is small, and should fit cache. Before each accumulation of N rows, the buffer needs to be reset to zero. If the memset is a bottleneck, it would be faster to do the first row without an add, storing to the accumulation buffer, and then add for the remaining rows.
|
2015-06-09 01:05:18 +00:00 |
|
scale_argb.cc
|
odd width support for scale by even scale factor and box scale down by 4. scale down by 4 uses scale down by 2 internally.
|
2015-05-26 17:56:51 +00:00 |
|
scale_common.cc
|
Box filter for YUV use rows with accumulation buffer for better memory behavior. The old code would do columns accumulated into registers, and then store the result once. This was slow from a memory point of view. The new code does a row of source at a time, updating an accumulation buffer every row. The accumulation buffer is small, and should fit cache. Before each accumulation of N rows, the buffer needs to be reset to zero. If the memset is a bottleneck, it would be faster to do the first row without an add, storing to the accumulation buffer, and then add for the remaining rows.
|
2015-06-09 01:05:18 +00:00 |
|
scale_gcc.cc
|
rename source files from row_posix.cc etc to row_gcc.cc to avoid gyp build filtering out source files from build when on windows with clang. The source code contained in row_gcc.cc is gcc syntax inline assembly available for any platform that supports gcc or clang for intel cpus.
|
2015-06-09 17:27:52 +00:00 |
|
scale_mips.cc
|
only enable mips assembly for old 32 bit abi. new 32 bit abi and 64 bit bit able remove t4 to t7 and add a4 to a7
|
2014-06-24 23:40:52 +00:00 |
|
scale_neon64.cc
|
work arounds for ios 64 bit compiler where int passed into assembly needs to be explicitely cast to 'w' register.
|
2015-05-05 22:46:16 +00:00 |
|
scale_neon.cc
|
fix for arm builds where tmp for assembly produces an error if its uninitialized.
|
2015-04-30 18:21:19 +00:00 |
|
scale_win.cc
|
rename source files from row_posix.cc etc to row_gcc.cc to avoid gyp build filtering out source files from build when on windows with clang. The source code contained in row_gcc.cc is gcc syntax inline assembly available for any platform that supports gcc or clang for intel cpus.
|
2015-06-09 17:27:52 +00:00 |
|
scale.cc
|
Box filter for YUV use rows with accumulation buffer for better memory behavior. The old code would do columns accumulated into registers, and then store the result once. This was slow from a memory point of view. The new code does a row of source at a time, updating an accumulation buffer every row. The accumulation buffer is small, and should fit cache. Before each accumulation of N rows, the buffer needs to be reset to zero. If the memset is a bottleneck, it would be faster to do the first row without an add, storing to the accumulation buffer, and then add for the remaining rows.
|
2015-06-09 01:05:18 +00:00 |
|
video_common.cc
|
Remove bayer format support from libyuv. This format is very rare and used on legacy hardware. Its not well optimized and has bugs related to odd widths. Removing the format will allow tests to pass under more circumstances, run faster and allow focus on higher priority quality and performance issues.
|
2015-02-09 19:58:19 +00:00 |
97b35daf75