- 2e9f3e5 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. by fbarchard@google.com · 9 years ago master
- d880a90 make randomize take 64 bit count to allow larger images to be tested by fbarchard@google.com · 9 years ago
- 05416e2 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. by fbarchard@google.com · 9 years ago
- b07de87 enable intrinsics for clangcl if -mssse3 is enabled. by fbarchard@google.com · 9 years ago
- 965ef27 bump deps to head chromium to get prebuilt clangcl by fbarchard@google.com · 9 years ago
- b3d3db1 align clangcl using declspec instead of gcc style vector by fbarchard@google.com · 9 years ago
- da6ee35 Revert "libyuv: Include foo_posix.cc in the gn build, like it's done in the gyp build." by johannkoenig@google.com · 9 years ago
- d7242b5 libyuv: Include foo_posix.cc in the gn build, like it's done in the gyp build. by johannkoenig@google.com · 9 years ago
- 684c09c roll chromium revision to head in attempt to resolve build time performance hit due to external compilers on windows. by fbarchard@google.com · 9 years ago
- a09fd1a scale tests use int64 for size BUG=none TESTED=untested R=harryjin@google.com by fbarchard@google.com · 9 years ago
- 01ca071 Test rotation with negative height by fbarchard@google.com · 9 years ago
- d3d8e0d make source for planar tests contiguous to test planar functions coalesce into a single low level call. by fbarchard@google.com · 9 years ago
- 00c0141 Add scale color test place holder source BUG=401 TESTED=untested by fbarchard@google.com · 9 years ago
- e5fbad8 Add test for scaling color accuracy BUG=401 TESTED=scale_color test by fbarchard@google.com · 9 years ago
- e787144 adjust dimensions for scale factor tests to ensure the scale factor tested is actually used. by fbarchard@google.com · 9 years ago
- bd2d903 odd width support for ARGBSobel functions. Improves performance for images that are not a multiple of 8 pixels. by fbarchard@google.com · 9 years ago
- cfce47e 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. by fbarchard@google.com · 9 years ago
- 535a714 Scale Down by factor tests scale down to specified ratio rather than up. This ensures the alignment constrains on the destination dont cause a different factor to be used. by fbarchard@google.com · 9 years ago
- 7c09264 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. by fbarchard@google.com · 9 years ago
- 80be246 add test for box filter before improving odd width. by fbarchard@google.com · 9 years ago
- c38aeec scale down by 2 on argb images support odd widths using _any function. by fbarchard@google.com · 9 years ago
- 632c50f include posix source for 64 bit clang builds. by fbarchard@google.com · 9 years ago
- 3666015 add nacl macros for arm to YUV422TORGB_SETUP_REG. by fbarchard@google.com · 9 years ago
- 7be3bc6 enable speed optimization for libyuv by fbarchard@google.com · 9 years ago
- b33dc47 sobel use LL for constants to be passed in as int64 by fbarchard@google.com · 9 years ago
- 1be66a7 psnr tool, work around for ios 64 bit compiler where int passed into assembly needs to be explicitely cast to 'w' register. by fbarchard@google.com · 9 years ago
- d3f51b5 work arounds for ios 64 bit compiler where int passed into assembly needs to be explicitely cast to 'w' register. by fbarchard@google.com · 9 years ago
- b0f8352 row_neon64 additional fixes for warning on ios where int doesnt match %2 size which is 64 bit by default. change size to explicitely 32 bit with %w2. by fbarchard@google.com · 9 years ago
- a20e2c6 row_neon64 fix for warning on ios where int width doesnt match %2 size which is 64 bit by default. change size to explicitely 32 bit with %w2. by fbarchard@google.com · 9 years ago
- 6d55546 scale 64 bit fix for warning on ios where int width doesnt match %2 size which is 64 bit by default. change size to explicitely 32 bit with %w2. by fbarchard@google.com · 9 years ago
- e8c90c3 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. by fbarchard@google.com · 9 years ago
- 54e78d1 make windows code built with clangcl include the _posix source code. by fbarchard@google.com · 9 years ago
- 2c44965 make row_win windows code built with clangcl include the _posix source code. by fbarchard@google.com · 9 years ago
- 484e5d2 make windows code built with clangcl include the _posix source code. depot_tools excludes these source files now, so they need to be manually included. by fbarchard@google.com · 9 years ago
- 1ec12e5 Roll chromium deps to version that includes vs2015 link error fix. by fbarchard@google.com · 9 years ago
- ab6b224 fix for arm builds where tmp for assembly produces an error if its uninitialized. by fbarchard@google.com · 9 years ago
- 31806d7 scale to 3/4 bug fix for odd widths. multiply to index into source by scale factor should be 4 / 3 not 3 / 4. by fbarchard@google.com · 9 years ago
- 9f4636e AVX2 port of ScaleDownBy4. by fbarchard@google.com · 9 years ago
- 5abb6d4 disable stucture padded warnings on win64 builds. by fbarchard@google.com · 9 years ago
- e23274c remove unused function SumBox. BUG=432 TESTED=untested R=bcornell@google.com by fbarchard@google.com · 9 years ago
- 428fce6 remove unused function ScalePlaneBoxRow_* which was for slow box filter that is no longer used. by fbarchard@google.com · 9 years ago
- 31c875f enable chromium_code=1 to increase warning level. this will enable the same warnings as when built within chrome, helping avoid breakage in future. by fbarchard@google.com · 9 years ago
- 35aa92a fixed unused variables/code warnings in scale box function by fbarchard@google.com · 9 years ago
- 9870a11 Add PROJECT setting to codereview.settings by kjellander@google.com · 9 years ago
- ab8bf44 Enable FORCE_HTTPS_COMMIT_URL in codereview.settings by kjellander@google.com · 9 years ago
- f995021 Work around casting warnings in scale_neon64.cc for ios 64 bit. by fbarchard@google.com · 9 years ago
- a81da96 Work around for ios 64 bit build warning - use explicit word register for int. by fbarchard@google.com · 9 years ago
- 4e78b8d scale to 3/4 or 3/8 with odd width destinations efficiently. previously if width was not multiple of what the simd loop would do (24), scaling would fall back on slower C code. This change allows SIMD to be used for most of the scaling and C for the remainder, improving efficiency. by fbarchard@google.com · 9 years ago
- 1ffb04b Allow ScaleRowDown any functions to accept non-power of 2 for destination SIMD multiple. by fbarchard@google.com · 9 years ago
- c8a2c23 NaCl/GYP: remove references to prep_toolchain from libyuv. prep_toolchain is now a no-op. by fbarchard@google.com · 9 years ago
- 2b7f6b7 ScaleAddRows_Any_SSE2 functions for handling odd widths. by fbarchard@google.com · 9 years ago
- 01db3d1 Remove declspec(align(32)) from AVX2 functions. by fbarchard@google.com · 9 years ago
- 812f59e box and point sampling use scaledownby4 but linear and bilinear do not. by fbarchard@google.com · 9 years ago
- e52b9c3 make box filter upsampler consider a pixel width of less than 1 to be 1. This makes it behave as a point sampler. by fbarchard@google.com · 9 years ago
- 8eb887f add empty header for backwards compatibility. BUG=none TESTED=lint passes by fbarchard@google.com · 9 years ago
- bb43f06 Add a scale down by 3 test for testing general purpose box/bilinear with scale factor of 3, but allowing width/height specified to do odd with testing. by fbarchard@google.com · 9 years ago
- c998631 lsl by 2 requires a number sign for xcode on ios 64 bit build. add the # sign for ios compatibility. remove legacy x86 asm files that are unused. the unused files cause complications in build systems that build all files. by fbarchard@google.com · 9 years ago
- 32ad6e0 Remove unused variable 'I422ToRGB565Row' that breaks osx builds. by fbarchard@google.com · 9 years ago
- 013e812 Port box filter to AVX2. by fbarchard@google.com · 9 years ago
- b5ea79d add rows handle height of 1 using a more general while-style loop. by fbarchard@google.com · 9 years ago
- c7161d1 Remove code alignment declspec from Visual C versions for vs2014 compatibility. by fbarchard@google.com · 9 years ago
- 1eb51bc Fix bug in YUV to RGB for gcc/clang and enable affected functions. by fbarchard@google.com · 9 years ago
- bb5a009 ARGB4444ToARGB and ARGB1555ToARGB ported to AVX2. by fbarchard@google.com · 9 years ago
- 8b9f908 RGB565ToARGB AVX2 vzeroupper before the ret, not after. by fbarchard@google.com · 9 years ago
- 5f60985 Add ScaleARGBFilterCols_NEON for ARM32/64 by yang.zhang@arm.com · 9 years ago
- 3d1176a ARGBToYJRow_AVX2 hooked up for ARGBToJ422 BUG=none TESTED=ARGBToJ422 unittest by fbarchard@google.com · 9 years ago
- 8f0b327 ARGBToUV AVX2 functions hooked up. by fbarchard@google.com · 9 years ago
- 9afabe2 Add ARGBToY AVX calls. by fbarchard@google.com · 9 years ago
- 2827277 port RGB565ToARGB to AVX2. by fbarchard@google.com · 9 years ago
- e2ea106 shift for arm wants a # sign for nacl and ios. by fbarchard@google.com · 9 years ago
- 44b6ba9 Scale down by 4 for odd number of destination pixels using 'any' that handles SIMD for multiple of 8 pixels, and C for the remainder. by fbarchard@google.com · 9 years ago
- 62a9fe3 code style cleanup of scale functions. no functional change. by fbarchard@google.com · 9 years ago
- c00796c Fix the issue of q4 register not in clobber list for ARMv7 by yang.zhang@arm.com · 9 years ago
- c70c7c0 scale to half size optimization for avx2 - use pmaddubsw instruction to horizontally add bytes, then pavgw to round and divide by 2. by fbarchard@google.com · 9 years ago
- f23d622 Add ScaleARGBCols_NEON for ARM32/64 by yang.zhang@arm.com · 9 years ago
- 416c48d add new scale_any to build files. BUG=419 TESTED=untested R=bcornell@google.com by fbarchard@google.com · 9 years ago
- 72673ac linear and point sample scale to half size for AVX2. by fbarchard@google.com · 9 years ago
- 9ef8999 scale to half size use pmadd/pavgw to horizontal averaging. by fbarchard@google.com · 9 years ago
- e6ca9cc Scale down by 2 AVX2 port. Processes twice as many pixels as SSE2 and takes advantage of 3 argument instructions to reduce register usage and number of instructions. by fbarchard@google.com · 9 years ago
- f16f33d All cpu flags to be set so that instead of comparing C code, compare assembler to assembler, for benchmarking purposes. by fbarchard@google.com · 9 years ago
- 4f46a3e remove 1 tab from each row of histograms in libyuvTest.TestFullYUV and libyuvTest.TestScaleFullYUV. by fbarchard@google.com · 9 years ago
- d41fbf4 Handle scale down by factor of 2 efficiently by calling SIMD for multiple of 16 destination pixels, and C for remainder. by fbarchard@google.com · 9 years ago
- 0d3bfab Add nacl macros to ScaleFilterCols_NEON on ARM32/64 platform by yang.zhang@arm.com · 9 years ago
- d28cd77 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. by fbarchard@google.com · 9 years ago
- d6d7de5 Add ScaleFilterCols_NEON for ARM32/64 by yang.zhang@arm.com · 9 years ago
- 70e5c81 copy width to int64 to pass to assembly to avoid warning on ios 64 bit for implicit: value size does not match register size specified by the constraint and modif by fbarchard@google.com · 9 years ago
- 0e4388a I422ToRGB24 AVX2 and I422ToRAW by fbarchard@google.com · 9 years ago
- 4d387fc Add ScaleARGBRowDown2Linear_NEON for ARM32/64 by yang.zhang@arm.com · 9 years ago
- e246e6c Add ARGBToRGB565DitherRow_NEON for ARM32/64 by yang.zhang@arm.com · 9 years ago
- 3b4f5eb Port J422 colorspace to GCC BUG=414 TESTED=try bots R=tpsiaki@google.com by fbarchard@google.com · 9 years ago
- 92f7f42 rename I400 to J400 and I400 reference to I400. J400 is a simple replication of values to convert to RGB, which is what the old I400 was. I400 reference is the Y part of the YUV formula, so renaming that to I400. by fbarchard@google.com · 9 years ago
- 35f0add cpuid ifdefs fixed to remove some duplicate code cases. by fbarchard@google.com · 9 years ago
- df90670 clangcl build fix for __cpuid in psnr util. Since clangcl provides the intrinsic thru its Visual C emulation, don't duplicately define the function with an inline version, which is normally needed for gcc/clang. by fbarchard@google.com · 9 years ago
- f301777 Fix YToARGB and tweaks to thresholds in YUV tests. by fbarchard@google.com · 9 years ago
- bfba8e4 Add test for color accuracy of J422ToARGB by fbarchard@google.com · 9 years ago
- f2fad0f Optimized J422ToARGB. BUG=414 TESTED=J422ToARGB unittest R=tpsiaki@google.com by fbarchard@google.com · 9 years ago
- e408a37 Improve accuracy of J422 color space using higher precission fixed point and bias. by fbarchard@google.com · 9 years ago
- ca5b1bd Add ScaleAddRows_NEON for ARM32/64 by yang.zhang@arm.com · 9 years ago
- 63726ed test different ways to round and clamp by fbarchard@google.com · 9 years ago
- 952ca5f Fix for planar functions SSE2 enable when building with clang for Windows. by fbarchard@google.com · 9 years ago