opencv/modules
Kai Hugo Hustoft Endresen 551b5d3e1a StereoSGBM.cpp - use SSE2 for pass 2 using MODE_HH
With a test image set of 2800x1400 bytes on a Intel Core i7 5960X this improves runtime of MODE_HH with about 10%. (this particular replaced code segment is approx 3 times faster than the non-SSE2 variant). I was able to reduce runtime by 130 ms by this simple fix.

The second part of the SSE2 optimized part could probably be optimized further by using shift SSE2 operations, but I imagine this would improve performance 10-20 ms at best.
2016-01-08 00:32:52 +01:00
..
calib3d StereoSGBM.cpp - use SSE2 for pass 2 using MODE_HH 2016-01-08 00:32:52 +01:00
core Merge pull request #5883 from alalek:fix_5876 2016-01-04 05:28:41 +00:00
cudaarithm fix #5780 : cv::cuda::createContinuouseffectively ignores its cols argument 2015-12-10 14:20:28 +03:00
cudabgsegm move obsolete algorithms from cudabgsegm to cudalegacy: 2015-01-19 14:17:00 +03:00
cudacodec Adding support for WinRT(WinPhone 8/8.1 and Win Store) via CMake 3.1 2015-03-02 22:26:43 +03:00
cudafeatures2d Adding support for WinRT(WinPhone 8/8.1 and Win Store) via CMake 3.1 2015-03-02 22:26:43 +03:00
cudafilters Adding support for WinRT(WinPhone 8/8.1 and Win Store) via CMake 3.1 2015-03-02 22:26:43 +03:00
cudaimgproc disable sanity tests for AlphaComp and PyrLKOpticalFlowDense 2015-11-27 16:46:47 +03:00
cudalegacy Some changes to support mingw-w64 2015-09-01 00:59:08 +03:00
cudaobjdetect change links from samples/python2 to samples/python 2015-12-18 11:00:30 +03:00
cudaoptflow Modified sparse pyrlk optical flow to allow input of an image pyramid which thus allows caching of image pyramids on successive calls. 2015-12-29 10:48:14 -05:00
cudastereo increase epsilons in some tests: 2015-11-27 16:46:20 +03:00
cudawarping Modified sparse pyrlk optical flow to allow input of an image pyramid which thus allows caching of image pyramids on successive calls. 2015-12-29 10:48:14 -05:00
cudev fix BORDER_WRAP processing on Maxwell generation 2015-11-27 16:45:26 +03:00
features2d features2d agast: clear in/out keypoints container 2015-12-25 13:49:24 +03:00
flann Added missing copyright headers. 2015-10-13 15:02:38 +03:00
highgui WinRT consistency 2015-12-31 14:31:26 +01:00
imgcodecs update osx and ios build_framework.py 2015-12-17 16:17:31 +03:00
imgproc fix cv::MARKER_TRIANGLE_UP 2016-01-01 10:10:01 +01:00
java Fix java version++ 2015-12-18 18:02:16 +03:00
ml ml: fix NormalBayesClassifier bulk prediction(#5911) 2016-01-04 11:47:08 +01:00
objdetect change links from samples/python2 to samples/python 2015-12-18 11:00:30 +03:00
photo change links from samples/python2 to samples/python 2015-12-18 11:00:30 +03:00
python Merge pull request #5821 from alalek:build_framework 2015-12-18 10:12:55 +00:00
shape fix mat copy in shape 2015-12-23 10:48:36 +01:00
stitching HAL moved back to core 2015-12-17 12:33:23 +03:00
superres Fixed cmake and build issues when using Visual Studio 2015 2015-10-29 11:50:48 +03:00
ts disable ts module if it is not used: BUILD_TESTS=OFF and BUILD_PERF_TESTS=OFF 2015-12-12 15:23:31 +03:00
video change links from samples/python2 to samples/python 2015-12-18 11:00:30 +03:00
videoio solve videoio/src/cap_mjpeg_decoder.cpp compiling problem using VS 2008 2015-12-25 13:57:26 +03:00
videostab WinRT consistency 2015-12-31 14:31:26 +01:00
viz Added new functionalities to viz module 2015-08-15 01:03:59 +02:00
world update osx and ios build_framework.py 2015-12-17 16:17:31 +03:00
CMakeLists.txt world fix 2014-08-05 20:12:35 +04:00