After setting depth to CV_32F for non double types this produces difference in results. I've attached test that reproduce this issue.The issue was found during anali broken GPU tests  for multiply/device functions.

2. Two new NeonOptimized functions subtract_8u8u32f and subtract_8u8u8s were added

2. Two new NeonOptimized functions subtract_8u8u32f and subtract_8u8u8s were added

Conflicts:
	modules/core/src/arithm.cpp

fixed and improving formatting in opencv2refman.pdf. added support for n-channel mask in Mat::copyTo() and n-channel images in cv::compare(). fixed 2 compile warnings in opencv_python.

added COVAR_ and SORT_ enums to core.hpp; fixed many, many VS2005, VS2010 and MinGW (GCC 4.5.2) warnings

replaced "const InputArray&" => "InputArray"; made InputArray and OutputArray references. added "None()" constant (no array()).

a big patch; use special proxy types (Input/OutputArray, Input/OutputArrayOfArrays) for passing in vectors, matrices etc.

reworked nearly all of the OpenCV tests (except for opencv_gpu tests) - they now use the Google Test engine.