Good day! I've just built the latest master branch of OpenCV with CUDA support. So, I tried some sample code:
Mat shot = imread("D:\\shot.bmp"); // Shot.bmp is a picture 1571x1002
gpu::GpuMat gpuShot;
gpuShot.upload(shot);
The last line "gpuShot.upload(shot);" takes huge time to invoke! It takes about 4-5 seconds to upload to GPU! Is it normal for that? I don't think so... BTW: My hardware is GeForce GT 520.
And second question: the file "opencv_gpu249d.dll" has size about 250 Mb! Meanwhile, "opencv_gpu242d.dll" from official prebuild boundle (v.2.4.2) has size approx 800 Kb. What the hell of a size? And what I doing wrong? Please help, thanks.