Well, I'm trying to do matrix multiplication with gpu::gemm, but occurs fatal error: The function/feature is not implemented (The library was build without CUBLAS) in unknown function and arithm.cpp:72: error: (-213) The library was build without CUBLAS.
I built the OpenCV 2.4.9 with Cmake 2.8.1 and set ON the tag WITH_CUBLAS, but the fatal error occurs again.
Already I used CUDA 5.0 and CUDA 4.2, but yet not working.
What should I do?