Cuda OpenCV - Gpu API call (initialization error)

asked 2014-10-03 13:41:37 -0600

andreiMtl gravatar image

updated 2018-12-24 16:03:12 -0600

sturkmen gravatar image

Hello, I'm trying to compile CUDA6.0 inside OpencCV 2.4.9 with CMake 2.8.12 and Visual Studio 2012 . I have a Nvidia GTX760.

Cmake recognizes cuda properly and it makes a solution without generating any errors or flags. I can also build this solution without generating any errors. My problem is when I try to use the generated libraries. All the non-gpu libraries work just fine, but if I try to use any gpu function I get:

OpenCV Error: Gpu API call(initialization error) in CudaDeviceInfoFuncTable:getCudaEnabledDeviceCount, file C:/opencv/sources/modules/dynamicuda/include/opencv2/dynamicuda/dynamicuda.hpp, line 652

Any hints on how to solve this?

edit retag flag offensive close merge delete

Comments

If you install CUDA, do you succeed in running the deviceQuery sample? This is the first thing you should do before moving to OpenCV. I am guessing your card is not found by your CUDA installation.

StevenPuttemans gravatar imageStevenPuttemans ( 2014-10-06 07:45:50 -0600 )edit
add a comment