clGetPlatformIDs : Device Not Found

Accelerated Computing CUDA CUDA Programming and Performance
1

Hello,

I’m trying to use OpenCL on two different GPUs : Tesla C 1060 and Quadro FX 5800. When I use this code :

cl_uint platNum;

cl_platform_id platform;

cl_int err;

err == clGetPlatformIDs(1, &platform, &platNum);

The function returns the error “Device not found” (on the two GPUs). I searched what’s the problem but I can’t find the solution or even the cause (according to the doc, we can’t even have this error with this function). Does someone have an idea of the problem and how to solve it please ?

On the Quadro FX, I’m working with OpenCL 1.0 and CUDA 3.2 ; on the Tesla C 1060, I’m working with OpenCL 1.0 and CUDA 4.0 (both on Linux).

Best regards

2

I believe you should use the assignment operator (=) not comparison operator (==), that is:

cl_uint platNum;

cl_platform_id platform;

cl_int err;

err = clGetPlatformIDs(1, &platform, &platNum);