Pycuda gives 'cuModuleGetFunction named symbol not found error !!!

Accelerated Computing CUDA CUDA Programming and Performance
1

Hello,

File : cuda.cu

extern “C” host void cublas_gemm(float* A, float* B, float* C, int m, int n, int k){

    cudaMemset(C, 0, m * k * sizeof(float));
	cublasHandle_t handle;
	//cublasCreate(&handle);
          ...
           cublasSgemm()
          .....

}

File python.py
{
mod = SourceModule(open(“cuda.cu”).read())
cublasGemm = mod.get_function(“cublas_gemm”) <<<<--------------
}

The above line gives me error as below :

pycuda._driver.LogicError: (‘The following error happened while compiling the node’, cublasGemmEx(GpuContiguous.0, GpuContiguous.0), ‘\n’, ‘cuModuleGetFunction failed: named symbol not found’)

I have read about name mangling as I have used extern “C” in above code. But still it gives me error.
Please let me know how do I solve it ?

Thanks !!!