.cpp + .cu to call kernel function

Accelerated Computing CUDA CUDA Programming and Performance
1

Hi
I’m trying to call kernel function .cpp + .cu

In .cpp ,
I called
…
extern “C” void Kernel_Hilbert(cufftComplex *a, float *b, int Width, int Height);
…
Kernel_Hilbert(d_input_value, d_pointwise_coeff,(int)TWIDTH,(int)THEIGHT);

and

In .cu ,

#include <cuda.h>
#include <cufft.h>
#include "device_launch_parameters.h"

extern "C" void Kernel_Hilbert(cufftComplex *a, float *b, int Width, int Height)
{
	Kernel_pointproduct<<<1,Width>>>(a, b,Width,Height);
}


__global__ void Kernel_pointproduct(cufftComplex *a, float *b, int Width, int Height){
	
	int tid = blockDim.x * blockIdx.x + threadIdx.x;
	
	if(tid < Width){
		for(int i=0; i<Height; i++){
			a[i*Width+tid].x = a[i*Width+tid].x * b[tid];
			a[i*Width+tid].y = a[i*Width+tid].y * b[tid];
		}
	}
}

But there is an error.
the error said
“Kernel_pointproduct” is undefined

anyone have an idea to solve it?

Thank you!

2

Move the definition of the kernel to before the place where you invoke it. Or else use a prototype to provide a forward reference. This is a basic C programming concept and has nothing to do with CUDA.