cudaMemset fails on __device__ variable

Question

I am having trouble using cudaMemset on a device variable. Is it possible to use the reference to the device variable for cudaMemset, or is it just a m

Answer 1

I believe you can also use cudaMemcpyFromSymbol: A function, such as the following kernel, can change the value of the variable declared in global memory (outside of the main function)

__global__ void kernel1() { d_test = 1.0; }

Inside your main, you can obtain the value using cudaMemcpyFromSymbol

cudaMemcpyFromSymbol(&h_test,"d_test",sizeof(float),0,cudaMemcpyDeviceToHost);

Of course, there is also cudaMemcpyToSymbol to change the value of the global variable.

The idea came from here: Having problem assigning a device variable in CUDA