Hello,
Is it possible to change only a few values at different locations in a CUDA array (in GPU memory) without running a kernel through the whole array ?
If yes, how ?
Thanks
Hello,
Is it possible to change only a few values at different locations in a CUDA array (in GPU memory) without running a kernel through the whole array ?
If yes, how ?
Thanks
up
Sure, you can address specific elements of your array with cudaMemcpy. Or you could just use thrust::device_vector for memory management:
#include <thrust/device_vector.h>
int main(void)
{
thrust::device_vector<int> v(100);
v[50] = 13;
std::cout << "element 50 is: " << v[50] << std::endl;
}