__kernel void foo(__global float* a, __global float* b, __global float* c)
{
    unsigned int i = get_global_id(0);
    float temp;
    for(int j = 0; j < 1024; ++j)
        temp += cos(sin(a[i])*b[i])*a[i] + b[i];
    c[i] = temp/1024.0f;
}