Parallelizing vector addition using multithread

==6564== Profiling application: ./vector_add_grid
==6564== Profiling result:
Time(%)      Time     Calls       Avg       Min       Max  Name
 55.65%  20.312ms         2  10.156ms  10.150ms  10.162ms  [CUDA memcpy HtoD]
 41.24%  15.050ms         1  15.050ms  15.050ms  15.050ms  [CUDA memcpy DtoH]
  3.11%  1.1347ms         1  1.1347ms  1.1347ms  1.1347ms  vector_add(float*, float*, float*, int)