This version provides a better for multi-gpu calculation. Besides, user can adjust chunk_size_x to have better performance for specific gpu apparatus. Also, different method is supported here.
2.1
This version provides a better for multi-gpu calculation. Besides, user can adjust chunk_size_x to have better performance for specific gpu apparatus.
2.0
This version adds the function of importance sampling. To make the integration more suitable to very deep and high peak integrands, a heuristic tree search method is applied so that the domain can be amplified again and again.
1.0
This version only accomplishes the random sampling.