Sebastian Aaltonen
Every system has limited amount of memory. This implementation wastes a lot of memory. Allocator overhead, allocator fragmentation, unique_ptr overhead. An optimized version scales to larger numbers in actual hardware. Uses less memory and runs much faster.