经MMX, SSE 2优化后的MEMCPY,比VC中的MEMCPY快不少-as MMX, SSE, SSE2 optimization of MEMCPY than the VC lot faster MEMCPY

memcpy函数测试，在linux和VC编译器下面运行，效果不一样（不能重叠copy）。linux内核memmove函数实现。-memcpy function tests, and VC compiler in linux running below, the effect is the same. linux kernel memmove function implementation.