I was told that using memcpy from string.h is faster than CopyMemory from windows.h. Is this true?

memcpy() is in windows.h (or stdlib.h, I forget which).  Unless you're doing something which would require absolute speed (like calculating intercept vectors for missiles), it really doesn't matter.

When you use memcpy it will typically be inlined and optimized by the compiler, while CopyMemory might not be. That's a possible explanation for a performance difference between them.