Performance of copy algorithm

[Jerry Quinn]

Dhruv Matani writes:
 > A) Open coded loop on char			 7.20s
 > B) memcpy					 0.80s
 > C) Loop using doubles				 1.68s
 >  (cheats on alignment and block size)
 > D) memmove					18.96s
 > E) std::copy (which translates to memmove)	18.96s
 > 
 > I have just copy pasted this form a post in the past.
 > 
 > I'd like to say, that on my system (RH-9, 2.4.20-8), memmove 
 > is _*actually*_ faster than memcpy!!!!! That comes as a surprise to me.

What CPU?