C for loop indexing: is forward-indexing faster in new CPUs?

Question

On a mailing list I\'m subscribed to, two fairly knowledgeable (IMO) programmers were discussing some optimized code, and saying something along the lines of:

Answer 1

When optimizing loops I'd rather look into loop unrolling (as it cuts down the number of comparisons vs. the exit value, and it may be optimized for parallel processing (MMX) depending on what goes on inside the loop).