Not saying I get this whole CRC thing, just looking at the 2 snippits, the biggest thing I see is you have a nested loop and Darrel doesn't. But that seems like it would change things by more than double. (approx difference between the 2)

Try using his poly instead of yours in your code to see if that changes your timing. I don't know if you can just do that, but if so it will eliminate that as a factor.