The subtimings would have an effect for sure (the ram used can affect this due to the SPD timings of each set of ram). I did do some comparisons changing subtimings using memset and I only found a difference of +/- 100 MB/s changing everything I could in memset.

The results show a rough comparison, but you are correct in saying it would be best to set everything the same on the 2 boards, and to be certain, use the same ram for the tests.