An 8 mb/s difference is statistically well within the standard range of deviation for most RAM when it comes to benchmarking bandwidth. In other words, your difference is so statistically insignificant, it means nothing. Even if you're only running DDR 2100, you should get a theoretical bandwidth of around 1900-2000 mb/s. An 8 mb/s difference constitutes a .004 % of a difference in performance. Moving your mouse around during benchmarking would probably cause a bigger difference as Sandra scores ALWAYS vary a little bit between tests.