SuperPi 32M mainly, the difference might begin to show up in 3DMark01 if you're really squeezing for points. Tighter CAS helps a little in PCM05.
CAS 4 Vs 5 and the MHz difference that comes with it... for most benches you can write any effects off as test discrepency, the differences are probably that small.
(especially true for C2D chips)
Been thinking about doing a chart with different timings and MHz to show this, but my GPU is holding me back.