Let me repeat: triple-channel memory at 1066 MHz has the SAME BANDWIDTH as dual-channel memory at 1600 MHz. So, in the OP's case, his chosen memory configurations will make absolutely no difference in any workload except the fact that his i5 system has 2GB more.
I`m to guess your taking this from a theoretical point of view...? Again the smoking of crack has been performed...
Just ran Aida64 on both of my machines at the same clock speed and memory speed then even ran the 2600K with the memory at 1333mhz... Both sets of memory at the same timings... Note also the I7 920 was at 191fsb as opposed the 2600K was at its stock setting of 100...
2600K @ 3.8Ghz 1600mhz mem
mem read...18786mb
mem write..19229mb
mem copy...20825mb
mem late...46.4milsec
I7 920 @ 3.8Ghz 1530mhz
mem read...15618mb
mem write..13743mb
mem copy...20083mb
mem late...53.5milsec
Extra test...
2600K @ 3.8Ghz 1333mhz mem
mem read...16316mb
mem write..18588mb
mem copy...17771mb
*
mem late...54.1milsec
Only instance the I7 920`s triple channel memory beat the 2600K was with it running slower speed of 1333mhz in the copy part of the benchmark... BTW you dont want to see what the 2600K will do with it @ 4.6Ghz and the memory @ 1866mhz... It will flat clean the I7 920`s clock... Just for shits and giggles I may buy a 2500K just to see if there is much of a difference between it and my 2600K... (bet not much)
You just cant tell me better newer architecture wont prevail...? That extra channel is going through old school tech as are the other two...