Divider kan give tab i benchmarks, men i reallife vil et højere overclock med divider IMO oftest give et bedre resultat end et mindre overclock 1:1.
I skal jo også lige huske på at vi i dette tilfælde ikke sammenligner 1:1@2600 Mhz med 5:4@2600 Mhz. Med #0's ram snakker vi jo med en 3000+ Venice 9*250 Mhz (1:1) = 2250 Mhz mod måske 9*300 (6:5) = 2700 Mhz (ramhastighed på 300/(9*6/5=10,8~11) => 2700Mhz/11=245 Mhz) ;)
Jeg vil vædde min pensionsopsparing på, at det sidste owner det første i alle reallife apps... :)
I øvrigt er her lige et citat fra overclocking.com:
<i>"Ram dividing
Unlike in Athlon XP and Pentium 4 systems, Athlon 64 take absolutely NO performance hit when using a memory divider, so don’t be afraid to divide that memory when you hit a wall with your ram.
The real speed of an A64 speed is the clock of the CPU, every other value is derived from this. So if your CPU runs at 2200Mhz (11x multiplier) the bios devides the CPU clock by the multi (2200/11) and you have the ram speed. This means that the ram on an A64 is ALWAYS divided. So placing it under (for example) a 200:166 divider means you “slow down” the ram a bit more. But now if you speed up the HTT so that your ram is running at 200Mhz again, your CPU will be overclocked. And pretty much the only thing you did was change your multiplier up. To about 13.2 to be exact, if you’re a fanatic you can try to recalculate that"</i>