More PLL?

[cje]

I've got a Q6600 that needs a lot of volts to be stable at 3.6 (1.58vcore). On air, that can be a bit toasty at times. is there anything I can do to lower the dependence on vcore? Like increasing PLL voltage?

It's a Q6600 SLACT G0, so I am unsure what PLL would be safe, and if it does anything. I've read conflicting arguments wether it actually does something or not.

 

[vertica]

not sure about pll voltage, but getting the gtl lane ref# voltages should help reduce vcore.

can only remember the calculation for 45nm tho, give us i sec and i'll have a look for the 65nm, its similar but subtly different.

 

[vertica]

Stolen from another forum:

for 65nm chips equation:

Vtt x 0.667 = (Vtt x 0.635) + Y

where Y is your gtl ref voltage, (vtt is also know as fsb termination voltage)

rearranged this gives:

Y = 0.032 x VTT

notes:
y comes out in volts, most bios require the input in mV multiply the answer by 1000 to give you input.
Also it is more than likely that you wont be able to match y exactly in the bios so get as near as possible.
Set all four (for a quad) chip lanes to the answer above.


for 45nm the +Y is on the other side so Y = -0.032 x VTT (gives a -ve value) but otherwise the same.

enabled me to drop vcore by 0.05V and load temps by 5C accordingly

 

[-)G(-Sawyer]

I tried 3.6 GHz on my CPU and to honest I didn’t see any real world gain from it, sure benchmarks scores went up but so did the temp and voltage.

 

[cje]

Yeah. I'll have to run some games benchmarks. Because I can run 3.2 Ghz at near stock voltage (1.35). Much cooler.