How hot should a Q6600 G0 be on a Tuniq Tower at idle, mine seems toasty.

[Cob]

I'm sure it does affect lifespan, but unless you're pumping 1.7v through it 24/7 I doubt it affects it too much. At 1.5v it might last 10 years instead of 15 years. Ok that's a third, but I'm really not worried in the slightest if it's still working in 10 years.

Strictly speaking CPU clockspeed should win every time, but raising the CAS to 6 may well negate the faster clockspeed at those settings. You'll need to benchmark them to see.

 

[Cob]

edit: the prior just failed prime95, shall i lossen the timings some more? up the ram voltage? it's already at 2.0v, or shall I just leave it at 3.24?

cheers mate.

Give the CPU a touch more vcore.

 

[yantorsen]

Give the CPU a touch more vcore.

i dunno how you change the vcore, its just on auto atm.

but i upped the ddr2 voltage and it seems to be doing the trick, just gonna see if i can get those timings down.

edit: p95 just failed, apparently i have a fatal hardware failure :S

 

[Cob]

i dunno how you change the vcore, its just on auto atm.

Tut-tut!


What does CPU-Z say the vcore is at idle at 3.33ghz? And which board is it?

 

[yantorsen]

Tut-tut!


What does CPU-Z say the vcore is at idle at 3.33ghz? And which board is it?

1.328v idle at 3.29ghz.

and the board is a P35C-DS3R

edit: and 1.296v load :S, that seems odd.

 

[Cob]

Vdroop causes the difference under load. And that 1.328v at idle is probably nearer 1.4v in BIOS due to the usual bad calibration/vdrop.

CPU Voltage Control is the vcore setting in BIOS. Set it to 1.4v and see what CPU-Z says once you get back into Windows. If it's equal to or less than 1.328v then set it to 1.425v.

Also change the memory option in the MIT (overclocking) menu that has Standard, Turbo and Extreme as its choices to Standard.

 

[SkeeterUK]

Should stick with the 3ghz you got, 100 more mhz aint gona do squat for performance anyhow. Only be worth it if you can go from 3ghz to 3.3ghz.

 

[yantorsen]

Vdroop causes the difference under load. And that 1.328v at idle is probably nearer 1.4v in BIOS due to the usual bad calibration/vdrop.

CPU Voltage Control is the vcore setting in BIOS. Set it to 1.4v and see what CPU-Z says once you get back into Windows. If it's equal to or less than 1.328v then set it to 1.425v.

Also change the memory option in the MIT (overclocking) menu that has Standard, Turbo and Extreme as its choices to Standard.

yeah just tried that and it failed prime95 after 8minutes at 3.29GHz.

I'm just priming it at 3.26, timings 5-5-5-18. ram overvoltage +0.1v

 

[yantorsen]

Should stick with the 3ghz you got, 100 more mhz aint gona do squat for performance anyhow. Only be worth it if you can go from 3ghz to 3.3ghz.

do you think it would be more benificial to just try and tighten the timings up maybe? to 4-4-4-15 maybe?

 

[SkeeterUK]

Yeah probably go back to 3ghz, and try tighter timings on the ram, you may need a slight increase in ram volts tho, just know what its limits are before doing that tho (if its even needed).

 

[yantorsen]

hmmm well it's about 1hour stable so far with settings in sig, and timings of 5-5-5-18, and ram at +0.1v

cob seemed to think that cpu speed would help more than ram timings, I'd imagin the difference in negligable anyway i guess.

 

[Cob]

CPU speed helps to a point. But raising the timings to CAS 6 slows things down until you gain a bit more CPU speed.

Strictly speaking CPU clockspeed should win every time, but raising the CAS to 6 may well negate the faster clockspeed at those settings. You'll need to benchmark them to see.

 

[yantorsen]

CPU speed helps to a point. But raising the timings to CAS 6 slows things down until you gain a bit more CPU speed.

it's running at 5-5-5-18 which isnt quite CAS6

so basically its between, 3.26ghz 5-5-5-18, or 3ghz 5-5-5-15

i dunno how much difference timings really make, so i'll ask a more direct question, which one will give me better in performance in games?

 

[Cob]

Well to get above 3.26ghz you were going to have to loosen the timings.

3.26ghz 5-5-5-18 will be faster than 3ghz 5-5-5-15. The last timing is the least important by miles.

 

[yantorsen]

so if I could get it to 3ghz 4-4-4-15, would 3.26ghz 5-5-5-18 still be quicker?


and would it be worth going to 5-5-5-20 to get a higher clock say?

 

[Cob]

It's very unlikely your memory will do 4-4-4-12 at 3ghz. But 3.26ghz 5-5-5-18 would still be quicker.

5-5-5-20 won't get you a higher clock. The last timing makes very little difference.

You've hit your memorys wall at its current voltage. I wouldn't loosen the timings any lower than 5-5-5-18. Have you tried 2.0v?

Otherwise your CPU is as fast as it will go with your current memory.

 

[yantorsen]

memory is at 2.0v, as my mb overvolts ram by 0.1v above what the bios says, i know you'll ask why or how i know this, and i can tell you if you want, but its a longish story.


cheers for the help mate

 

[tommysdad]

Also change the memory option in the MIT (overclocking) menu that has Standard, Turbo and Extreme as its choices to Standard.

Does standard setting make the system more stable when overclocking ?.

tommysdad

 

[Cob]

Does standard setting make the system more stable when overclocking ?.

tommysdad

Depends. Setting the option to Turbo or Extreme tightens some hidden memory/northbridge timings and increases memory bandwidth quite substantially.

But if your memory is clocked to near its limit on the current timings (4-4-4-12/5-5-5-15/whatever) then Turbo and Extreme will start to cause instability a lot sooner than Standard.

 

[tommysdad]

Thanks for the explanation Cob, I got my cpu stable. I`ll have a crack at the memory now

tommysdad