Q6600 1.45 too much voltage?

chevmaro

Member
Dec 30, 2005
113
0
0
So I OC'd my quad and I can go 3.0ghz at stock voltage. But it seems once I get past that I have to be very generous with voltage. I got it running at 380x9 3.42ghz. But I cant get stable unless I huck a bunch of voltage at it. In the bios on my P35 board I have to run at 1.48 vcore. In windows with cpu-z it reads 1.45 at idle and 1.4 under full load. My temps look ok at 60-62 C but I am scared that I am overvolting the cpu. Is this too high? I ran prime for an hour and was fully stable. I got scared that I would be running too much voltage and then reboot and backed it down to 333x9.

Just wondering why it takes so much voltage to be stable at 3.4 and if it is too much voltage to be safe.
 

Idontcare

Elite Member
Oct 10, 1999
21,110
59
91
When you say "stock" voltage for 3GHz...do you mean to say you manually set the BIOS voltage to be equal to the CPU's VID or do you mean you leave the Vcore alone in the BIOS (in which case the BIOS is likely ramping your CPU voltage for you as you overclock the FSB)?

I can say two of my G0's can do 3GHz at "stock voltage"...but the truth is the mobo increases the voltage automatically as I increase the FSB. (mobo is DS3L)

Use CPU-z at idle and loaded in Windows when you are running "stock" voltage at 3GHz. Don't be surprised if you see it >1.3V in CPU-z.

The reason this is relevant is it will go a long ways to explaining why it appears that your CPU voltage takes a huge ramp to go from 3GHz to 3.4GHz. You perception of the voltage ramp it took to go from 2.4GHz to 3GHz is likely in error.

On two of my G0's (VID's are 1.2675 and 1.2750) the "stock voltage" the mobo puts to them when running Prime95 (small FFT) and be stable is around 1.3500V...nowhere close to their "stock" VID setting.

Edit: the analogy I wanted to include here is that CPU voltage and FSB is like SPD settings on your ram...at least it is for me and my DS3L's...clock the FSB up and if you leave your RAM settings to be determined by the mobo then the timings will be automatically relaxed for you (to a point)...

You can prove this to yourself by keeping the chip at 3GHz and manually setting the voltage in the BIOS to something around VID and then watch the sucker crash and burn in Prime95 small FFT.

And is 1.48V (bios set) too high for Q6600? I don't think so provided you are handling the temperature that the voltage is generating at your desired clock frequency then your chip will still live for 3-5 years (long enough to be obsolete when you replace it).

I have my G0's under Tuniq120's and I don't volt them higher than 1.40V (bios setting) or else my full-load temps climb above 69C (my arbitrary comfort limit)...but they run at 367MHz FSB (3.3GHz).
 

chevmaro

Member
Dec 30, 2005
113
0
0
Sorry, stock voltage meaning my P35 reads the default voltage in the bios as 1.325. It does not have an auto setting so it does not increase voltage with FSB. It has default or manual. I take the default number 1.325 and set it manual as 1.325. In Windows CPU-Z reads 1.25. Im just wondering if theres something im missing. It seems as if most quads dont run past 1.4ish volts. My temps are much lower then most at high voltage but I didnt think I would need this much to remain stable. I got to be missing something else and maybe the voltage is compensating for another area? If its not dangerous ill do it. It did feel fully stable at that voltage and if it passes prime after given a longer time.
 

ArchAngel777

Diamond Member
Dec 24, 2000
5,223
61
91
Provided you can keep the temperatures under control, you can safely move vcore 15% above VID.

If the VID=1.325 you can set it at 1.52. Keep in mind though, that without good cooling you will hit the thermal limit before the vcore limit.
 

chevmaro

Member
Dec 30, 2005
113
0
0
Prime after an hour only showed temps of 60ish. Do you think after an hour it would max out on temp? I know I need to run it longer to determine if its stable but I just didnt want to push it if I shouldnt be. It sounds like now though I might go home and try again with the higher vcore.
 

ArchAngel777

Diamond Member
Dec 24, 2000
5,223
61
91
Originally posted by: chevmaro
Prime after an hour only showed temps of 60ish. Do you think after an hour it would max out on temp? I know I need to run it longer to determine if its stable but I just didnt want to push it if I shouldnt be. It sounds like now though I might go home and try again with the higher vcore.

I think one hour is a great test for Prime95, provided you are not working with sensative calculations. If you are running programs to cure disease and all of that good stuff, then you need to run very long tests to ensure 99.9999999% stability. You don't want to be sending incorrect data to those programs/projects. So, if you are like me, and just want to play games, edit pictures, and read, you are perfectly safe to stop P95 after 1 hour and consider it stable.

As far as your thermals, did you take the temperature with CoreTemp? CoreTemp is the only program that I find to be 100% accurate. SpeedFan had (has?) an issue with detecting the proper core temperatures for the Quad Cores.

Be sure to keep the cores under 70c if you are someone who likes to be extremely safe, or keep them under 80c if you are like other people who believe the chip will long outlast it's usefullness. The choice is entirely up to you. Keep in mind, though, that while gaming you will never hit the temperatures that you do while running P95. In other words, P95 is very intensive and the closest thing to working as hard as it does would be encoding and rendering applications.

Good luck with everything!
 

heyheybooboo

Diamond Member
Jun 29, 2007
6,278
0
0
Originally posted by: chevmaro
Prime after an hour only showed temps of 60ish. Do you think after an hour it would max out on temp? I know I need to run it longer to determine if its stable but I just didnt want to push it if I shouldnt be. It sounds like now though I might go home and try again with the higher vcore.

Don't forget to account for lower ambient temperatures this time of year - unless unlike me you actually have heat in your room :D


 

chevmaro

Member
Dec 30, 2005
113
0
0
Room temp was about 70 F. I got the temps from both speedfan and coretemp. Speedfans core temps are lower but the actual CPU temp that speedfan reports is similiar to the core temps in coretemp. Each were reading 60-62 while running at the same time.
 

Idontcare

Elite Member
Oct 10, 1999
21,110
59
91
Originally posted by: chevmaro
Room temp was about 70 F. I got the temps from both speedfan and coretemp. Speedfans core temps are lower but the actual CPU temp that speedfan reports is similiar to the core temps in coretemp. Each were reading 60-62 while running at the same time.

Speedfan is off by 15C (too low) because it does not account for the G0 quads having a different TJunction than the B3's.

Graysky has a cool picture and explaination of how to configure speedfan to work correctly by way of adjusting the offset. It worked perfectly for me.

Here is Graysky's link, look about 1/2 down his OP for "Section 5 Required Software": http://forums.anandtech.com/me...=2057083&enterthread=y