Prime Currently running temps @71C - need quick reply!

rob_cunningham

Distinguished
Nov 12, 2008
89
0
18,630
Hi,

I'm trying to test if my OC is stable on my Q6600 which is at 3.6ghz. I've just managed an hour (and its still running now!) without any errors.

My Question is should i abort becasue my core temps are sat around 70C? I don't want to cause any damage to my cpu. My Ind' core temps in prime are:

Core 0: norm' 70-71, max 73
Core 1: norm' 70-71, max 72
Core 2: norm'66-67, max 68
Core 3: norm' 65-66, max 67

i've read that my threshold on a q6600 is 71C and that any prolonged use can damage the chip.

How long should i carry on testing in prime for at these temps?

I do have all fans on max and CPU internal Thermal Control enabled.

I would be really appreciate a quick response.

Thanks,

Rob.

 

rob_cunningham

Distinguished
Nov 12, 2008
89
0
18,630
I have an Asus Silent Knight II.

After Prime failed HWmonitor would read any temps or volts and system wouldn't shut down, i had to use reset button.

Reboot failed so i allowed to cool down before trying again. I am back in and everything seems fine.

Do you think it faile due to temp or lack of voltage? if the latter then i guess 3.6 may be too much with current cooler.
 

rojito

Distinguished
Jan 5, 2007
209
0
18,680
I would suggest you to find a stable speed with coretemp always below 71º. If not possible with 3.6 GHz you should lower your speed.
I'm not sure you can get 3.6 on a Q6600 with that cooler, I'm already impressed you managed to run prime for more than an hour at that speed. Are you using stress test (the one that says no memory tested) or the blend (some or lots of memory tested)?
 

rob_cunningham

Distinguished
Nov 12, 2008
89
0
18,630
I ran Prime using large ffts it says it produces max heat, little ram tested.

Do you not rate my cooler? I was recommended to get this one. I thought it was good because it is one of the few that are 100% copper. i'll see if i can lower my voltages further and then re-test until i can keep all cores below 70C


 

rojito

Distinguished
Jan 5, 2007
209
0
18,680
It seems to be a good cooler, but getting 3.6 on a Q6600 is for the ones that use 120mm fans (and only a few of them).
 

htoonthura

Distinguished
May 21, 2006
651
0
18,980



what is your vid and your current voltage?
 

rob_cunningham

Distinguished
Nov 12, 2008
89
0
18,630
VID= 1.250v

These are my voltage settings:

CPU = 1.4375v
PLL = auto
VTT = 1.46v
Mem = 2.1v
NB = 1.52v
SB = 1.55v
1.2 HT = 1.25v
GTL_REF = +50mv
Loadline is enabled.

I'm running 400fsb x9.

Any suggestions would be great thanks.

 
G

Guest

Guest
Seems like its too high for you... I for one have seen max temps... under 12 hour prime stable on my own q9550 at OVER 80 Celcius... way to high for me... left it on over night and i came back in the morning and saw it I was like sh*t!!!!.... apparently no dicernable harm came from it though so I backed down on everything and now i'm at 3.85 ghz at 1.25 volt core with max load 71 C in the summer so now i'm happy

BTW for stability always run small fft first and then run the Blend test and then the Large FFT... thats the order I do my tests in.... if you can get small fft stable but can't pass blend that means A. Your ram is going to fast, or B. you need to raise northbridge voltage and then for large its just a test to make sure under max temperature it still runs ok
 

rob_cunningham

Distinguished
Nov 12, 2008
89
0
18,630
just dropped cpu voltage down one click (1.43125v) and raised PLL up one (1.5v) and its looking ok.

You are probably right its too high for a stable OC but i am a persistent "bar-steward".

I am going to see how low i can get my cpu voltage before i give up.
 
G

Guest

Guest
1.4 volts for 3.6? thats seems high to me... I'd think 1.38 or something would be ok for a q6600 @ 3.6
 

rojito

Distinguished
Jan 5, 2007
209
0
18,680
As I said before, I don't think that cooler can handle 3.6 GHz, although for a 92mm one, its very good.
Remember that just a few apps are able to stress your cpu nearly as much as Prime, so if you're not using any of them, you may keep 3.6 GHz, but for your safety I would lower the speed to 3.4GHz or buy a very good 120 mm fan CPU Cooler. There are some no to expensive that use direct touch technology that should be able to let you stay at 3.6 with temps below 67
 

CompuTronix

Intel Master
Moderator
Guys,

Please read: Core 2 Quad and Duo Temperature Guide - http://www.tomshardware.com/forum/221745-29-core-quad-temperature-guide


Section 6: Scale

Scale 2: Quad

Scale 2: Quad
Q9x50: Tcase Max 71c, Stepping E0, TDP 95W, Idle 16W
Q9x50: Tcase Max 71c, Stepping C1, TDP 95W, Idle 16W
Q9400: Tcase Max 71c, Stepping R0, TDP 95W, Idle 16W
Q9300: Tcase Max 71c, Stepping M1, TDP 95W, Idle 16W
Q8200: Tcase Max 71c, Stepping M1, TDP 95W, Idle 16W
Q8200: Tcase Max 71c, Stepping R0, TDP 95W, Idle 16W
Q6x00: Tcase Max 71c, Stepping G0, TDP 95W, Idle 16W (Q6600 G0)

-Tcase/Tjunction-
--70--/--75--75--75--75-- Hot
--65--/--70--70--70--70-- Warm
--60--/--65--65--65--65-- Safe
--25--/--30--30--30--30-- Cool


Tcase = CPU temperature
Tjunction = Core temperature

The Guide shows how to calibrate your temperatures, and will bring you up to speed on how it all works. Also, Intel's Thermal Specification, shown in their Processor Spec Finder - http://processorfinder.intel.com/details.aspx?sSpec=SLACR - is for CPU temperature, not Core temperature, which is a common misconception among many users.

As you may not be aware, Prime95 Small FFT's is the standard for CPU thermal testing, as well as CPU stability testing. Small FFT's provides a steady-state 100% workload for the processor Cores, while Large FFT's and Blend do not. The description for Large FFT's is misleading, since it in fact does not produce 100% CPU workload and heat, although power consumed in Watts at the power cord will be slightly higher due to higher memory workload. The CPU temperature and Core temperatures will be ~ 4c higher using Small FFT's. Try it and see for yourself.

In order to compare apples to apples, thermal testing should be done as close to 22c ambient as possible. If ambient is higher or lower, then simply add or subtract the difference from your test results, so that your temperatures are adjusted to reference 22c, which is specified as the "standard" for ambient. Case covers should be removed, and all fans should be at 100% RPM. Since thermal saturation is typically reached within 7 to 8 minutes, a 10 minute test run with Small FFT's is sufficient. For stability testing, 8 hours of Small FFT's is the minimum, and 24 hours is considered proof of stability.

Currently the 2 best air coolers available are:

(1) Sunbeamtech Core Contact Freezer - http://www.newegg.com/Product/Product.aspx?Item=N82E16835207004
(2) Xigmatek HDT-S1283 - http://www.newegg.com/Product/Product.aspx?Item=N82E16835233003

Google the reviews and see for yourself - http://www.frostytech.com/articleview.cfm?articleid=2271&page=5


Hope this helps to clarify these issues,

Comp :sol:
 

rob_cunningham

Distinguished
Nov 12, 2008
89
0
18,630
Hey Thanks for the info. There is a lot of interesting views. Its a little reassuring the post from Compu that i didn't exceed too high temps on my testing - although it still wasn't great.

Also Thogrom, do you really think 1.38v is realistic to be stable at 3.6ghz? if so then there is hope for me yet. And do you mean 1.38v set in the bios or as CPU-Z reads? you can see that the vdrop on my board is quite high - running at 1.4375v in bios gave 1.376 in CPU-Z.

I think you are probably right that i would need to upgrade my cooler if i want to get 3.6 at a sensible temp but in the meantime i am only going to lower my voltages and see what i get. I'll keep an eye on any Prime tests just incase temps get too high again.

Thanks again from all you.

I'll post any news of sucess or probably failures.

 

Liderc

Distinguished
May 12, 2008
251
0
18,780
For a 1.25 vid you shouldn't need 1.40+ for 3.6, I know each chip can vary, but I would imagine you should be able to get 3.6 with 1.35ish. Lupiron can help you with Q6600's but I'm betting he tells you to get a better cooler
 

rob_cunningham

Distinguished
Nov 12, 2008
89
0
18,630
Tried contacting Lupiron a couple of times becasue he has the same board as me and had the same chip. unfortunately i haven't heard from him. I've read all his posts on the P5N72-T mobo but he only hinted at a few of his settings.

If there's a possibilty of acheiving a stable 3.6 below 1.4v then this itself could solve my cooling issue.

I'm a little reluctant to change my cooler as i have only had it a couple of months.

I'm still very much a "noob" to overclocking and i do take on board all of your comments and advice.

Thanks again.
 

rob_cunningham

Distinguished
Nov 12, 2008
89
0
18,630
Hey,

I running again at 3.6ghz with now only 1.35v on the CPU.

My settings are:

CPU = 1.350v
PLL = 1.58v
VTT = 1.38v
Mem = 2.1v
NB = 1.52v
SB = 1.60v
1.2 HT = 1.40v
GTL_REF = +50mv
Loadline is enabled

Before i disappoint myself by running Prime and watching it fail (again) i thought i would just try and see how the system runs first. Like Rojito said I am not going to be using any programmes that can stress my pu as much as Prime will.

Before reaching these settings I did run Prime using small ffts. my voltages were as above expcept PLL was on auto and 1.2HT was on 1.30 & 1.35v and both BSOD in about 10sec. Is that an indication of voltages being too low somewhere? If so, does anyone know which is the likely culprit?

BTW my Vcore in CPU-Z is now only 1.280v when set at 1.35 in the bios. Loadline seems to cure any/most vdroop (is that one correct?) when system is stressed.
 

rojito

Distinguished
Jan 5, 2007
209
0
18,680
1) I would say, forget (for now) about the difference between mobo and CPUZ voltages, and use your mobo's for reference. The good thing is you can load windows with that setting, thus making possible the testing.
My advice would be to use all the settings you had when you had an almost stable and overheated CPU but keep the 1.350v. Load Prime and test, if it fails take it up a bit and retest, and so on until you get a stable setting. Then try lowering other voltages like your NB's.
2) About the difference between Mobo's and CPUZ's voltage reading, probably CPUZ is telling the truth, but that's just a guess.
3) I repeat, it`s almost impossible to get 3.6 (at decent temperatures) out of a q6600 with even the best 92mm cpu fan and keep in mind that with time dust reduces the cooling capability of your system and temps could reach some degrees higher, so make sure you can't stay at about 3.4 instead of 3.6. The difference on performance will be much much less than the idea of using a number like 389 (fsb) instead of a round and pretty 400
 

rob_cunningham

Distinguished
Nov 12, 2008
89
0
18,630
Thanks. i'll certainly give it a go. I'm useless at the math for working out clock speeds from the Fsb and multiplier. Am i assuming you are suggesting an fsb of 400 and reduce the multi to 8.5? becasue i don't have the option to use 1/2 multipliers.

otherwise i guess i would be dropping the fsb.
 

rob_cunningham

Distinguished
Nov 12, 2008
89
0
18,630
Hi,

I think 3.6 may be out of my reach for the time being. I tried what RoJito suggest and used same voltages as my best Prime run but with CPU voltage of 1.35 and then creep up until stable. This was working up until about 1.3625v, it was gradually getting better results in Prime but then anything higher wouldn't even boot. I kept creeping up until 1.38125v and still wouldn't boot. I figure much higher than this and i'm back into the high temps again.

I've just had a go at 3.4GHz using 425 x8. Prime is currently running and "looks" like it going well. Max core temps are between 57-63C.

Is there a performance difference between the same clock speed but using lower multipliers with higher fsb?