Concerned with Temps

Associate
Joined
14 Nov 2007
Posts
85
I've recently started to attempt to overclock my computer abit more (in sig) I before had it stable at 3.1Ghz (3105 = 345x9) My PWM temp was pretty high back then but as far as I were concerned at a safe temp... But I can't remember off hand at this moment. Anyway I've currently upped it, by pushing the bus speed to 378x9 (3402Mhz) meaning I've had to increase the voltage multiple times in the BIOS, I've been running OCCT, Orthos & 2 instances of Prime95 (Total of 6 threads altogether) I'd ran it for 6 hours yesterday before sleeping when I woke up I had found it had once again restarted so I bumped up the voltage again to like 1.43 or 1.45v in the BIOS currently CPU-Z is showing a jump between 1.360 & 1.390v the VCore. So far the stress tests have been running for 9 hours & I've to say that the temperatures are somewhat making me twitchy but I can't find a single straight answer where I look, as even people with Q6600's just say "My CPU is *C" * being a the value, without going into core temperatures, my actual CPU temperature in general is reading 49-52C in Everest & has been around that temp for a few hours now. What is more worrying me however is the fact that the Core & PWM temps are a fair bit higher...

Core1 & 2 are currently: 68C (And have been around 64-68C for a few hours.)
Core 3 & 4 are currently both 61C but they've been jumping around from like 57C to around 63C.

I'm somewhat worried about that, I'd assume I should be more worried about core temps than actual CPU general temp I'm guessing but not 100%, also my PWM temperature has been jumping between about 71-77C after this many hours, but I don't know a truly safe temperature for it. Would any of you say these temps are normal or need to be cooled down? I am wondering mostly however is that heat = decreased lifetime performance, would these temperatures lower the life too much? (Or even much at all?)

Last question I need help with is Voltages in the BIOS, Do ICH, ICHIO, MCH or similar voltages need to be increased when overclocking at all? As to be on the safe side I've notched those voltages up a little aswell when attempting to see if they would help stability however if they do NOT need to be I'd like to turn them back down as they are bound to be creating a little more heat than necessary. I've been searching hours for answers & everything I've read so far has either been confusing or vague... Thanks for any help you guys can give me.
 
Soldato
Joined
17 Aug 2007
Posts
2,896
Location
London, Moscow & Samara
Last edited:
Associate
OP
Joined
14 Nov 2007
Posts
85
Well Core Temp & Everest have in the past had both the same temp readings & after so long core temp seemed to crash on my system (only the program not the computer), the Prime95 I have only stressed 1 core at a time ever... I've always been wondering how people have it working on multiple cores.... Actually just trying Core Temp stays open for some 30 seconds before an unhandled exeption & crash, but I did get screenie in before it happened, took quite awhile to get this up since I didn't want to stop the stress tests just yet, so everything feels super slow on it! (Prime95 isn't crashed, it just whites out like that when I start to open another window but eventually if I leave the comp it'll show that as per usual.)

 
Associate
OP
Joined
14 Nov 2007
Posts
85
I'm sure that is where I got the prime I got, but I will download & reinstall anyway as a double check.... Also I am dual-booting, I can boot into Vista however so far I've been booting into Windows XP for the stress testing.
 
Associate
OP
Joined
14 Nov 2007
Posts
85
Ok using XP only open prime and no other program select the above as i mentioned in post 4. Post back with a screen shot whilst you are running prime leave for about 30 seconds before taking the shot. I need to see if the problem is still there as you describe

prime down load link from here ftp://mersenne.org/gimps/p95v254.zip

Thanks for the link, seems I was downloading different version or something, Now I've started it up & just 1 prime stressed all cores as you say.

Ouch, Now it's warmer! I'm somewhat not suprised in 1 way, suprised in a bad way in the otherway, As said before I had it at 3.1Ghz abit lower voltage however even at full load the temps stayed quite abit lower than what they are suddenly reaching.

 
Soldato
Joined
17 Aug 2007
Posts
2,896
Location
London, Moscow & Samara
Ok excellent stuff;),

(Leave prime on one side of the page and the others CPUZ, Core temp on the other side. Just makes it bit more neater to read off your desk top;))

Back of the v-core notch by notch ;).

Then run prime for stability. Doing it this way will lower your temps;)

If prime detects a failure on one of the cores you will need to up the voltage again and re-stress.

Make a note of your present VCORE this will be your guide
 
Last edited:
Associate
OP
Joined
14 Nov 2007
Posts
85
Ok excellent stuff, now if you back off the CPU core voltage by one notch as your temps are a bit high and re-run prime and core temp including CPUZ. close everest .

(Leave prime on one side of the page and the others on the other side. Just makes it bit more neater to read off your desk top)

By backing down the v-core will lower the temps, as prime runs using longer iterations the temps will go over 70, but you will need to ensure stability using prime.

Alright, I'll try that... If you could give me a quick guideline I'm in the BIOS at the mo (I have 2 comps setup next to each other so very convenient actually) currently I have each set

CPU Core Voltage: 1.4500 V
DDR2 Voltage: 2.100 V (GeIL's recommended voltage)
CPU VTT 1.2V Voltage: 1.27 V
MCH 1.25V Voltage: 1.37 V
ICH 1.05V Voltage: 1.12 V
ICHIO 1.5V Voltage: 1.60 V
DDR2 Reference Voltage: 0%
CPU GTLREF 0&2: 67%
CPU GTLREF 1&3: 67%

As said before I upped MCH, ICH & ICHIO a little before to attempt stability & just haven't changed it back, I've also upped CPU VTT a little aswell, Should I turn them back down? I don't know how much they are needed to be raised or not. (Btw that is exactly how it is listed in BIOS)
 
Soldato
Joined
17 Aug 2007
Posts
2,896
Location
London, Moscow & Samara
Ok i am not totally familiar with the IP PRO Bios as i use a 680i motherboard.

At the moment i would just reduce the CPU core voltage, as that is the main temp zone as core temp reads the temps directly from the chips die (DT).

Leave the other settings for the moment and concentrate on the CPU core voltage. Attain lower temps by reducing the VCORE.

I am sorry that i can offer no further advice regarding the other settings you point out.

Although restoring them to their original settings or "auto" should not pose a problem as your main problem is clearly high CPU voltage and high temperatures associated with it.

The MCH voltage or commonly referred to Northbridge voltage as you are not using 4gigs of memory it can be lowered to its original setting.

Your memory voltage is fine.

In the mean time i will go read up on this Bios and hopefully expand my knowledge


You may find this thread useful for information

http://forums.overclockers.co.uk/showthread.php?t=17739133&highlight=Official+Abit+pro
 
Last edited:
Associate
OP
Joined
14 Nov 2007
Posts
85
Ok i am not totally familiar with the IP PRO Bios as i use a 680i motherboard.

At the moment i would just reduce the CPU core voltage that is the main temp zone as core temp reads the temps directly from the chips die (DT).

Leave the other settings for the moment and concentrate on the CPU core voltage. Attain lower temps by reducing the VCORE.

I am sorry that i can offer no further advice regarding the other settings you point out.

Although restoring them to their original settings or "auto" should not pose a problem as your main problem is clearly high CPU voltage and high temperatures associated with it.

In the mean time i will go read up on this Bios and expand my knowledge

Alright I've put them back to their stock values which are the lowest values, except for CPU VTT 1.2V Voltage which is up from 1.20 to 1.23 (still lower than 1.27).. I've put VCore to 1.4400 V & Will try it there, The reason the voltage was up at 1.4500 (Yea it were actually that high) is because the last attempt at stress testing the computer must have restarted sometime between 6 hours & 13 hours, I went to sleep just when it was about 6 hours & woke up some like 7 hours later to find it'd restarted, it'd been doing the same but quicker earlier when testing it at lower voltages.

I think it's easy to say that temps aren't much lower currently, although it didn't take long for 2nd thread of prime to fail although I had forgotten to change to round off checking to on.. (Note: I have everest on to check PWM temp) Screenie:



I've ticked "Round off checking" now, Next step?
 
Soldato
Joined
17 Aug 2007
Posts
2,896
Location
London, Moscow & Samara
Last edited:
Associate
OP
Joined
14 Nov 2007
Posts
85
Alright, I also see I had MCH set to absolute maximum, I feel somewhat lucky I didn't attempt to put it higher! :D I've been running Prime with round off checking since the last post & it's been ok, but I will try your suggestion, as it feels like that would be the better idea... I will edit this post when it's done.

Edit: :o Computer crashed before I could shut down, anyway, I've had to set voltage at 1.3000 as that is the lowest setting. OK, Well as soon as I started Prime to stress, blue screen auto restart.
 
Last edited:
Associate
OP
Joined
14 Nov 2007
Posts
85
Well it is somewhat ironic, but with me setting the voltages back it's nearly all back to defaults the only things aren't are USB Legacy at BIOS instead of OS or something like that, It is getting pretty late, so I will continue tomorrow after a quick look at what I can change & hopefully I'll find a nice middle ground again, I have a profile saved of my last overclock... so if I need I can just load my last stable overclock back up.

Thanks much for your help & spending the time to keep replying... I will continue this thread tomorrow/in future if I need too again.
 
Associate
OP
Joined
14 Nov 2007
Posts
85
o/cing at 3 in the morning - hardcore!

:D Or Crazy, 1 or the other! :)

Looking forward to an update from Yiuca, hopefully those temps are down
I've not got them down yet however today I haven't been able to do much since I've been so busy, well in fact it's about 6Am now after being up all night & all of yesterday & I've not finished sorta shows how it's been...

I yesterday before I went to sleep took a look at my voltages when overclocked to 3.1Ghz (345x9) the voltage for vcore in BIOS is 1.35v with vdroop in effect (under load) the vcore is around 1.280v - 1.300v so I'm currently wondering why so stable with that voltage when 300Mhz rise makes me have to put the voltage up & hence the heat up so much? Unless I was just unlucky with stability tests last time I ran through the voltages, I think I will try that when I next get the chance, I'll loosen RAM timings to help make sure, & try stressing it again incrementing in voltage by 1 increment each time until I get stable & hope I can get stable on lower voltage thus lower heat. (For comparison sake, the difference between 3.1Ghz & 3.4Ghz temps for me is that 3.4Ghz temps are a full 10C+ hotter at full load, seemingly because of the voltage needing to be applied is quite abit higher I think.)

Edit: I've just been messing around with voltages before getting some sleep, making my way up incrementally through the voltages & as far as I'm aware I've just found the temp jump, from 1.4100v (in BIOS) to 1.4200v the temps seem rise several degrees while under load, unfortunately voltages under & even at 1.4200v are NOT stable for me to retain for long periods of time, it takes a little while but eventually the comp does decide to "give up" as per usual. From voltage change of 1.4100v & 1.4200v I see approx 8C change in core temp & 12C change in PWM temp... Massive rise in temperature & I have no clue why.
 
Last edited:
Associate
Joined
6 Dec 2004
Posts
1,293
Location
Scotland
Hi guys, can I add a couple of things, or rather ask advice? I too have the Q6600 chip and have overclocked it by increasing the board multiplier from 8 to 9. I have been running the Mersenne prime program all night and the temps have settled nicely.

My temps have settled between 57C and 58C which seems rather warm but I am on a stock cooler. Are these temps OK or am I going to have to get a bigger cooler?

I am using Core temp -.95 as I have seen suggested on this thread but already have Speedfan running. Now Speedfan give temperatures about 12C lower the care Temp. Is there a way of knowing which is right?

Thanks for any help.
 
Back
Top Bottom