Sign in with
Sign up | Sign in
Your question

Q6600 Stock Speed idles at 47C wtf mate?

Last response: in Systems
Share
September 20, 2007 6:31:10 AM

So I just got my new system up and running. My first system build I might ad. I have a q6600 running at the stock 2400Mhz. I went with an ArcticCooling Freezer 7 Pro with the stock 120mm fan that it comes with, and Arctic Silver 5. Coretemp and Speedfan are both recording an idle temp of anywhere from 46C-48C, and anywhere from 59C-60C under stress. None of my cpu settings have been changed, everything is the stock setting.

I think I used way way way too much thermal paste the first time I installed, so I dis-assembled and cleaned it off, then I started over. I used one line down the middle the second time, but that didn't change anything at all. Since then I've also zip-tied all my wires and cords out of the way so the air can flow better. I have a Lian-Li case with 2 front panel 80mm fans, an 80mm exhaust and a rooftop 80mm as well. Any ideas about why I am running so hot?

Sidenote: Within the next month or so, I will be investing in a Thermalright Ultra 120 heatsink, with plans to o/c to 3Ghz, but I need to get things cooler first. I've thought about lapping my processor, but I'm not sure I want to take that risk if I don't have too. I know most people lap the Ultra 120's, and I wouldnt mind doing that as its quite a bit cheaper, and as far as I can tell, harder to screw up as far as the electronics is concerned.
September 20, 2007 7:40:16 AM

Do you know what stepping Q6600 you got?
Sounds like you have the B3 stepping which ran about 45c or so.
The G0 runs about 10 degrees less.
Check it out and let us know. Not sure whether CPU-Z would give it to you.
September 20, 2007 8:00:44 AM

hi man,

i have q6600 G0 ... and a TR120 Ultra Extreme

my cpu is at 3.6ghz 1.52v idle .. 1.45v full load (big drop i know :(  )

anyway ... in idle @1.52v and 3.6ghz i have 37-38C

and in full load about 73-74C

and i dont think its perfectly seated.. cuz the cpu needs some lapping i think..

so ur temps are really high even for a b3 imo

cheers
Related resources
September 20, 2007 9:14:23 AM

though you mentioned efforts to improve case airflow... what is your ambient case temperature? if your ambient case temperature is even remotely close to 40C, then your cpu temp really isnt doing that bad, tbh (since youre only a few degrees above that at idle). cleaning out any and all dust from all areas inside your case with a can of air, and having non conflicting airflow paths helps alot too... as does having a room running at somewhat cooler temperatures; warmer room = warmer case and a hotter cpu

im sure the core marchitecture is able to withstand much higher temperatures than what youre experiencing currently (in excess of 80C or so). so no worries at all yet it seems.

also, as far as AS5, it takes up to a week to fully set in anyhow after being properly applied with included instructions [on AS site] followed till that point, so, temperatures will be a few degrees higher up until then, too.
September 20, 2007 9:57:19 AM

Nothing personal, but if this is your first build, you shouldn't be thinking about overclocking.

Your first post was probably "MY CPU is too hot".
Your second post will be "OMG did I fry something?".

When it comes to lapping, my thought is "Well, if we haven't had a need to do it for how many years, why suddenly THIS year we need it?" P4s ran hotter than Core2 Duos, yet we're so quick to say 'omg you need lappzorz!" You shouldn't NEED anything above and beyond what the computer comes with to get it working! AS5 can take a week or so to work, but even then you're only gonna see a 1-3C difference IIRC. Did you use AS5? Did you install it just like:

http://www.arcticsilver.com/arctic_silver_instructions....

?

September 20, 2007 1:16:46 PM

airblazer said:
Do you know what stepping Q6600 you got?
Sounds like you have the B3 stepping which ran about 45c or so.
The G0 runs about 10 degrees less.
Check it out and let us know. Not sure whether CPU-Z would give it to you.


CPU-Z said
"Stepping: B,"
"Revision: G0"

----------------
Midpost - Rant:

I'm not trying to pick fights, but every post you make cyberjock is negative. I understand ther are risks, but if you take them and are successful there are benefits to lapping and overclocking, and just because it's my first build and I'm concerned that it's running hot doesn't mean Im going to try and oc/ to 4Ghz and kill my computer, I'm not an idiot.

----------------

I did follow AS's guide when installing, but I did't realize how long it takes to ge the full effect. My room is probably 75F so that may have something to do with it, maybe not. I don't know my current ambient case temp, but as soon as I find out, I'll let you know. Oh, by the way, there's no dust in there yet, it's only about a week old.
September 20, 2007 1:44:15 PM

I assume this is with the intel's heatsink/fan.

EDIT: Oh AC 7.... Check it. Well, I would check the possibility of a convex/concave CPU...
September 20, 2007 1:59:26 PM

I have the thermalright ultra 120 etreme and I idle at 46-48C. I am overclocked to 3Ghz. I don't really see the need for anything faster.

For Slickku:
Did you lap your thermalright ultra 120 extreme? That is serious cool temp!
September 20, 2007 2:32:11 PM

Yeah slick, are you using speedfan for your readings? If you are, you need to adjust your temps by 15C, if not, you are a god among cpu coolers.

If my cpu is concave, I could lap it to smooth it out, but I've only ever seen people do that all the way down into the copper. Is this necessary, or would it be bad to just go into the nickel enough to be flat and smooth?
September 20, 2007 3:06:08 PM

What is your vcore? I have a B3 and the motherboards "stock" vcore was high. I lowered my vcore and overclocked at the same time at it was still stable. It is stable at 2.7ghz@1.23 vcore. Right now its at 3.0ghz@1.31vcore. Running a 120 extreme.
September 20, 2007 3:20:15 PM

Hey Frankenstyle - I have a Q6000 with a Thermalright Extreme 120 and it idles at 47c at stock speed. I have an Antec Sonata II case it idles at 39c with the side off. I have the older B3 stepping and my room temp is 78c (Arizona). I overclocked it to 3.4 on stock voltage (1.3v) but I got an error in Orthos so I didn't play any games or benchmark it. I left the side of the case off and the idle temp at 3.4 was 47c (same as stock with the side on). I don't want to mess with the cpu voltage adjustments yet and I don't need the extra horsepower right now so I changed it back to stock. I did however run some 3DMark06 benchmarks at 3.0 and 3.2 and they ran fine. I used Coretemp and Intel's TAT and the load temps were under 70c but I don't remember the exact temp. My B3 stepping processor overclocks well so your G0 stepping processor should do better.
September 20, 2007 4:31:59 PM

Ok, thanks. 78 in your room must suck. Maybe your used to it. Thanks for the numbers though!
September 20, 2007 7:55:51 PM

Wow I just checked my other post and I meant 78 degrees farenheight for my room temp not 78c. I am use to it and I can't afford to cool my house any cooler.
September 21, 2007 3:31:08 AM

Quote:
Well 80F in my room. My my E6300 @3.22 (1.325v) idles 48c and loads at 60c so Id say its not that bad. You have 4 cores so I wouldnt be too concerned, cooler temps are on the way. When we play games, the cpu never reaches full load, only gets to about 53-54c, so not concerned at all.


So with 4 cores I can allow it to run a little hotter and it won't matter so much? Im also not sure what you mean by cooler temps are on the way, but it makes me happy just to have hope! From what I have read and seen on the subject, the q6600 would probably be able to stand 70C, but I really don't want it to get over 60C. Would that be a safe assumption?

Quote:
Wow I just checked my other post and I meant 78 degrees farenheight for my room temp not 78c. I am use to it and I can't afford to cool my house any cooler.


I feel your pain. Last winter my house was 55F on a good day, and 80-85F all summer, but like the deadbeat student I am, I moved back home, so now I can blow all my cash on my computer and save on heat and a/c!
September 21, 2007 3:39:35 AM

core marchitecture can withstand easily over 70C (85C, i believe), so at 70C it is warm, but no dangers yet. at 60C is the same situation, 60C is even within a comfort zone, considering how many individual cores you have running. if it was a single core cpu, 60C could be considered too warm (considering the how much less it can even process simultaneously, P4s were examples of single core cpus that just ran too hot), but you have 4 times that many cores, so 60-70 really isnt all that bad.

with an X2 for example, 60C is safe to run at even, but again, still warm, but no dangers at that temperature.

a great way to lower temperatures though, is to reduce the voltages. voltages are the make or break deal for keeping something cool, or causing something to fry if theyre set too high... so, try undervolting. there are a few different applications you can download that will do that.
September 21, 2007 3:58:02 AM

Can't you change that in your bios? I think I may just have. I took it from 1.26 to 1.12, and I forget the term, but i'm registering on cpuz now that its even a little lower than the 1.12v. My temperatures aren't showing any signs of going down, if anything it is hotter now than before.
September 21, 2007 4:00:07 AM

reducing the voltages will definetly reduce the temperatures. even if you OC by 1GHz, your temperatures will most likely be lower than stock speeds, if your voltages are low enough. voltages are what drive temperatures up, not really clock speeds. you need a high enough voltage to maintain stability though at a certain clock speed, but the lower voltages the better.

for instance, my X2 3800+ @ 2.0GHz 1.35v will reach as high as 60C+ under load on a warm day. but if i reduce the voltages to 1.125v, the temperature reduces down to low 50s under load, sometimes high 40s, depending on ambient temperatures too. (1.15v is needed for absolute stability though under full load, but it also results in a slightly higher temperature too). at 1.1v though at stock speeds, my system will just freeze, because there just isnt enough current going to it. and at 1.6v+ my cpu will probably fry, and have tempatures well above 70C too. (its only rated up to 71C though, so itll probably either fry my cpu much above that or just crash my system, but hopefully crashing is all that would happen)

current OC using rightmark cpu clock utility, is an increase of 1 degree @ load (alternating 50 to 51C) going from 2GHz to 2.2GHz @ 1.15v
September 21, 2007 8:15:48 AM

Ok the OP has the G0 version of the Q6600. Temps for this should be around the 35c mark with a good heatsink.
What thermal paster did you use? If it was Artic Silver 5 only a thin line across the middle is needed for quad cores as it will spread out. It might also be a good idea to check that your heatsink or cpu are not concave (known with Intel cpu's) as this means that the heatsink is not making proper contact with the cpu and thus the temps are higher.
You could try moving your pc to a different room but I don't think it's your ambient temps (75f isn't that high).
Check your cpu with a razor blade to see if it's completely flat..

September 21, 2007 10:12:08 AM

Im not sure if any1 has said this but you should change the vcore if its on auto because it will always be volted higher... somewhere around 1.25 should be stable...

Cheers
September 21, 2007 2:15:45 PM

I have an update, after raising my speeds and lowering my voltage just a hair.

My voltage is at 1.216 and I'm o/c to 3Ghz
My idle temps are 48C and under load I stay around 62C average.
I had prime95 stressing it all night.

From what I've heard, seen, read, etc. these temps are fine and within the range of safe numbers. I would like to lower them, so I may lap the processor (already checked the heatsink) when I get up the courage and something to clean thermal paste.

September 21, 2007 5:26:15 PM

Oh, lol. I need to get my head out of my case once in a while.

So running 4 cores instead of 2 puts less stress on each individual core.

Thanks for the help everyone
September 21, 2007 8:40:37 PM

Frankenstyle said:
So running 4 cores instead of 2 puts less stress on each individual core.


i guess in a relative sense, not to confuse things at all... so, if you have 4 cores each running at 1 GHz, you then have 4 GHz running total. if you have 2 cores each running at 1 GHz, you then only have 2GHz running total... so the temperatures should naturally be lower then with the reduced total frequencies.

with your current quad core cpu running at 3GHz, you then have 12GHz total running under your cpu, which is naturally going to run hotter than a 3GHz dual core, which only has frequencies of half that amount.

im sure it makes sense though... but another example is with intels 80 core cpu, i can imagine the operating frequencies probably wont be all that high, for the above reasons. then again, they could find some way around that.
September 21, 2007 9:37:07 PM

This whole thread exactly describes my experience with an e6850 (temps are under load, Prime95 both cores):
1) stock HSF: core temps of 65C or so
2) Arctic Cooling Freezer 7 (with included thermal compound): cores = 62C or thereabouts
3) Zalman 9700 (plus Arctic Silver 5), just installed: cores at ~48C

So I got very little improvement with the AC Freezer 7. Maybe it would have worked better if I'd used AS5. After 3 weeks of playing with this, I'm just glad temps are finally reasonable, and I guess they'll get a little better as the AS5 cures.

I'm new to a lot of this stuff but it's gone okay besides blowing $20 US on the ACF7.

One more thing: my CPU temp (tcase) has generally been just about equal to the individual core temps, maybe even a degree or two higher, and this has held regardless of cooling.
September 21, 2007 10:25:45 PM

You said "WTF mate", Aussie?

If he is then hotter temps are on the way, not cooler:p 
September 21, 2007 10:32:32 PM

choirbass said:
i guess in a relative sense, not to confuse things at all... so, if you have 4 cores each running at 1 GHz, you then have 4 GHz running total. if you have 2 cores each running at 1 GHz, you then only have 2GHz running total... so the temperatures should naturally be lower then with the reduced total frequencies.

with your current quad core cpu running at 3GHz, you then have 12GHz total running under your cpu, which is naturally going to run hotter than a 3GHz dual core, which only has frequencies of half that amount.

im sure it makes sense though... but another example is with intels 80 core cpu, i can imagine the operating frequencies probably wont be all that high, for the above reasons. then again, they could find some way around that.

The way you describe it is true yet not true. For example if your dual core runs at 4ghz, 2ghz a core, it is going to produce more heat than a quad core running at 1ghz each. Alot more. This is because the function of heat in relation to speed (ghz) is most likely the square of the speed (along with some other constants). Like how the saying goes, speed kills.
September 21, 2007 10:34:29 PM

hmm, i guess i can see how that makes sense
September 23, 2007 5:24:11 AM

choirbass said:
reducing the voltages will definetly reduce the temperatures. even if you OC by 1GHz, your temperatures will most likely be lower than stock speeds, if your voltages are low enough. voltages are what drive temperatures up, not really clock speeds. you need a high enough voltage to maintain stability though at a certain clock speed, but the lower voltages the better.

for instance, my X2 3800+ @ 2.0GHz 1.35v will reach as high as 60C+ under load on a warm day. but if i reduce the voltages to 1.125v, the temperature reduces down to low 50s under load, sometimes high 40s, depending on ambient temperatures too. (1.15v is needed for absolute stability though under full load, but it also results in a slightly higher temperature too). at 1.1v though at stock speeds, my system will just freeze, because there just isnt enough current going to it. and at 1.6v+ my cpu will probably fry, and have tempatures well above 70C too. (its only rated up to 71C though, so itll probably either fry my cpu much above that or just crash my system, but hopefully crashing is all that would happen)

current OC using rightmark cpu clock utility, is an increase of 1 degree @ load (alternating 50 to 51C) going from 2GHz to 2.2GHz @ 1.15v


Your comment about frequency having no significant change in heat generation is wrong. Why did AMD and Intel both offer technologies to lower your frequency without turning anything off BEFORE they added features turning off unused CPU parts? Even now, laptops gain alot of power by lowering the CPU frequency. Heat loss from electricity is calculated from:

Heat loss = I(squared) * R

I = current flow
R = resistance

Resistance is constant for any given manufacturing process. Ideally you want both to be low so that the heat loss is also low. Obviously since current is squared, lowering your current makes a VERY VERY big difference. The 2 ways to lower current on a CPU after manufacture are:

1. Lower the voltage(the push to get electrons to a give place). If you don't send as many electrons as you possibly can to the transistor, that's less heat. Ideally you want enough electrons in the transistor for the signal to go to the next location without losing the signal. It's kind of pointless to try to conserve heat if your computer can't even think straight.

2. Lower the frequency(speed at which the transistor switches). The transistor has to resaturate the electrons after each cycle. If the transistor switches faster, that's more current. If I double the clock cycle, I'll create twice as much current at any given time. That's bad j00 j00. So how do you fix this problem? Make the transistor smaller. If you make the transistor smaller, the number of electrons you can shove into the transistor becomes less. That means less current and therefore the square of current saved is not lost as heat. This is probably the main reason that Intel competes with AMD for performance/watt. They have a pretty big technological advantage over AMD on transistor design. If AMD could market smaller transistors on par with Intel, AMD would probably own Intel in performance/watt.

Just as a side note, some of you might wonder why the nm numbers are chosen. You probably all remember the 130nm -> 90nm -> 65nm -> 45nm -> 32nm(someday).....

Those numbers aren't just pulled out of a hat. In the 3D world each die shrink represents approximately 1/2 the volume of the previous design. Remember you have to think about length, width, and height. Cool huh? So a 45nm transistor is about 1/4 the size of a 90nm, and 1/8th the size of a 130nm.

Also, having a dual core at 2Ghz is not like running at 4Ghz, entirely. If you run a single program, it will only run at the speed of 1 core. 2Ghz in this example. No faster and no slower. The advantage to running multiple cores is:

1. You have a program optimized for SMP where it IS optimized to equally load several(hopefully all) processors and no 1one task is waiting on the next. This will yield a situation that's VERY close to what you described. Unfortunately as far as I know there aren't any programs on the market that are optimized that well to where this possibility exists. It's a great programming feat to write programs to take advantage of SMP to the max. The programming prowess required to make sure both threads aren't waiting on the other is hard to do, and very few customers would truly enjoy the benefit of this anyway.

2. If you are running a CPU intensive program(I'll just use WOW as an example), your OS and other background tasks can run on the second core. This prevents your computer from appearing frozen. Your OS and WOW can both be responsive since they aren't competing for CPU usage. Sure they might compete for FSB or RAM, but that's a different problem. If you remember 10 years ago if you wanted to ZIP a big file(like 1GB back then) it could take a while, and your computer was almost worthless while it was zipping. I remember setting up batch files to encode my .wav files to .mp3 after ripping them from my CD and my CPU could sit at 99% CPU usage overnight and I'd encode a whole 5 songs. It sucked, and your computer was completely busy the whole time. Don't even think about trying to multitask during it, because it was pointless.

If you do a benchmark of zipping a file on a dual core and a quad core where the only difference is the number of cores, the time won't change at all. AFAIK all zip programs are single thread, so they'll only process the data as fast as 1 core. Check out some of the numbers for the old Pentium-D benchmarks. They were fantastic compared to their single core brethren IF the program was optimized for SMP. Otherwise, it was pointless. This single fact is why when Pentium-Ds first came out the Ghz was lower than single core, and gamers had a choice to make. Go buy a dual core and run their single thread game at the slower 2.8Ghz, or buy the 3.4Ghz computer that was single core. It was a though choice, and some people argued both ways on which was better because of #2 reason above. Some swore that the 3.4Ghz was faster in single core, others swore a 2.8Ghz was faster because the OS didn't have to compete with the game, so the game could use that few % points that the OS wasn't using now.

Think of your computer as a highway:

60MPH = 2Ghz
2 lanes of traffic = 2 cores available

Just because you have 2 lanes doesn't mean you can do twice as much traffic because you always have that jerk that wants to swerve between lanes slowing everyone else down because they don't want to get hit.

Sorry if this is a long post, I like to explain everything so that everyone can take something away from knowledge. Don't feel too bad choirbass. When I spent 5k on my quad core Xeon machine a few years ago(before there was such a thing as 'dual core' CPUs) I really wanted to think the same thing. I had almost '10Ghz' then. Believe me I wanted it to run at 10Ghz for the money I spent on it! But nobody could compete with my machine's ability to do a virus scan of C, defrag of D, zip a file on E, and still play Everquest at 60+ FPS. That's where my geek points were.
September 23, 2007 8:57:46 AM

i feel humbled, lol. ive read a somewhat decent amount, mostly from various hardware threads though for the past couple years, only following from time to time... for multiple cores and heat output, that was the closest example that came to me to give a linear example of heat comparisons, without mention of manufacturing processes (dont understand it too well myself), other than smaller manufacturing processes primarily resulting in reduced cost, and an increase in transistor counts able to be added, and a reduction in operating temperatures and possibly power required as a result, not sure of too much other than that as far as manufacturing and all... even though smp performance is heavily dependant on how well optimized the software itself is, as you said... didnt mean to have you type all of that, though its appreciated to help clarify for everyone.

though i do appreciate the explaination of heat output and whats involved :) 
September 23, 2007 9:31:57 AM

I love writing stuff like that. I do it occasionally for posts where I know someone will actually walk away with some knowledge. My thirst for knowledge knows no bounds. I feel like my goal in life is to know everything about everything. Of course, that's not possible, but I really love learning stuff. If I had it my way I'd go to school for a living. The volumes of information I could write in the forums would be a great read for alot of people. Unfortunately I rarely feel motivated to write posts like that which is above because I'm not sure how many people would actually use the search feature and use the information provided. Sure, alot of the information is extra reading material. But it's totally fascinating to me. I'm sure others(probably like you choirbass) enjoyed the rather long post because of all the good info you can get out of it.

If there's a few topics like this one that you'd like to see covered on how/why things are how they are I can do a few writeups. I enjoy doing it when I know people will actually benefit from it. I'd do a writeup on the whole CPU/heatsink/glue relationship and how it all works out thermodynamically so that everyone can read and understand how complex, yet simple it is. Unfortunately I think if I wrote it the next post I'd see would be "OMGZ my CPU is hot! What is wrong?" I'd really like to write up the article, post it on the forum as a sticky. And when someone posts some useless post repeating the previous 25 people who all posted the same topic just link to the sticky and say have a nice day.

The other one I'd love to do a writeup on is UPSes. The article on TH didn't really do much to explain the inner working of UPSes. There's some things that customers should know when picking out a UPS. It's nice to understand how each UPS works(each company handles their inputs and outputs a little different. Some are better than others. Unfortunately these companies aren't quick to just and over the secrets to how their equipment works, you have to disect them by hand to understand what's so much different about them. I will say that I buy APC usually, but my Cyberlink is awesome too.

I've spent LOTS of spare money(haha.. I operate a nuclear reactor so my hobby can afford to be expensive) building systems for the sole reason of learning. I've learned alot from being able to blow $5k on a computer because I wanted to learn something. I've bought quite a few toys just 'because I could'. My friends are always asking what parts I have lying around that I don't use anymore. They are usually fairly new parts that I sell for dirt cheap.

I'm sure if I talked about some of my previous monster computer's I'd make 99% of the geeks in this forum wet their pants every night for a year.

So you want me to write up a post about what the heck a transistor is and how it works? =) Quite a few people see a transistor as some magic switch that uses magic pixie dust to work some kind of super cool physics defying magic. If the magic smoke capsule breaks it's broken.
September 23, 2007 10:36:46 AM

nope my tr120 ultra is not lapped.. neither is my cpu ( i may sell it so i dont want to void warranty) ... anyway after i put some more AS5 on it it dropped to about 70 full load @ 3.6ghz... sadly i need quite some voltage as this cpu is not particulary good...
September 24, 2007 2:14:10 AM

I appreciate the explaination there. Just letting ya'll know.
September 24, 2007 6:28:49 AM

Awsome write up and would appreciate some more for sure!!
September 24, 2007 1:44:09 PM

I just oced to 3.2 ghz and I am now idleing at 117F.
September 29, 2007 8:38:08 PM

Which app are you guys using to read out the core temps?
My Q6600 G0 is idleing at 26 Celsius and 41 after running Prime95 for an hour. But that's according to the Asus PC Probe II util so maybe those readings aren't any good?
!