Sign in with
Sign up | Sign in
Your question

Not another Core 2 Duo "temp" thread?

Tags:
Last response: in CPUs
Share
September 24, 2006 8:36:49 PM

As some one pointed out, one of the single most repetitive threads around here right now is Core 2 Duo CPU temps.
Its just my opinion, but one of the mods should create a sticky specifically for this subject as it recurs so often. I would recommend the older thread started by lcandy for this purpose.

lcandy's thread

Why does this thread recur so often?
It seems to stem from 3 things:
1) People improperly attaching their CPU HSFs
2) A program called "Core Temp"
3) Case design:
...Fan size
...Fan installation/orientation
...Cable routing


IRT problem 1: This seems to be one of the most repetitive problems. Manufacturers use several different methods to attach the HSF to the CPU. The simplest solution to mounting problems is to actually read and follow the manufacturer’s installation instructions. You don’t really need Artic Silver 5 compound, water cooling and you certainly don’t need to lap the HSF or CPU heat spreader, though all these things (especially liquid cooling) will in fact help increase the efficiency of the heat transfer process.

Of greatest importance to problem 1 is cleanliness. When spreading the heat transfer compound, both the CPU heat spreader and the HSF must be absolutely clean, a tiny speck of grit will cause a gap between the face of the HSF and the CPU heat spreader drastically reducing the heat transfer efficiency. If you are going to lap the components, you must be especially conscientious in cleaning.


IRT problem 2: I would recommend people do NOT use the "Core Temp" program or trust its readings.

I have tried the "Core Temp" utility. I have disregarded its readings as they did not match either the BIOS or 3 other programs I have used. In fact, "Core Temp" utilities readings were off by such a significant margin that I suspect has a serious formula error.

For the E6600 (rev B2 step 6) that I am running on an Asus P5W DH motherboard with a stock Intel HSF, I have looked at idle temps in various "cooling" configurations using the BIOS. For both idle and loaded temps the other programs I have used are:
Asus PC Probe II
MBM5
Hmonitor 4.2

Core loading was accomplished using multiple instances of "CPU Burn in", as well as D3 and some other progs, with both cores verified loaded to 100% These 3 temp monitoring programs all indicate within +/- 1'C of each other. They are also +/- 1'C of the BIOS for idle temps.
With no modifications to the CPU cooling, the temps are:

Open air idle............23.5'C

Open case idle........ 26'-27'C
Open case loaded....29'C

Closed case idle......29'C (New intake filter)
Closed case idle......32''C (Dirty intake Filter)

Closed case loaded..32-34'C (New intake filter)
Closed case loaded..36-38'C (Dirty intake filter)
(I finally got it to break 36'C under load a short while ago. The filter is getting severely blocked)

IRT problem 3:
Bigger, faster fans do not always aid cooling and can in fact hurt it. The most important factor is flow path, seconded by volume. Velocity really doesn’t do much unless you are limited to smaller (80mm) fans. Reality is: you can flow 3000 CFM through your case and still have unacceptably high CPU temps IF the path the air is flowing does not feed/scavenge air from the CPU HSF. Fan placement, orientation and cable routing all impact airflow to the CPU HSF significantly.

For example, the reason my system runs so cool, even at full load is that the case design provides a single, uninterrupted "layer" of airflow to cool all the components including the HSF. Some of the newer BTX cases provide internal ducting to insure the HSF is being fed by drawing the exhaust air directly over the CPU HSF. Other cases in ATX form factor offered over the past 2 years or so have ducts in the side to feed outside air directly to the HSF

There are many different case designs available at this time, and they are not all created equally in terms of airflow. In most instances, your goal should be unidirectional airflow i.e. air in one side, out the opposite, unless you are going to feed the HSF directly. Placing multiple fans in parallel (all fans "suck" or "blow") at all sides of the case and relying on random openings or gaps to provide intake or exhaust is NOT the way to go. The simplest and cheapest way to go is to place all fans at the rear of the case, drawing air from inside the case. If you want to add fans to the front, sides or top, orient them to feed into the case. Rule of thumb if you are going to have fans feed into the case: For every fan "blowing" into the case, have one equivalent fan "sucking" from the case. Insure all cables are routed in such a fashion as to minimize the "open" space they consume and that they do not impede flow into or around the HSF.

More about : core duo temp thread

September 24, 2006 10:30:31 PM

The following is a dupe from another C2D temp thread :

You need to understand that Core Temp, just like Intel's TAT (Thermal Analysis Tool), read the temperature straight from a digital thermal sensor (DTS) located near the hottest parts of the CPU core(s) (TM2). All the other apps (including the BIOS) read from analog sensors located elsewhere within the packaging (TM1) or even under the socket for some motherboards.

You're just deceiving yourself when you dismiss the readings from Core Temp and TAT as innacurate because they are too high for your taste, deep inside, the core(s) do get that hot, hiding your head in the sand won't change anything about that fact.
September 24, 2006 11:28:09 PM

As an owner of one of the Core 2 Duo threads...

If a sticky is made it should cover the following:

-- What Heatsink is being used. (not fair to compare a watercooled rig to stock heatsink)
-- Ambient temps
-- OS being used (i.e. Win2K & speedstep)
-- Coretemp seems to be the 'gold standard' for measuring temps
-- Last but not Least: Temperatures at Stock and optionally overclocked.

Most people don't say the heatsink they are using or the ambient temp. This info is very important. Next winter I'll put my PC outside, and start ragging on all the people bragging about how their Core 2 Duo is running at 35C.

Would make a good poll, what are your Core 2 Duo temps...

i.e. at 78F ambient, with ARCTIC COOLING Freezer 7 Pro/ AS5, Win2K aka no speedstep, coretemp 42C.
Related resources
September 25, 2006 1:28:22 AM

Quote:
The following is a dupe from another C2D temp thread :

You need to understand that Core Temp, just like Intel's TAT (Thermal Analysis Tool), read the temperature straight from a digital thermal sensor (DTS) located near the hottest parts of the CPU core(s) (TM2). All the other apps (including the BIOS) read from analog sensors located elsewhere within the packaging (TM1) or even under the socket for some motherboards.

You're just deceiving yourself when you dismiss the readings from Core Temp and TAT as innacurate because they are too high for your taste, deep inside, the core(s) do get that hot, hiding your head in the sand won't change anything about that fact.



I am not "deceiving" myself by dismissing the results from "Core Temp"
The readings are not "too high for"my"Taste"

I am disregarding "Core Temp"'s results, as all the other data I have is contrary to the data "Core Temp" is providing.


You "need to understand" that TAT (which I use and trust) and "Core Temp" measure in die temp, as opposed to the other programs I mentioned which measure on die temp. I do not refernce the erroniuos "socket", "CPU surface" etc temp readings some other programs produce as they are misleading.

What I would have assumed you understood, since you know about TAT, is that for the consumers purposes, it is the on die not in die temp that is of practical concern, since it is the one refernced by HSF/ water cooling systems manufacturers (at least all the ones Ive used) and the one to which consumer results must be refernced for accurate comparative purposes.

What you "need to understand" is that my results, comparing "Core Temp" to TAT readouts show "Core Temp" about 10'C higher. Since both systems refernce the same (in theory) data, one of them must be wrong.
What you also "need to understand" is that watching the "Core Temp" products results closely over time, (weeks, not minutes) in relation to TAT as well as the other programs I have used has shown a significant lag in relative temp changes.
What you "need to understand" is that I have seen "Core Temp" freeze, providing no reliable data while instances of MBM5, Hmon4.2 TAT and PC Probe running at the same time have all shown normal fluctuations IAW system loading.

What you "need to understand" is that "Core Temp" has been around for a whopping 5 months now, (with the current version 0.9.0.92, released barely over 2 months ago) and is still a beta version as clearly stated on the web site, located here:

http://www.thecoolest.zerobrains.com/forums/viewtopic.p...

What you "need to understand" is that everything Ive looked at and done, along with all the posts concerning the subject of Core 2 Duo operating temps on THG, shows the "Core Temp" product to be unreliabale and inaccurate. Not because I dont like it or because you think I want to see something that "Core Temp" is not showing me, but in relation to all the other data.

What you "need to understand" is that the reason Ive spent a ridiculous amount of time playing with temperature monitoring programs and montoring the results is that it is important to me, for the purposes of proving the design of the enclosure Ive built. I would look very silly if I tried to sell the high efficiency airflow case design Ive been working on in my spare time for the last 9 months if it wasnt efficient, or couldn't feed a CPU HSF sufficiently.

Now, is it possible everything else I've run including TAT is wrong? You bet it is. Is it likely? I dont believe so. Has it been important for me that I find a program that is accurate, regardless of the results it give me? Well, I think its obviuos, but the answer is "yes".

On that note, if people want to know what the in die temp of their core 2 duo is, I would recommend using TAT, located here,
http://shintai.ambition.cz/files/tat.exe
instead of "Core Temp".

So, is there anything else you feel I "need to understand?"
September 25, 2006 3:44:47 AM

Quote:
1. I have never seen TAT and CoreTemp reporting more than 1C from each other, they both use roughly the same calculation method from TM2 and so makes sense.
10C difference between TAT and CoreTemp is absolute FUD. PERIOD


Wusy, I cant believe you. I cant believe you accused me of speading Fear Uncertaintly and Doubt? How could you lower me to the class of BowelMatrix, 9nm, Moronically Misguided Mike and Sharinincompoop? Seriously? Because (God forbid) I dont trust a program you think is reliable? Ive have never, ever claimed to be an expert, nor am I "making sh!t up as I go". I am presenting the results of the data I have seen.

Well, heres the screen shot. Perhaps you and the other "elites" can explain to me why Im seeing the disparity between Core Temp and TAT



Personally, I thinks it because "core temp" is crap.

Quote:
2. There's no on-die, in-die or socket temperature bullcrap. It's either calculates from TM2 through DTS or TM1 as SidVicious said.
People who call those those in/on-die sensor are amateurs, they know jack arse of what they're talking about when it comes to Intel's DTS. PERIOD.


So Im old. Congrats, you've proved it. On die, In die and socket temps are old terms. I know you're young, but you're not that young and you are expert enough that surely you remember these terms? Seriously. I know with you're knowledge that you are familiar with them
Oh yeah, as I stated, Im not an expert. I do not get paid therefore, by definition I am, as you stated, an amatuer. Of course, if Im not mistaken, since you are still in school and not getting paid for this you, by definition are also an amatuer. If you meant to say, I lack expertise. Yes I do, and I have said so many times. So, uhh whats your point?

Quote:
3. CoreTemp updates its temperature by around 5s slower than TAT. That's known.

Whats this have to do with the subject? Are you implying that the lag and "freezing" Im seeing is due to this? If so, allow me to clarify: The lag and freezing is on the order of minutes, not seconds. I dont consider the cyclic sample time as a factor in the disparities Im seeing.

Quote:
4. Because Intel never intended people to calculate temperature from DTS of TM2, they never gave any instructions on how and in fact hides that info very well. The reported temperature is a relative thing and different for everyone. Temperature sensors never reads absolute value. PERIOD
What is your point? You are quite right, temp is a relative thing which is an important point. Comparing the (old term) on die temp with the core temp, and mixing the limitations is wrong .

If someone says they are seeing a 55-60 with TAT, its fine. If they say 55-60 with Core Temp, Id say get a different prog. If the said 55-60 with Probe or MBM, I'd say holy baking CPUs Batman, you've got a problem. Which is exactly what is happening here. People are using "Crap Temp" and seeing 50, 60s 70's etc, and are wound up becuase they want to see 20s-30s. They are not going to see that. with either TAT or crap temp.

Quote:
5. Due to #4, read the C2D OC Guide on how to determine proper temperature when overclocking. Reading from temperature sensor has always been a relative thing for a well trained overclocker. PERIOD
Ive read your guide, and as I have told you previously, it is excellent. Again, I thank you for taking the time and effort to provide us with that.

That said, do you really believe anyone who is posting the "OMG, my Core 2 is burning up" here is an overclocker, or is going to progress to this level of tinkering any time soon? Its only an opinion, but I dont think so.

Most of these people seem to fall into the catagory of "cant seat the HSF properly". Again, my opinion, but wouldnt you agree? Do you honestly think the people who cant get the HSF on correctly are at the level that they should be playing with TAT or Core Temp? I dont.

Clearly, by the number of repeat "are my temps too high" posts, these folks do not understand TAT or Core Temp, and dont need to be playing with them.
Again just my opinion, but MBM, Hmon and the rest are more these folks speed, for the time being.

Quote:
6. Whatever, me and Sid are too elite for you guys to handle.
Talk to our hands cuz we ain't listening

Wusy, you are sharp. I know it. So explain to me why it is everything else is working, on a normal-stock Asus E6600 setup EXCEPT Core Temp.
I dont really think you read the thread. I think you saw Core Temp doesnt work and "knee jerked", because I dont see a whole lot of controversy in your points to me.

Just as an aside, BM lables himself "elite" as well. I think we all know how elite he really is. Maybe you should stick to "expert" so as not to stick yourself in the same class BM claims. You two are definately not in the same class.
September 25, 2006 4:28:58 AM

Quote:
Wooowwaaaa, calm down guys.... this is a good thread, I am learing here. I am the first to admit, I am a little bum-foozled at the disparity I am getting from temperatures.

I need to read through the links you guys have posted. But this is informative.

On that note, I have seen Core Temp reporting tempertures in agreement with Everest but about 10 degrees off from BIOS and Asus probe.

This is indeed confusing myself and many people.

Jack


Both Sid and Wusy gave good explanations of where Core Temp, TAT and the others are getting their data.

Sids is here:
http://forumz.tomshardware.com/hardware/trust-temps-fto...

My contention is that the reading provided by Core Temp is wrong.

For that read out, I would go with TAT (its free) over Core Temp, though apparently, according to Wusy's implication, I seem to be the only person having a problem with it.

As for the other monitors, they will always (if they are fuctioning properly) show a lower reading than TAT.

Your temps (from the other thread) seem good to me. In hind sight, I probably shouldnt post my temps. They are a little low, and I expect to drop them another 3-5'C as I tinker with my enclosers flow path.

The point of this whole thread was that SOMEONE (Wusy would be a good candidate) should get a sticky going to explain these sort of disparities, along with the standard "make sure your HSF is seated before you spend more $$$ because your temps are high"
September 25, 2006 5:31:45 AM

Quote:
Regardless this is a good thread to have, you, Wusy and others have expertise that I lack, so this has been informative. (Extremely informative). It would be a good thead to condense down and ask Jake to sticky for a few months as more and more C2D systems come online.

Jack


I would second that.
Even with wusy venting I also find this very informative .
A Sticky would help reduce the redundant threads.
Peace all 8)
September 25, 2006 12:24:02 PM

OK, let me add this to the mix. Everyone is talking about 'free' programs to measure temps. Do the reviewers use these or do they use hardware thermistors to measure temps? Wouldn't hardware be more accurate? If so, why are we wasting time on something inherently inaccurate? My only conclusion is because it is free. Then we go back to that OLD saying you get what you pay for.

That being said I was running a stock E6400 and my Bios temp was 40C on idle with a Zalman 9500 CNPS AT. Yep, I reseated ambient temp was 75F, and I was using Artic 5. for me that seems high. When I OC'ed the CPU to 8*375=3.0Ghz same Vcore my idle temp in the bios went to 60C.

So I am looking for some hardware to check my temps. I am also getting a liquid cooling rig.

Has anyone used the Coolermaster temp probes with any success as far as accuracy? OR, would this be trading software inaccuracy for hardware inaccuracy.

Thanks guys,
September 25, 2006 8:05:28 PM

one of the biggest problems is airflow, i did a quick test to check this by doing the following.

I opened the PC case so that both sides of the case did not have the side boards, placed a normal standing Fan on one side and switched it on, thus allowing the air to pass freely from one side of the case and through the other side.

The temps that i had before i did the test ( some people might think this a bit drastic ), were as follows all the following are at Idle:

Core Temp: 45/45 *C (Core 1/ Core 2)
TAT: 44/45 *C (Core 1/ Core 2)
SpeedFan: Temp1: 36*C, Temp2: 39*C, Local temp: 38*C, remote Temp: 46*C, HD temp: 29*C

after the Fan test..:) 



so as you can see, just simple airflow can make a huge difference.

oh and that using the stock HSF, I had originally installed a ThermalTake Golden Orb 2 and the temp difference between that and the stock HSF where like 1*C, I will probably reintsall the golden orb as its a lot quieter and has a funky blue light..:p 
!