Overclock.net banner

1 - 20 of 24 Posts

·
Banned
Joined
·
2,407 Posts
Discussion Starter #1
I still don't understand why my e5200 doesn't get below 34C.

I'll try not to bore you but I don't want to leave information out.

Ambient Temps. I live in the lower North Eastern US. Our temps range from about 0F (-17C) to 90F+ (32C). Right now, it's not too cold out. About 34F (1C) outside. Inside is a bit warmer but not much.

So I've run this in three different configurations:

1. Stock Intel HSF. Idles at 34C.
2. Dark Knight. Idles at 34C.
3. Dark Knight, OCed to 3.3ghz, 1.25v vcore. Idles at 34C.

I can't give you specs on my load temps. I don't think I've gotten over 60C though. I tested stability so long ago that I can't remember. Nothing dangerously high, if I recall correctly.

I'm thinking three things are possible.

1. The motherboard's temps are either stuck at 34C or bottom out at 34C. It has no problem going up when I use the computer.
2. I flubbed the dark knight install. I've done it at least twice though.
3. The Dark Knight needs to be lapped.

I just don't know what's wrong that I can't get below 34C.

34C is okay with me but when I move to the Q9550, I want to get this figured out since I know I'm going to have more heat with 4 cores, right?
 

·
Not new to Overclock.net
Joined
·
77,943 Posts
Having stuck sensors with Intel's 45nm Core2s is normal, but not common. But fortunately, all that matters is the load temperature.
 

·
Registered
Joined
·
1,440 Posts
I doubt they're all running at 34C. I've heard of sensors not working properly.
 

·
Premium Member
Joined
·
3,864 Posts
mine idles about that man. The only temperature that matters is the load temp, idle doesn't mean jack! I wouldn't worry to much about that, as long as your load temps are low then there shouldn't be any complaints.
 

·
Registered
Joined
·
2,625 Posts
Quote:


Originally Posted by Protezione
View Post

34 is pretty darn cold. My idle [email protected] runs at 32*c on water in 24*c ambients.

That's not cold. My e6300 idles at about 28c with the stock cooler. I think it's a sensor problem, probably nothing to worry about since idle temps don't matter.
 

·
Not new to Overclock.net
Joined
·
77,943 Posts
I'd like to add the following to my reply above: ....Besides, software can't give us 100% accurate temperatures anyway. So, don't worry so much.
 

·
Premium Member
Joined
·
8,033 Posts
as others have stated, idle temps don't mean squat and are often not reported accurately anyway. Load temps are what is important and luckily do get reported properly with Real Temp.
 

·
Banned
Joined
·
2,407 Posts
Discussion Starter #12
Wow, that was a boatload of responses in very short time... Thanks.

How can idle temps not matter? How can you accurately get a load temp if your idle temps aren't accurate?

If it is a stuck temp sensor, any way to unstick it?
 

·
Not new to Overclock.net
Joined
·
77,943 Posts
Quote:


Originally Posted by mtbiker033
View Post

as others have stated, idle temps don't mean squat and are often not reported accurately anyway. Load temps are what is important and luckily do get reported properly with Real Temp.

With all due respect, here's the way it is: temperatures below 50°C are nothing more than temperatures below 50°C. The best we can say is, "well, it's below 50°C." But as it gets higher and higher above 50°C, the accuracy also increases. At around 60, the temperature is reasonably accurate enough to say that it's pretty close to 60°C.

But this is where it begins to get complicated, and this is why it's important to remain casual about all of this. The software is likely using a Tj. Max value of 100°C. This dictates what the temperature reading is. Then you have the Tj. Max offset. This also dictates what the temperature reading is.

So, here's what's important: as long as the overclock is remaining stable through rigorous stress testing, then this is an indication that the temperatures are fine. However, if you're seeing temperatures well into the 70's and 80's, then you can say "well, I guess it's time to do something about the heat". I mean, don't go "omg, it's exactly 76°C" because all that reading would mean is "hey, cool it down a bit".

So I'll say it again: we need to stop worrying so much so that we can focus on what matters.
 

·
Not new to Overclock.net
Joined
·
77,943 Posts
Quote:


Originally Posted by TFB
View Post

Wow, that was a boatload of responses in very short time... Thanks.

How can idle temps not matter? How can you accurately get a load temp if your idle temps aren't accurate?

If it is a stuck temp sensor, any way to unstick it?

Intel didn't intend for anyone to use these sensors as veritable thermometers. So, our temperature monitoring methods are very crude and rudimentary. They're very inaccurate by nature. You see, the sole purpose of these sensors is to either throttle the CPU down or to shut it off when the temperature reaches the Tj. Max which is usually about 100°C for these 45nm Core2s. At least, 100°C was the target when your particular E5200 was made. For all we know, it could be 98°C, or it could be 103°C. Or, it could be 97°C. Or, it could be 99°C, 101°C, 95°C, 106°C... either way, it's somewhere around 100°C.

Therefore, due to the purpose of these sensors, the temperature is only 100% accurate when the temperature is pretty much at the Tj. Max value. So this is why the accuracy increases as the temperature rises. This is also why that anything under 50°C is considered "less than 50°C" instead of "48°C" or whatever. I mean, if my copy of Real Temp says "47°C", then I say "well, it's less than 50°C".

It's a complicated thing. So to keep it simple, just be casual about it.
 

·
Premium Member
Joined
·
7,891 Posts
Just to throw my experience in here my xeon 3110 (e8400) that I use to have had one sensor that was stuck at 32c while the other sensor would show lower ambient temps say like 25c or whatever.

When the cores heated up above 32c it would track properly.
 

·
Banned
Joined
·
2,407 Posts
Discussion Starter #16
Thanks. That's too bad that we can't accurately get the temperature. I like to be in control of things like this.
 

·
Not new to Overclock.net
Joined
·
77,943 Posts
Quote:


Originally Posted by TFB
View Post

Thanks. That's too bad that we can't accurately get the temperature. I like to be in control of things like this.

But this aspect isn't that big of a deal. What's more important is the stability of the overclock. So, having a general idea of the temperatures will give you a really good idea of what caused any instability you might encounter during stress testing. I mean, if the temps were in the 70's or 80's, then you know for sure it was heat. But if the temps were in like the 50's or 60's, then perhaps it wasn't.
 

·
Registered
Joined
·
2,835 Posts
Quote:


Originally Posted by TFB
View Post

Wow, that was a boatload of responses in very short time... Thanks.

How can idle temps not matter? How can you accurately get a load temp if your idle temps aren't accurate?

If it is a stuck temp sensor, any way to unstick it?

The temperature sensor is made to sense if it's too hot, not too cold. Therefore it works much better for reading load temps than for idle temps.

34C isnt bad. My i7 goes down to about 32 with speedstep enabled. Before that I had a Q9550 and it idled around 34C also.
 

·
Registered
Joined
·
989 Posts
for everyone who says that their processor idles at xx temperature, what is your ambient? your idle temps are below my ambients, so your rooms must be very cold. OP, make sure you take ambient temperatures into consideration.
 
1 - 20 of 24 Posts
Top