Ocing - Which temp readings to believe?

[nightscope]

Alright, I got a e2180 along with an Abit IP35-E motherboard. I want to OC the cpu, but I'm in a bit of a dilemma.

I haven't oced yet because I'm receiving different temperature readings.

Bios reports CPU at 37 Celsius
Core Tempt reports 24 Celsius
Speed Fan reports 24-25 Celsius

So which should I believe? Remember the cpu is at stock and idle.

Thanks in advance!

 

[dragonsprayer]

bios is never the same as the desktop
if you mean your bios on the dsk top the it should read the ame as the other ones

regarless it does not matter since your temps are low


temps tell you if your too high - thats 70c

of it you have a problem if they change or increase suddenly

so who cares if its 24 or 37 its fine!

 

[nightscope]

37 is kinda high for this chip from what I've read...people get 25-30 on average. And I applied some mx-1 paste on it.

 

[nightscope]

Ok I got prime95 and I'm running one blended test...do I need to run two tests since I'm using a dual core? it's version 24.14...How do you run two tests?

 

[CompuTronix]

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

 

[dragonsprayer]

37 is kinda high for this chip from what I've read...people get 25-30 on average. And I applied some mx-1 paste on it.

first mx - sucks, all marketing. the paste dries fast its a blend if silver and ceramics ..alumina and boron nitride. you get good numbers on newly installed chips due to the fast evaporating solvent

the draw back is if you bump yor cpu cooler/heatsink few days later that stuff is like power and it will not flow like AS

arctic silver is pliable and drys slow it works much better over a wider range of applications - i am not impressed with mx.

i love people who say cpu's run at room temps 25c? come on! 25c? is that in a 25c room too? which you know is impossible. i have seen those readings too!

i assume everyone by now knows overclocking improves performance with little negative effects - so i have 3ghz factor in!

why use a cooler at all!

 

[nightscope]

I don't quite understand the temperature guide. So the maximum tjunction depends on the processor? Well it's an e2180 with M0 revision, what does that mean?

 

[nightscope]

So my e2180 is 100 Celsius or 85 max tjunction? I really would appreciate an answer, don't want to fry my one day old computer...

 

[Grimmy]

So which should I believe?

CoreTemp should provide the digital readings fine. SpeedFan usually needs to be calibrated, or configured with the offset setting. Kinda suprized that you don't have a 15C cooler reading off it. I'd say you have the HS on correctly.

I've read...people get 25-30 on average. And I applied some mx-1 paste on it.

I wouldn't worry about what other people get, since your ambient room temps and airflow are contributing factors. For example my room temps around the PC case are 79-80F (26C). CoreTemp shows 36C (Core0) 39C (Core1). SpeedFan Tcase 29C. Now if my room temps are 68-70F, then I would get lower readings. I used the MX-1 paste that came with my Tuniq. Only thing I hate about it, was applying it. I've use AS5 in the past, but to tell you the truth, I really don't see that much of a difference at all. Since DragonSprayer talks about paste that dries out, well AS5 needs 200 hours to cure, meaning to become solid. So it's not based on how moist it stays.

Ok I got prime95 and I'm running one blended test...do I need to run two tests since I'm using a dual core?

A more simple way of testing CPU load on your cores:

Stress Prime 2004 Orthos Edition (SMP Version)

Or if you really want to use Prime95:

X2 dual core cpu's + dual Prime95 for stability testing (fixed)

I know its about X2 CPU, but it still works for C2D's as well.

I don't quite understand the temperature guide. So the maximum tjunction depends on the processor? Well it's an e2180 with M0 revision, what does that mean?

heh, I'll let CompuTroni x explain that.

So my e2180 is 100 Celsius or 85 max tjunction? I really would appreciate an answer, don't want to fry my one day old computer...

I'd say the max is 85C. From what I remember I thought the quads had max of 100C.

I wouldn't worry too much about frying it, since it has a thermal trip, which should shut the PC down.

 

[nightscope]

Yeah that's what it says in the overclocking guide, that you had to add 15 Celsius to coretemp temps only to quads.

In the temperature guide it says the max tjunction is 100 celsius for the e21x0 series...but I really don't think that's it. That means my temps are close to 40 Celsius at idle! and 75 at load!

For applying it, I just put a small dab in the middle and let the heatsink spread it out.

Just want to make sure that the max is 85, it's worrying me.

 

[Grimmy]

I had to add 15C to speed fan for my E4400. So it doesn't make a difference if its dual or quad.

The only other thing about the M0 Stepping, is that it would be the newest stepping which provides a lower vcore. Here's info base on E4400 (L2 stepping) E4500 (M0 stepping)

Intel Core 2 Duo E4500 M0 Stepping CPU Review

Not sure if you over looked it off that guide:

Note 1: Core Temp is an excellent utility, however, it has a fundamental flaw in terminology, which creates confusion in the CPU temperature community, by obscuring the distinction between temperature and specification. Core Temp shows Tjunction 85c (or 100c), which is an incorrect term. The proper expression is Tjunction Max 85c (or 100c), which is the term defined by Intel as shown above in the Specifications section, and as represented below:

So what you were looking at, is what CoreTemp says for Tjunction Max, which isn't correct according to the guide. I'm pretty sure it's 85C. But the actual temps its showing for your cores in realtime, should be correct or trusted.

 

[nightscope]

Well they look correct, but I can't base it just on that. I need to know for sure

 

[Grimmy]

Run Prime95 on both cores then or use Orthos for about 6 hours at stock. That way you know your system is stable at stock, and you can take note of the temp readings.

 

[nightscope]

How would that help? There's barely 2-3 degree difference between stock and oc. Which temp do I read off of speedfan for the cpu? The core 0 and core 1? Or just Core:? Core 0 reports 28 celsius, core 1 reports 28 Celsius, and core: reports 48 celsius...which is right? CPU: reports 30 celsius, and system reports 27 celsius.

 

[Grimmy]

Heh.. so you've already OC to see the difference?

The Tjunction Max on CoreTemp just has the wrong terminology. However it should be 85C for that CPU. So that means your system should or better turn off if one of those cores hits that temp of 85C. So the same applies to my E4400 has the same max reading of 100C. That that part of it is just wrong, from what I understand.

Now your actual temps shown in CoreTemp can be pretty much trusted, since it is a digital sensor, and there's basically nothing to offset or configure. Your idle temps are not that important, your CPU load temps are what really matter. So that would be the main difference between your thermal limits on stock vs OC.

When testing your rig at stock settings you should get an idea of what your thermal limits are when it is or not OC'd. As far as calibrating SpeedFan, I use a IR thermometer to figure out my Ambient temps around and in my PC case.

You can PM CompuTronix, or you can wait for him to reply. He usually does.

Edit:

Here are screen shots of my testing (it wasn't extensive) when I had my E4400 at 2.8ghz (room temps were 75-76F):

CPU on Load

CPU Cool Down
**note that the cores are 15C cooler on speedfan, since I didn't set the offset at the time.

 

[nightscope]

So my core temps are probably correct and the max tjunction really is 80 celsius? Are you 100% sure?

 

[Grimmy]

It should be 85C..

But I've been wrong before...

. o O (gulp..)

 

[nightscope]

Yeah I mean 85...

Oh if you're wrong, then ;lkasjdf;lkjasdf

 

[nightscope]

Does anyone know for sure what the max tjunction for a e2180 is? M0 revision?

 

[CompuTronix]

NightScope, I can assure you that the information in my Guide is correct, and is in accoradance with Intel's specifications, which you can see for yourself by using the link provided to Intel's Processor Finder in the Specifications Section. Your Tcase Max (CPU temperature) is 73c, and your Tjunction Max (Core temperature) is 100c, as shown in Scale 1. This means that Hot is Tcase (CPU) 70c, and Tjunction (Hottest Core) is 80c. Safe temperatures are Tcase (CPU) 60c and Tjunction (Hottest Core) 70c, which are again, just as shown in Scale 1.

As explained in the Guide, the utility "Core Temp" you should be using is Beta 0.95.4, so use the link in the Tools Section to download it. 0.95.4 should show your Tjunction (again - WRONG terminology - correct terminology is Tjunction MAX) as 100c.

Hope this answers your questions,

Comp

 

[nightscope]

I am using the beta 0.95.4 version, and it still says the tjunction is 85 Celsius...

 

[nightscope]

Plus, if the tjunction truly is 100 celsius, then I've reached 76-79 celsius temperatures...and that truly worries me.

 

[CompuTronix]

Follow the Calibrations Section, Parts 3 & 4, and use Crystal CPUID to read the DTS register values which will determine EXACTLY what your Core temperatures are. These results are virtually indisputable, and will reveal whether Core Temp 0.95.4 is unable to correctly decipher Tjunction Max for your M0 Stepping.

Comp

 

[nightscope]

Alright, I got 43 Celsius for tjunction in part 3. So that means the core is running at 43 Celsius? That is awfully hot for idle temperature

 

[CompuTronix]

Did you check the values for both Cores? Also, did you work through Parts 1 & 2? If so, what were your results?

Comp