Printable View
As per my Sig. I have an E2180 clocked to 2.88ghz (288x10) with a OCZ Vendetta.
The Core Temp monitor show the cores to be at 27-30c when idle or just surfing/word processing, this looks so low, do you reckon the reading is wrong
Coolermaster case has a 120mm fan in the rear, 92mm fan at the front and a VX450 psu.
I did OC the chip to 3ghz and the temps showed at 35c
Cos I run Ubuntu on it I can't use the standard Utilities everyone with Windows uses
At idle that temperature is normal, you should be check the load temperature because at idle the CPU would normally turn the multiplier down to save energy which would make it less hot. The problem people have had with Core Temp is that it reports higher temperatures than Realtemp.
This is just a precaution but I doubt your load temps would high, it seems fine to me.
seems quite low :o
what vcore. I think I'm using 1.42v for 3ghz and idle temps are 37-40c.
You also need to check the reading for the Cores with something like Core Temp or Everest. My suggestion is to take a reading after 10 minutes of doing nothing (make sure that there is no CPU usage in the background) and again after 10 minutes of Prime 95 (or Orthos). This is a good check on both the efficiency of your cooler and whether the program used is accurately calibrated. Also if you know what the ambient (room) temperature is, that is useful too.
From memory I think vcore is around 1.425, I've turned off the Auto multiplier adjustment in the BIOS.
Thorsson, I'm running Linux, no Windows boot available on the machine so I don't think I can run any of those Utilities. I'll probably have to look around for some Linux ones again. I'm new to Linux and its not the easiest to install 3rd party sotware on.
Room temp probably around 20c
well 27c - 30c is pretty good as I just switched out that cpu and checked before it was at 40c in bios on the 1.425v setting. thats with the eist on too.
Idling with the room below 18C at floor level, I'm seeing 28/29C rising to 46C stressed (Prime95 small FFT).
The CPU is at 2.9GHz (290 x 10) with Vcore 1.40V in BIOS, Ninja B on top.
Numbers from Speedfan 4.35 (Coretemp agrees).
I do find going over 3GHz requires a fair bit more voltage and produces a lot more heat (and higher fan revs); 2.9 gives me a very quiet system. :cool:
Coretemp and (latest) Speedfan use same assumptions, so it would be odd if they differed. That's why I like to use RealTemp as the secondary program. But these figures look about correct. Is that the CPU (die) temp? What do the cores show?
Core temps. 'CPU' shows about 2C higher on mine.Quote:
Originally Posted by Thorsson
In that case I would suspect both of them are undercalling your cores slightly, probably 5C. CPU should always be a few degrees cooler than the cores under full load.
Great nice to know these temps are realistic, I'll try and get some other Temperature software for linux, the BIOS and Linux temp monitor do agree. Just does seem very cool for an OCd chip, although its a cheap cooler (OCZ Vendetta) I believe its more than enough to cool an E2180.
I also found that pushing it to 3+ghz gave 5-6C increase at idle and needed a lot of voltage thats why I backed off to 2.88
Just a couple of thoughts - Since the readings are "too-good-to-be-true", I'd satisfy myself that the system is stable after lots of heavy stressing e.g. P95 and 3dmk06. I think it's safe to say most of us OC until it becomes unstable and then back off a bit. We tend not to stop OC just because its 80C or whatever. In other words we all know some people say, my rig runs stabe at 90C while another can't get past 60C.
The other thought is to bite your lip and install Windows on a spare disk (temporarily) just to run the famous temp monitoring apps, then you'll get a better feel of how hot your rig is.
It really is increasing the voltage which pushes up temperatures dramatically, I notice very little difference in temps between 2.33GHz (stock) and 3GHz (about as high as I can go without increasing voltage) but pushing the voltage up to 1.33v (from 1.28v) to have it stable at 3.5GHz makes a marked difference to load temperatures.
Increasing the clock has a linear relationship with power used, increasing voltage has an exponential one. And if you're raising voltage so that you can raise clock that's a triple whammy. I try to avoid it.