I recently built a new PC (specs at the bottom), this is the first PC I built with overclocking in mind. I have not overclocked yet.
What's bothering me is the two cores of the processor (E8400) are showing drastically different temps- Core 0- 37 and Core 1- 49 degrees Celsius seems to be average and are relatively stable there. Is this normal? If not, what is going on?
I applied the thermal compound myself, and I am sure the whole chip is coated evenly. I used "OCZ Freeze Thermal Extreme Conductivity Compound" which has been shown to be extremely good.
Also one final question- what are good "safe" temperature limits? IE, when overclocking, what should I keep it under? (I know I need to read the overclocking guide, but haven't had time yet and wont be OCing for a few weeks anyway)
how did you apply the thermal coupund that you know it's even? hopefully right in the center. check with task manager if one core is working more than the other. this would explain the diff in temps. another matter is which program you are using to check the temps? i think that the mostly likely explanation is that one core is being used more than the other, as in a single threaded application is running that core and second one is idle.
I think I may have used too much, could that be part of the problem? What is the best method to apply the compound? Any good websites to read about it?
As far as one core being hit and the other not, that is a negative. They are both being hit equally. It's also been consistent over a day, and doesn't change no matter what I do- from Crysis to just internet browsing it is pretty constant.
These temperatures check out with both Coretemp and HWMonitor as exactly the same. Havn't checked the BIOS yet to see if they match that.
Almost all games I play are single threaded, and only Core 0 gets hotter under load, usually around 46C and Core 1 is cool at around 25C on my 8400 @ 3.6GHz using CNPS9700. I applied my thermal paste using the nice brush applicator that came with the Zalman cooler, very easy to use and spread perfectly evenly; a nice thin coat on the CPU & heatsink did it for me.
Perfectly normal to see the 2 cores fluctuate in temps, only program I saw that brought the 2 cores to their max temps, was Prime95, for obvious reasons.
Message edited by Dechy on 05-06-2008 at 05:56:35 PM
The problem I have with that theory is that its been core 1 that's been hot over 2 days and I monitor its activity. Even idling, the temp difference is there.
I looked on OCZ's webpage and found the following:
"The presence of an Integrated Heatspreader (IHS) on the CPU greatly facilitates the application of the thermal compound. Place a small amount of Freeze (roughly the size of a grain of rice) on the center of the IHS. Lower the CPU cooler directly down onto the CPU to minimize smearing, and push it gently into place. It may appear that only a fraction of the heat spreader has paste installed, the cooler will squeeze the paste out to the appropriate width over time. If necessary, the CPU cooler can be wiggled slightly to facilitate the initial spreading of the thermal compound."
I put A LOT more on than that. I will re apply tonight and see what happens.
you are not supposed to use a lot of it. like it said, just a grain in the center. the point of the compund is to fill the tiny imperfections in the surface of the heatsink, so there is no air pockets and the surfaces touch.( air gaps are good insulators, as in it doesn't transfer the heat), thermal compund takes care of that. but if you apply too much of it, ur transfering heat from processor through the paste, NOT directly into the heatsink. uneven distribution of too much of it will raise the temperatures.
You'll know you put too much if your heatsink feel like it's on a skating rink. I almost always put too much with seringe applicators, never found them good.
I applied mine on the CPU just so that I could barely see the grey color of the IHS & same for the copper color of the heatsink, needs to be thin, a lot thinner than most people think.
And even idling, I too have a core that's a bit higher than the other, but only by 1-2C
Message edited by Dechy on 05-06-2008 at 06:13:39 PM
So after reapplying the thermal paste, the temperature was still 49 degrees on the second core.
I downloaded Real Temp 2.5, and it reported the temps to be 32 and 39 directly after boot. (HWMonitor still said 42 and 49). However, Real Temp has a nice "test sensor" function and showed that the second core's sensor is stuck. It had no change in temperature after a few seconds of 100% activity, whereas the other did.
Is this going to be a big issue in overclocking, or can I just assume that both cores are within 5 degrees celcius of each other? I just got the processor from tigerdirect... should I try and use the warranty?
What you need to do is run orthos or prime to load, and see what the temps do.
Many of the 45nm's have dysfunctional DTS, but almost all are nonlinear and highly inaccurate at idle. Just that some happen to show the same inaccuracy at idle, ie both high, others one is high and one low. Also some will not even report temps at all, ie no movement, until a certain temp is reached.
If on load, they dont even out, and you really need to get it to 35 from tjmax, 70C with coretemp or 60C with realtemp to test the sensors. If they are within a few degrees C, might as well keep it, the next one you get may be worse....assuming they would allow an RMA in first place.
