i5-2500 Voltage and Temperature

[mrcaffeinex]

Hello again everyone!

I am in the process of assembling a computer for a friend. The specs are as follows:

Intel Core i5-2500 (not the "K" model)
MSI P67A-G45 Motherboard
8GB (2x4GB) G.SKILL Sniper Series DDR3-1866
EVGA GeForce GTX 460 1GB SuperClocked Edition
Western Digital Caviar Black 500GB SATA III HDD
Seasonic S12II Bronze 620W PSU
Cooler Master HAF 912 Case

So here's the problem: when I got everything assembled, I started up the computer and went into UEFI setup. The system was showing a CPU temperature of 60C. I have never been a big fan of stock heatsinks, but since the chip came with one, I thought I'd try it out and see how it did...needless to say I was still unimpressed.

We had already planned on using an aftermarket CPU cooler, so I installed an Arctic Cooling Freezer7 Pro. The model had served me well on another build, so with my recommendation, that is the one that was purchased. It definitely was cooler, but the CPU temperature was still much warmer than I would have liked at 55C according to the system information in UEFI.

I started searching around on the internet and on the forums here, only to find conflicting stories about how the UEFI lacks any of the throttling features that would result in reduced temperatures when not under load. I figured I would load up Windows 7 and see what HWMonitor reported back to me. HWMonitor showed a CPU temperature of 37C "idle" at the Windows desktop.

To see if I could replicate a higher temperature within Windows, I started a Prime95 in-place FFT torture test. The CPU steadily climbed up to 60C and I stopped the test when it reached 65C because I am not interested in letting this chip burn up just yet...

I had previously used the Freezer7 Pro on a socket 1156 i5 760 that was overclocked and that chip never made it to 60C running an overnight Prime95 torture test.

I used Arctic Silver 5 with the Freezer7 Pro. The stock Intel heatsink came with its own thermal interface material. I triple-checked to make sure that each heatsink was secured, since the push-pin style connectors do not fill me with confidence...

So...what is going wrong here?

P.S. On a potentially related note, what is the proper voltage for the i5-2500 (non-K) CPU? UEFI is telling me that it is running at 1.225V. HWMonitor states that it is at 1.20V and I have yet to find where the voltage control options are in UEFI (if there even are any)...

-MrCaffeineX

 

[Kenmitch]

Temps in bios are always higher than idle in the OS for reasons you stated. Your temps may improve over time once compound settles.

Your voltage looks to be in the normal range.

When you did prime what was CPU speed? 2500 can be overclocked 4 bins and might be depending on mb settings.

 

[mrcaffeinex]

I will have to check the clock speed tomorrow when I take another look at it. I didn't change anything in UEFI, I just viewed the system information, made sure it was recognizing the memory, drives, etc and then saw the temperature, which is what started the Googling...

 

[lopri]

Also most BIOS temp reads are taken from the board itself. There are sensors around the hot spots on the board, thus it differs from the "core temp".

 

[mrcaffeinex]

I checked on the voltage in UEFI later last night and found where to adjust it. I reduced the voltage from "Auto" which was keeping it around the 1.225V mark I mentioned previously to 1.18 and it ran Prime95 overnight last night without crashing, reaching a max of 68C. I was really hoping for better thermal performance, but since it should never reach those temperatures under normal usage, I guess I will just accept them.

It is still making me scratch my head that my Phenom II 955 BE Revision RB-C2 (which is supposed to be the warmest running version) in the same case with a comparable cooler never gets above the upper 40s, even moderately overclocked. I would have thought that a 1.35V 125W quad-core would have taken the heat generating crown...