CPU Voltages - CPU Temperatures

[Sahir Dizdarevic]

So here is my situation. I am having problems with cpu temepratures for like 3years. I used to hit 99 on full load. Got stock cooler on it and my solution was to try with aftermarket cooler, but before that I changed my thermal paste and I get my temp on full load at 85, but just for like 1month and then again it used to go up until it reaches 99.

Last night I found out that voltages could be my problem.

I found that my Core voltage was on 1.26V if HWmonitor didn't lie. I gone to BIOS and temp in BIOS was 60 while core voltage was 1.164V.. I set offset to "-" and put in there 0.164 so basically Voltage was around 1V now.

And then I tested my pc in call of duty ww2 game for 30mins it worked perfectly and my temperature dropped from 99 to 73 max..

But interesting thing is that HWmonitor was still showing that max voltages were 1.26V.. (didn't I reduce it to 1.000V ??)

And one more weird thing is that on idle, I still get same temperature as before and it is 50-55 on every core.

HWmonitor after 30-40min call of duty ww2 mp (it used to be 99c (max) on every core) - https://imgur.com/a/sUSrXUC
-------------------------------------------------------------------------

So basically I would like if someone could inform me more about these connections between voltages and cpu temperatures.. How can I reduce my idle temperature too, or do I even need??

Thanks in advance.

 

[Karadjgne]

Bios is a heavy loaded app, it runs everything, all drivers, all cores everything is on. And not even close to idle speeds. That doesn't happen until the OS is loaded and takes control of things, stopping some processes, parking cores, lowering voltages etc as they are not needed for basic is functionality. It's not until you open something up that things come back to life, but even then there's still a bunch of stuff that's not active and shutdown right after app start, so temps only see a momentary spike then settle down to app usage. Bios temps live by their own rules and are not viable comparison to OS temps.
Yes. You most definitely need to reduce temps, anything over @70°C is considered unhealthy, over @80 should seriously raise an eyebrow and over @90 brings out the 'excuse me? What temp was that?' Anything over 90 is getting into throttling, shutdown, damage areas.

Aftermarket cooler means nothing. There's many aftermarket coolers out there that are not intended for some purposes and can be seriously worse than the stock cooler. The only cpus that get away with using just any aftermarket cooler are i3 users who push @50w or so. There's many budget coolers rated @90w that are plenty worse than the stock 140w cooler on an FX 8 series cpu. So you might not be doing anything really about temps, all depends on exactly which aftermarket cooler is on what cpu and what kind of airflow is supplying the cooler with fresh, cooler air. Even the biggest and best coolers can't do much if airflow is so restricted that the case isn't much different to an oven.

 

[Sahir Dizdarevic]

Okay, thanks for that info. But.. still I am not sure if I understand you well about what did I want to know. I will upload some pictures to let you see what did I do to reduce my temp on full load from 99 to 73. And I am wondering did I do it well, since I've never changed anything like this in BIOS.

https://imgur.com/a/NaVg34e

Did I do everything fine? Did I reduce my CPU voltage too much? Am I going to notice some problems/crashes? And when I reduced it, why still in HWmonitor I get that my VID's goes to 1.26V, if I reduced it to 1.008V. Or is that even the same thing (VID and CPU Voltage) ? ? ?

These measurements are all on idle: https://imgur.com/a/83oJSHd

Btw my rig is:
CPU: i5-2300 2.8GHz
GPU: r9 270x 2GB
RAM: 8GB
MOBO: P8H61-M LE

 

[Karadjgne]

Vcore is what you or bios have determined to send the cpu. VID is what the cpu is actually using. Most times they ate close, but different. Bios might be saying to give the cpu 1.26v but the cpu might actually use only 1.192v.

There's no way for any of us to know your cpu, regardless of pictures. You could set vcore as 1.192v, it'd show something like 1.186v VID and still be perfectly stable. Or it could be 1.192v vcore, 1.191 VID and bsod repeatedly as it's needing 1.208v. And that's the trick with lowering voltages, finding the lowest it runs on, until you loose stability, then bump back up a notch.

Testing for temps, use Prime95 version 26.6 small fft torture test. This gives 100%cpu loads on all cores. (if pc is unstable, probably crash, but no guarantees) on Air test for 10-15 mins at most.

For stability testing (if p95 temps are OK) use Asus RealBench. This'll put everything in the pc through tests, includes ram, memory controller, gpu, pcie buss, USB, Sata etc. Test for @1 hr. This hard test the relationships the cpu has with all the components. If there's any instability in the cpu, this test will find it.

 

[Sahir Dizdarevic]

If I understood you correctly.. Vcore is actually voltage that bios send to my cpu, but VID is what cpu is actually using in that moment.

But if you look at my 2nd picture (with hwmonitor one).. How is it possible that if Cpu Vcore (under motherboards list) says that max voltage is 1.07V (I determind that?) but out of 1.07V that bios sent to my cpu at that moment, my cpu used 'only' 1.26V.

How is it possible that my cpu is using more then I (bios) send to him? If vcore is 1.07v but VID is 1.26v.. In one moment...

Sorry I am trying to get it, but seems I don't understand.