Possible broken cpu/mobo, heat readings off?

wolfgang784

Honorable
Aug 7, 2012
9
0
10,520
I cannot seem to find too much information on my new processor but from what I know of how cooling works these temperatures seemed too good to be true, and after monitoring them they are also extremely sporadic.

My new processor is the AMD FX-9370 4.4GHz 8 core, although for some reason it comes 3.8GHz and boosts to 4.7GHz. My motherboard is the GIGABYTE GA-970A-DS3P. I turned off the turbo, C1, C6, APM, CnQ and APM while boosting it to 4.4GHz at 1.45v and ran it in prime95. I had ran it at 4.4GHz before disabling any of that stuff first and it did fine in prime95 until the third test, when the cpu would throttle down to 3.4GHz on all but 3 or 4 cores. The temperatures were strange before I even started OCing and before I adjusted those settings in the BIOS.

During a prime95 test at 4.4GHz 1.45v with a room temperature of 69F / 20C and a corsair H100i water cooling system:

started at 40F / 4C according to speccy and core temps, and that is also mirrored in the bios some of the time. Readings in the bios are generally 59F-69F / 15C - 20C though, which seems more believable. They do go as low as 4C even in the bios readings though.

The temperature almost instantly jumped to 82F / 27C when the test started and topped out around 85F / 30C. When the third test began most of the cores dropped to 3.5GHz and the temperature went down to 81F / 27C.

When the test was stopped they dropped to 40-50F / 6-10C almost instantly.

After I finished that test and recorded those temperatures from core temps and speccy I went and changed all of the settings in the bios that I mentioned above. When I restarted from that, I noticed my idle temperatures reading quite strange. As soon as windows booted I was fluctuating rapidly between 59F-90F / 13C-32C. This went on for about a minute before jumping between 55F-70F / 13C-21C. After about another three minutes the temperature was down to a solid 60F / 15C and did not budge.

From what I know of cooling in general and from what I have read on here (been reading a lot in the OC section), I should not be able to cool my parts below room temperature. Mine however are reading as low as a 20F and 15F difference to the ambient temperature.

Now I did read somewhere on here someone mentioning that the whole FX series does not read accurately until reaching approximately 80F / 27C but only that one guy mentioned it and I have not heard anything else about it. The idle temperatures seem impossible some of the time and the OC'd prime95 temperatures still seem waaaaay too low to be true, but then again I am quite new to this stuff. I wasn't even quite sure which section to put this in but eventually decided on here. So is my CPU busted? Motherboard reading wrong? Do I need to rip this thing back apart and RMA it already? Are these temperatures actually possible / happening? Any help is greatly appreciated as I would love to solve this as fast as possible so I know if I need to RMA It or not.

EDIT: I also just remembered (half asleep) that at one point last night I tried pushing it to 4.7GHz with 1.45v still and it ran fine with the whole first two tests, but the third test again posed a problem. The temperature readings all showed it dropping from around 90F / 32C down to 40F / 4C and back up to 90F / 32C repeatedly for about 30 seconds before I stopped the test because something was obviously wrong. The H100i fans could be heard stopping when it reached 40F / 4C and starting when it bounced to 90F / 32C but there was no drop and rise time in between, just straight to 40F / 4C and straight to 90F / 32C with the fans going on and off as they thought they were and weren't needed.



EDIT::: Since the time of writing the temperatures have been at a solid 62-72F / 16C-22C which is more believable I think, so I decided to try another prime95 test. As soon as the test began the temps jumped to 91F / 32C and right before the second test they had risen to 94F / 34C, the highest I have seen even when previously testing and hitting the third test. When the third test was reached temps rose to 98F / 36C and within maybe 20 seconds all cores throttled down to 1.4GHz and temps dropped to 63F / 17C instantly. I waited about 10 seconds but the test continued at 1.4GHz so I ended it because obviously something went wrong.
 

wolfgang784

Honorable
Aug 7, 2012
9
0
10,520
Not sure how to mark this as solved so I will try to mark this post of my own as the answer so I can close this thread. Turns out the VRM on the board is being overloaded from what I am told and simply need to get a better board that has 8+2 power phase instead of 4+1.