Core Temp reliability

Toggle sidebar Toggle sidebar
kineticman

kineticman

Distinguished
Feb 6, 2011
45
0
18,540
I've been struggling with the stability of an FX-8350 on an MSI 990FXA-GD80 board for a while now (mainly issues with cores erroring out under load until I bumped the voltages around and went through a bunch of BIOS flashes). Everything seemed stable for the past few weeks at 1.344v in bios and Prime95 never got the temps over 32c (according to core temp). I know that core temp isn't entirely accurate since I highly doubt my CPU is idling at 5c in an 18c room, but today my computer shut down via core temps overheat protection saying that the CPU was 118c! Is this just a glitch in the software, or is it possible that my mobo is overvolting my cpu?
 
Sort by date Sort by votes
jay2577

jay2577

Honorable
Jun 17, 2012
1,844
0
12,460
Try CPUID HW Monitor and let me know what temp that says.
Is the fan still spinning on the CPU?
Is the heatsink firmly attached to the CPU? Try and gently wiggle it
How much thermal paste is on the CPU?
Also have a look in the bios when you boot up and let me know what the temp in that says:)
Are you overclocked and if so at what speed?
 
Upvote 0 Downvote
arges86

arges86

Distinguished
Aug 31, 2008
1,775
0
20,360
I've had no issues w/ core temp. but it could be the software is reading the temp wrong from the die.
Its been known to happen.
as Jay2577 said, HW Monitor is a solid program.

With any temp issues I would immediately put the voltage back at stock (I think your chip's is 1.15)
and remove an OverClock.
Check for stability and correct temp readings, then slowly increase your numbers if you still want to push the chip.

Is your BIOS set to turn off the PC if the temps get too high as well? Some modems have that safety feature built in
 
Upvote 0 Downvote
kineticman

kineticman

Distinguished
Feb 6, 2011
45
0
18,540
To answer some questions: I use a Zalman 9900 Max heatsink that's mounted properly and has a good working fan. The CPU is running at stock clocks at 1.344v set in BIOS (hardware monitor and CPUZ read 1.28). I applied only a thin layer of thermal paste, AS5.

I have HW monitor too and it's backing up the various readings of coretemp (currently sitting at 5c). Unfortunately I have coretemp set to shut the system down in two seconds once a certain temp is reached so I didn't have time to launch anything or check a voltage after I saw the notification that the temp was 118c while browsing the steam store. When I rebooted I went into BIOS and saw the usual 35c temp and at the desktop noticed nothing out of the ordinary when I ran Prime95 (32c max) to check. I have no idea where coretemp got that number from but given all the problems I've had with this motherboard and fiddling I had to do with voltages to get the chip stable at stock settings I'm a little concerned that the board randomly decided to overvolt the cpu.
 
Upvote 0 Downvote
arges86

arges86

Distinguished
Aug 31, 2008
1,775
0
20,360
That voltage seems high...
its strange that CPU_z and the BIOS red two different measurements.
I would try just resetting the BIOS just to make sure
You may want to re-seat the CPU and the heatsink on it, see if that alleviates some of your issues
 
Upvote 0 Downvote
kineticman

kineticman

Distinguished
Feb 6, 2011
45
0
18,540


I also have that software and it's in sync with what coretemp and hw monitor say. I have core temp set to shut down the pc if the temp gets above 60c. All of a sudden for whatever reason and with zero load (or close to it) on the cpu my temp shot up to -according to core temp- 118c. Coretemp shut down the pc, I checked everything in bios on a restart, and everything there and in the os is normal again. I'm trying to figure out whether core temp was out of whack when it read that or if my motherboard could have tried to give the cpu way too much voltage.
 
Upvote 0 Downvote
jay2577

jay2577

Honorable
Jun 17, 2012
1,844
0
12,460
The other option is a faulty temperature sensor. I have 4 temperature sensors which are independent and connected to my fan controller.
If you have no way to monitor the temperature without the motherboard sensor you could try touching the heatsink on your CPU as near as possible to the base when it happens.
Please try this and report back on how hot that was
 
Upvote 0 Downvote
kineticman

kineticman

Distinguished
Feb 6, 2011
45
0
18,540
No matter what I try I can't get the cpu temp to jump up that high again and I've been messing with voltages and stress tests all night. I'm just going to chalk it up to a momentary glitch in either the sensor or core temp. I think if my motherboard gave my cpu enough voltage to spike the temp to 118c in spite of all the cooling I have it would have more or less fried my cpu.
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom