So I did the latest BIOS and EC update to my GT72S 6QE and I'm noticing something very disturbing. Cores 0 and 2 temps are really hot, and cores 1 and 3 are where I expect them to be. The package temperature also shows really high as well.
Before the BIOS update, I repasted the CPU and GPU with Noctuna NH-T1 and made sure the fans were cleaned out. Ran the x264 stress test and the temps looked like cores 1 and 3 for all cores (fans on full blast).
Now my fans are still on full blast but now two cores and the package are reading very high temps. I wonder if I need to try and go back to an older BIOS revision. I can't imagine my thermal paste failed this quickly.
You'll see in the picture the temp graph. The spikes before the middle where from XTU running its stress test (with fans on full) and the second one is the x264 test (with fans on full)
-
Attached Files:
-
-
I doubt it is a BIOS thing. Might want to reopen and make sure everything is tightly placed and seated. You can try an older BIOS but I doubt that is the problem, because it would be widespread with other users.
-
Yeah I was mostly polling to see if it was wider spread. I re applied the thermal paste yesterday with this time spreading it out thin instead of using a pea size drop. I figured with the size of the CPU die that would be easier. It looks to be working now. Thanks for the assistance.
ryzeki likes this. -
Well now I'm at a loss for what is going on. For a few days, the core temps were all plus or minus 3-5 degrees of each other. Now I boot the system back up and there is a 20-25 degree gap from two of the cores to the other two. I can't believe that either (A) The cooler has come that far loose over the last two weeks, or (B) that the thermal paste (Noctuna NH-T1) has really degraded that much. I have the latest BIOS updates and the EC firmware. I think I'm going nuts......
Attached Files:
-
-
What is interesting is, at idle all the cores are running where they should in terms of temp. I wonder if two of the cores are turboing too high. I have tried locking in the core speed to "32" in the XTU and it didn't make the temperatures budge. Still seems way too much of a difference.
Attached Files:
-
-
Any suggestions?
-
Both the pictures you showed, show the cores being close in temps. Do you mean you are also getting very large differences in temp from the current setup, when you run at higher speeds?
-
Yeah I noticed that too, but way above he did have one screenshot with large differences while under load.
I just had nothing to add. I've never seen that large of a difference. Usually it's like the last two screenshots, within 5 or so degrees. No idea why they would read 25C diff unless it was just a fluke, or read error of some sort. -
There might be something wrong with the screws/heatsink too, that it is not properly being seated and with time, it makes a small enough gap to cause issues.
Are the same cores as the initial post, showing the same large difference or did the cores switched? -
Its the same two cores as what was going on before. I'll check the screws tonight and see if things have backed off. I'll re-tighten and check again.
-
So I reapplied TIM again, made sure everything was tight and clean. I ran this x264 test for about 9 hours and at 3.8GHz the temps look great to me. Now I shut down, went into the office, running the same test the temps are still great. So I'll keep testing and see what goes wrong when....
Attached Files:
-
-
Well I'm on Day 2.5 of this adventure, and the core temps are starting to split up a bit. Still not what I would call an issue at this point, I've been in somewhat temperature controlled environments and my overall temps have increased a bit. I'm going to go back to stock and see how it handles it from here....
Attached Files:
-
-
Things look much better at stock speeds. I may just leave it here and continue testing.
Attached Files:
-
-
Glad to see things managed to improve somewhat. It is weird that the temps are behaving that way. Maybe the CPU physically has issues at higher speeds requiring more voltage? worst case maybe the temp sensors might have issues or something, who knows?
Keep it at stock, or if you overclock, try to keep it at the BIOS settings instead of going higher. Sadly having a particular model CPU does not ensure we can run at a particular speed -
So I've been at stock since my last message. CPU is back to its old ways. I'm going to leave it stock forever, but this really sucks....
Attached Files:
-
-
Well the mystery may be resolved. I just popped open the bottom and would you look at that, the number 4 screw on the CPU heatsink was completely backed off....... So now my paste is dry and that's why those temps are worth junk. I'm off to contact MSI and see about getting a replacement Heatsink
-
I may decide to go the thermal pad route for the CPU vs. using paste as at least I can just tighten down the screws and not have my paste dry out. Any suggestions on what pads I could use?
Large Core Temp Fluctuation
Discussion in 'MSI' started by clayton006, Feb 16, 2016.