5960X overheating in BIOS

[Bob Wilson]

Hi all, I'd be grateful for any advice on troubleshooting this issue.

 

i7-5960X in an Asrock X99 Extreme4 board and Coolermaster Sedion 120 AIO.

 

3.0Ghz base, it was stable at 4.0Ghz but I ran it for the past 3 years at 3.7Ghz and never had temps above 85 degrees, even with sustained 100% loads.

 

Running Windows 10, received an update recently and the system went to shit. MBR corrupted, got that fixed and was loading into Windows when the machine just powered off. Wouldn't even turn back on. Left it for an hour, it booted, started loading Windows and then again turned off.

 

Waited an hour and turned it back on an into BIOS and using the hardware monitor in there just watching the CPU temp climb with no load, just sitting in BIOS. It got to 105 degrees and then powered off (obviously overheat protection kicking in).

 

So I remove the XMP profile and default BIOS back to default. Reboot, and exact same behaviour.

 

Thinking it's the cooler, I touch the heatblock and it doesn't even feel warm. That's weird.

 

Thinking it might be the TIM, I took it off and whilst it had dried out it wasn't flaky. But replaced it with some Coolermaster Master Gel, remounted and turned it on, exact same thing - temp just keeps climbing in BIOS and then turns off.

 

Thinking then that it might be the pump, I swap the AIO out for a Thermaltake Contac9 Air Cooler. Exact same behaviour.

 

Flashed latest BIOS. Exact same behaviour.

 

So now I'm stumped. Could it be a faulty sensor? Something wrong with the board? Given the rareness of both board and replacement CPUs, it's going to cost a bit of money (and time) to find replacement bits to swap out to troubleshoot. Any ideas would be greatly appreciated.

 

[Levent]

Sounds like pump is not connected or it died.

[Chris Pratt]

People seem to assume that the BIOS is a good place to check idle temps, because there doesn't seem be any direct usage. Nothing could be further from the truth. BIOSes as a general rule tend to be some of the most shight software out there. They're often bloated and unoptimized, and importantly, also don't generally have good CPU management, like letting cores sleep, etc.

 

The result is the BIOS actually puts a significant load on the CPU. Boot into Windows and monitor your temps there. If the CPU temperature continuously climbs like it did in the BIOS, then you have a real problem. Otherwise, don't worry about it.

[Bob Wilson]

Thanks for the replies.

 

@Levent that was my first thought, that's why I swapped out the cooler but same behaviour observed.

 

@Chris Pratt trouble is that even from a cold boot, I barely am able to log into windows before it thermal shutdowns. So I don't think the observed bahaviour in the BIOS is any different to what is actually happening, hence why I think it's a hardware fault somewhere.

[ZROCK]

19 hours ago, Bob Wilson said:

Hi all, I'd be grateful for any advice on troubleshooting this issue.

 

i7-5960X in an Asrock X99 Extreme4 board and Coolermaster Sedion 120 AIO.

 

3.0Ghz base, it was stable at 4.0Ghz but I ran it for the past 3 years at 3.7Ghz and never had temps above 85 degrees, even with sustained 100% loads.

 

Running Windows 10, received an update recently and the system went to shit. MBR corrupted, got that fixed and was loading into Windows when the machine just powered off. Wouldn't even turn back on. Left it for an hour, it booted, started loading Windows and then again turned off.

 

Waited an hour and turned it back on an into BIOS and using the hardware monitor in there just watching the CPU temp climb with no load, just sitting in BIOS. It got to 105 degrees and then powered off (obviously overheat protection kicking in).

 

So I remove the XMP profile and default BIOS back to default. Reboot, and exact same behaviour.

 

Thinking it's the cooler, I touch the heatblock and it doesn't even feel warm. That's weird.

 

Thinking it might be the TIM, I took it off and whilst it had dried out it wasn't flaky. But replaced it with some Coolermaster Master Gel, remounted and turned it on, exact same thing - temp just keeps climbing in BIOS and then turns off.

 

Thinking then that it might be the pump, I swap the AIO out for a Thermaltake Contac9 Air Cooler. Exact same behaviour.

 

Flashed latest BIOS. Exact same behaviour.

 

So now I'm stumped. Could it be a faulty sensor? Something wrong with the board? Given the rareness of both board and replacement CPUs, it's going to cost a bit of money (and time) to find replacement bits to swap out to troubleshoot. Any ideas would be greatly appreciated.

 

It sounds like the sensor on the motherboard that monitors the cpu temps is reading incorrectly. Conisidering its not hot to touch and it couldn't possibly get hot enough to restart the computer in bios even with no cooler installed. Being me, I would get a motherboard to throw at it and see if it fixes the issue. Im almost positive it would. However there is probably another way to fix the issue Im just not sure how. 

[Bob Wilson]

@ZROCK many thanks, kinda what I was thinking. I've reached out to a few places to see if they can diagnose this board before going down the replacement option, X99 boards are both rare and expensive (at least here in Oz).

 

[Guest]

On 2/13/2022 at 6:12 PM, Bob Wilson said:

Thanks for the replies.

 

@Levent that was my first thought, that's why I swapped out the cooler but same behaviour observed.

 

 

What other cooler? 

The first one wasn't enough for a 140w 16 threaded 20mb cache Extreme Edition processor in the first place..... Coolermaster Sedion 120 AIO

 

85c on that 120 AIO? Eh, that's hard to swallow.

[Guest]

On 2/13/2022 at 7:11 AM, Chris Pratt said:

People seem to assume that the BIOS is a good place to check idle temps, because there doesn't seem be any direct usage. Nothing could be further from the truth. BIOSes as a general rule tend to be some of the most shight software out there. They're often bloated and unoptimized, and importantly, also don't generally have good CPU management, like letting cores sleep, etc.

 

The result is the BIOS actually puts a significant load on the CPU. Boot into Windows and monitor your temps there. If the CPU temperature continuously climbs like it did in the BIOS, then you have a real problem. Otherwise, don't worry about it.

This is pretty correct. Was viewing my Dry Ice temps yesterday (and probably later this evening) and was reporting about double the actual temp. It was saying the Cpu 980X was running -108c when it was closer to about -58c. Core-Temp could only record down to -15c with a question mark next to it lol. 

 

Generally a bios will use 1 core. Newer UEFI perhaps 2 cores, but I don't think so. It would be overkill for display rendering, which is what uses the core. 

[Bob Wilson]

@ShrimpBrime Thanks for your thoughts. The other cooler I tested was a cheap air cooler I had on the shelf, Thermaltake Contac9 Air Cooler. It was more just to see if it made any dent in the rate that the temp climbed in BIOS, and it didn't. Re the AIO, I agree with you. I didn't spec it with that, I was handed the machine for work and now that it's died the company just gave me another one (i9-10900K) and at least this one has a 240AIO on it.

 

They've essentially told me to get rid of this machine but I'd love to be able to run it as a gaming rig - it's got 64GB of RAM, 1TB of NVMe storage and a GTX980 in it so it will still run a lot of the games that I play.

 

I might try booting into linux from a USB and see if that makes any difference, I can't imagine it will but its a cost-free test.

[Guest]

1 minute ago, Bob Wilson said:

@ShrimpBrime Thanks for your thoughts. The other cooler I tested was a cheap air cooler I had on the shelf, Thermaltake Contac9 Air Cooler. It was more just to see if it made any dent in the rate that the temp climbed in BIOS, and it didn't. Re the AIO, I agree with you. I didn't spec it with that, I was handed the machine for work and now that it's died the company just gave me another one (i9-10900K) and at least this one has a 240AIO on it.

 

They've essentially told me to get rid of this machine but I'd love to be able to run it as a gaming rig - it's got 64GB of RAM, 1TB of NVMe storage and a GTX980 in it so it will still run a lot of the games that I play.

 

I might try booting into linux from a USB and see if that makes any difference, I can't imagine it will but its a cost-free test.

You should really tackle the cooling issue....

 

ThermalTake Contac9 has all of 3 lousy heat pipes. I can't imagine it handles more than like 85w max.

 

You could try and disable the HT and clock to Base frequency and run the voltage low as you can. MAYBE you could use the system. But that's on you, I take no responsibility for burnt up hardware.  

[Bob Wilson]

Just to follow up on this, I finally had a chance to pull the MB from the case and do some diagnostics.

 

Pulled all the DIMMs, re-seated one in A1 slot, re-pasted and re-seated the AIO and booted, still running up to 105C and then thermal shutting down.

 

Let it cool and then flashed bios from backup bios (very useful feature that one).

 

Back into bios, it ran up to 105C but didn't shutdown, and then all of a sudden it just plummeted in temp and hung out around 35C. It sat there for 5 minutes, happily idling at 35C.

 

Re-seated all the RAM and sure enough, back to 105C and thermal shutdown.

 

Removed RAM and proceeded to add one stick at a time. Booted and hung out at 35C, until I put the seventh stick (there's 8 slots) in and then up to 105C.

 

Pulled it out and put the other DIMM in the same slot, 35C.

 

Thinking it was a dodgy DIMM, I seated it into the last slot and wouldn't you know 35C.

 

So in the end I really don't know what was wrong - could have been corrupted bios. Could have been that the DIMM is in fact dodgy but for whatever reason it doesn't get affected in the particular slot that it is in. All very weird.

 

All I know is that I now have a pretty decent gaming rig for free (well, for the cost of 2 hours of my life) and in the current state of desktop computing, that's a win!

 

Thanks everyone for your help!