CPU super hot under load. Idle temp fine.

[timberhoy]

Setup:
i7 4790k (no overclock)
Corsair H60
Gigabyte Z97X-UD5H
Nvidia GeForce 770
Corsair RM650 power supply
RAM: Crucial Ballistix Tactical Tracer 16GB Kit (8GBx2)
Corsair Carbide 500R (two front intake fans(120mm), one rear exhaust fan(120mm), one side intake fan (200mm), two top exhaust fans (120mm with one pulling air through the H60 radiator).

My setup has been stable and fine since I built it initially a little over a year ago. Last week I was playing CS:GO and my computer hard shutdown (ie straight black screen) and auto rebooted. I quickly removed my headphones once it black screened and realized that my fans were the loudest I had ever heard them. My PC finished booting back up and I was able to get back in game, but my fans were still running super loud. Finished the match I was playing, shutdown the computer, and went to bed. Installed Core Temp the next morning to see what was going on with my temps. In the past I've idled around in the lower 30s, and lived in the mid 50's during 2-3hr CS:GO sessions. Suddenly now while playing CS:GO I'm hitting 70's within 20 minutes of starting a match.

That lead me to believe that perhaps the pump was going in my H60, or maybe the thermal paste was going bad. I opened up my case, took a look at my H60 to check that it was mounted solidly (it was). I went ahead and removed it, cleaned off the OEM paste, applied Arctic Silver 5 (pea method), reinstalled the cooler. Booted everything back up and the temps were the same as they had been previous to the repaste/reseat. Went ahead and requested a RMA from Corsair for the H60 thinking that it was for sure my problem. Went to Fry's today and picked up a Thermaltake Frio Silent 14. Installed it, and my temps are still getting incredibly high under load. (But still idling at 33.) With the Thermaltake cooler my PC got through 6 mins of Prime95 before the MoBo auto killed it.

To me that seems to rule out the H60 as being the culprit. So now the question is, what else could be my problem? I haven't installed any new programs or changed any settings between when my H60 was doing perfectly fine keeping my CPU cool and now. I've tried swapping coolers to confirm a H60 problem, but the issue is persisting. What have I missed?

 

[Benjiwenji]

What is your temp under Prime95 SmallFFTs?

70c is safe for haswell. Junction point is around 100c, CPU won't even throttle below that temp.

What is your PSU. That would be the main suspect. PSU = most important piece of hardware but is rarely mentioned in spec sheets.

Also go to event viewer to research the BSOD codes.

 

[timberhoy]

What is your temp under Prime95 SmallFFTs?

70c is safe for haswell. Junction point is around 100c, CPU won't even throttle below that temp.

What is your PSU. That would be the main suspect. PSU = most important piece of hardware but is rarely mentioned in spec sheets.

Also go to event viewer to research the BSOD codes.

I have the Corsair RM650 power supply (added it to the main post).

I understand that 70c is safe, and that throttling isn't supposed to happen until 100c but what I've been seeing is extreme lag at 80c and the highest temp I've seen is 87c and the moment it hits 87 it shuts off.

Last time I started a Prime95 SmallFFTs test my PC turned off. Will try it again and report back error codes. I have not been getting a BSOD. It's been a straight black screen shutdown.

 

[timberhoy]

What is your temp under Prime95 SmallFFTs?

70c is safe for haswell. Junction point is around 100c, CPU won't even throttle below that temp.

What is your PSU. That would be the main suspect. PSU = most important piece of hardware but is rarely mentioned in spec sheets.

Also go to event viewer to research the BSOD codes.

Current temps with Prime95 SmallFFTs is 68-70c and then 6 minutes 30ish second mark a straight shutdown. Looked in event viewer and the only event that shows up is the "The previous system shutdown at 2:07:19 PM on ‎1/‎22/‎2016 was unexpected." from when the PC reboots itself.

 

[Benjiwenji]

No other errors before that?

70c is safe temp for your CPU under load, have you changed any settings in the BIOS? Voltage, RAM timings...etc? Try resetting your BIOS settings to default and see if anything changes?

Have you changed made changes to your RAMs lately?

The RM650 is an average PSU made by CWT, how long have you had it?

 

[timberhoy]

No other errors before that?

70c is safe temp for your CPU under load, have you changed any settings in the BIOS? Voltage, RAM timings...etc? Try resetting your BIOS settings to default and see if anything changes?

Have you changed made changes to your RAMs lately?

The RM650 is an average PSU made by CWT, how long have you had it?

No other errors are logged before that.

All my BIOS settings were defaults when the problems started. I've already reset my BIOS twice. The only thing I've changed since the BIOS reset is the CPU Fan to always on instead of the "normal"/default setting.

I haven't changed anything with my RAM since the initial build in Dec 2014. Same deal with the PSU, bought it in Nov 2014 and have been using it since I built everything in Dec 2014.

To further confuse things I was able to play CS:GO for a half hour earlier today without a force shutdown. The only thing right now that 100% triggers a black screen/power pulled shutdown is a Prime95 SmallFFTs test. I've never had the SmallFFT one last longer than 7 minutes.

 

[Benjiwenji]

All your drivers up-to-date? Particularly the Nvidia GPU driver?

If so, then try running the Prime95/GS:GO with just one stick of ram at the time, see if things changes?

Have a spare PSU around to test the system to see if PSU has gone bad? That's would be a possibility in random shut downs.

Check to see if any power connections are loose.

 

[Neur0nauT]

It sounds like the PSU's voltage regulator is faulty to the 24v connector. Definitely test with a different PSU. Don't risk anything else before you check this or you risk killing other components.

 

[timberhoy]

All your drivers up-to-date? Particularly the Nvidia GPU driver?

If so, then try running the Prime95/GS:GO with just one stick of ram at the time, see if things changes?

Have a spare PSU around to test the system to see if PSU has gone bad? That's would be a possibility in random shut downs.

Check to see if any power connections are loose.

Drivers are up to date. (First thing I checked.) I've also reseated my GFX card as well.

I'll open up the case and remove a stick of RAM and see how the Prime95 test goes then. If it still fails around the 6 minute mark I will reseat the CPU (because I haven't tried that yet and trying everything seems to be a good idea), reset the MoBo, and go from there with both RAM sticks installed.

I do not have a spare PSU around, and when I installed the air cooler I double checked all my power connections. I don't think I can straight up rule out the PSU as being the problem, but it doesn't seem like the shutdowns are random at all. They seem to be tied to high CPU temps. The one thing I'm unsure about is why my MoBo (or other component??) is literally yanking power when the CPU hits 80+. I can't seem to find any settings to change that number. If technically the CPU should still be capable of running at 90c, why is it dying at 80, 87, and such?

 

[Benjiwenji]

That's why I suspect a power issue, either the PSU or mobo, rather than a temperature one.

Is your mobo VRMs overheating? What are the temperatures?

 

[Neur0nauT]

I'll add that overheating is caused by excessive voltage.....My bet is on the PSU.

 

[Benjiwenji]

I'll add that overheating is caused by excessive voltage.....My bet is on the PSU.

Weird thing is that nothing has been overheating so far. Mobo VRMs temperatures are often overlooked, but I doubt that that is overheating either.

RAM can often cause crash as well.

 

[Neur0nauT]

I'll add that overheating is caused by excessive voltage.....My bet is on the PSU.

Weird thing is that nothing has been overheating so far. Mobo VRMs temperatures are often overlooked, but I doubt that that is overheating either.

RAM can often cause crash as well.

It could also be a faulty Northbridge on the motherboards chipset. Touch the heatsink on the Northbridge (check your mobo manual for the location) when things are overheated, versus the rest of the board. If it is crazy hot to the touch...then it could be to blame.

I'd still test with a different PSU before anything else though...since the rest of the system seems nominal for the most part.

 

[timberhoy]

That's why I suspect a power issue, either the PSU or mobo, rather than a temperature one.

Is your mobo VRMs overheating? What are the temperatures?

According to HWiNFO VR T1 is at 37c and VR T2 is at 60c at idle. I know VRMs can get a lot hotter than CPUs, but 60c at idle seems excessive. Going to run Prime95 and see what it gets to real quick before trying with only one RAM stick.

 

[Karadjgne]

I'm betting on the psu, but for different reasons.

First, I'd check windows errors. There will be a critical error for every shutdown. This'll be a place to start hunting.

You have a k cpu on a Z board. OC that cpu. But not by raising the multiplier, by undervolting the cpu. Intel cpu's are famous (infamous) for much higher than needed vcore, so instead of auto, which will be in the 1.2v+ range, try something closer to 1.08v. That alone can drop cpu temps 10° or more at load.

Back to psu. The original Corsair RM series were plagued with thermal problems, some (750/850) would not test under heavy loads without shutting down due to overheating. So as bad as the thermal properties on those designs are, if the fan is failing, it'll overheat easily, forcing an instant shutdown to prevent damage.

 

[Neur0nauT]

I'm betting on the psu, but for different reasons.

First, I'd check windows errors. There will be a critical error for every shutdown. This'll be a place to start hunting.

You have a k cpu on a Z board. OC that cpu. But not by raising the multiplier, by undervolting the cpu. Intel cpu's are famous (infamous) for much higher than needed vcore, so instead of auto, which will be in the 1.2v+ range, try something closer to 1.08v. That alone can drop cpu temps 10° or more at load.

Back to psu. The original Corsair RM series were plagued with thermal problems, some (750/850) would not test under heavy loads without shutting down due to overheating. So as bad as the thermal properties on those designs are, if the fan is failing, it'll overheat easily, forcing an instant shutdown to prevent damage.

This is all very valid if the CPU was being overclocked but its not. The PSU seems to be the first port-of-call.

 

[Benjiwenji]

60c is high at idle when you're running stock. What is it at load? I think that's more important.

Your case seems to have good air flow too....

 

[timberhoy]

60c is high at idle when you're running stock. What is it at load? I think that's more important.

Your case seems to have good air flow too....

VRMs under load were at 45c and 65c. So practically no change between idle and load.

 

[Neur0nauT]

60c is high at idle when you're running stock. What is it at load? I think that's more important.

Your case seems to have good air flow too....

VRMs under load were at 45c and 65c. So practically no change between idle and load.

All of these tests aren't proving much. The original source of overheating is from the power source. I don't know what else to say.

 

[timberhoy]

60c is high at idle when you're running stock. What is it at load? I think that's more important.

Your case seems to have good air flow too....

VRMs under load were at 45c and 65c. So practically no change between idle and load.

All of these tests aren't proving much. The original source of overheating is from the power source. I don't know what else to say.

I don't currently have a way of testing the PSU or getting my hands on another PSU. I understand that it could very well be the PSU, but since there are other things I can actually test I'm going to follow up on them until I can test the PSU.

 

[Benjiwenji]

Within normal range. So the suspect is the PSU. Although it is not certain, there isn't much else to do. Karadjgne's comment has been particularly helpful. Try to manually set the vcore as suggested by him.

Is there a part store near by or a family/friend that has a PSU that you can use?

 

[timberhoy]

Within normal range. So the suspect is the PSU. Although it is not certain, there isn't much else to do. Karadjgne's comment has been particularly helpful. Try to manually set the vcore as suggested by him.

Is there a part store near by or a family/friend that has a PSU that you can use?

Going to have to go to Fry's tomorrow morning to see what options they have in stock.

 

[timberhoy]

I'm betting on the psu, but for different reasons.

First, I'd check windows errors. There will be a critical error for every shutdown. This'll be a place to start hunting.

You have a k cpu on a Z board. OC that cpu. But not by raising the multiplier, by undervolting the cpu. Intel cpu's are famous (infamous) for much higher than needed vcore, so instead of auto, which will be in the 1.2v+ range, try something closer to 1.08v. That alone can drop cpu temps 10° or more at load.

Back to psu. The original Corsair RM series were plagued with thermal problems, some (750/850) would not test under heavy loads without shutting down due to overheating. So as bad as the thermal properties on those designs are, if the fan is failing, it'll overheat easily, forcing an instant shutdown to prevent damage.

Tried changing the vcore to 1.08 and couldn't get it to boot. Will try more options tomorrow (including a new PSU).

 

[timberhoy]

Went and got a Thermaltake Smart 750 from Fry's today. Installed it, tried running a Prime95 SmallFFT test and the PC died upon hitting "run". Ran a Prime Combo test and it lasted for 6 minutes 42 seconds before hitting 83c and doing a hard shutdown.

Soooo it's not the PSU? What next?

 

[timberhoy]

Somehow missed that I was running Prime95 28.7 instead of version 26.6. Downloaded 26.6 and was stable for a 10 minute SmallFFTs test (stopped it at 10 minutes, no failure).

Perhaps it was simply my H60 that failed during the original CS:GO match that crashed my PC and my testing method of Prime95 28.7 was at fault for the replacement parts test?