i7 4790K overheating, not the cooler

[zNitto]

So I am having a really odd issue with my almost 2 year old i7 4790K. I noticed yesterday that while playing battlefield 1 my cpu 1,2,3 and 4 temperature where spiking to 90c+ 1 going up to and past 100c. However it did not auto shutdown so I immediately shut it down myself and took off my h55i, I had only recently replaced the thermal paste so I thought I might have applied to little or too much so I cleaned the existing thermal paste off and applied a fresh squeeze. Upon reattaching my h55i I checked to make sure the bracket was secure and that I didn't apply too much paste, I tightened it down pretty much as hard as it would allow me. It was not loose and booted it straight to bios to check idle temps. My idle temp started at 50c and was soon at 60c on the bios temperature screen.

I started the think the issue must be in the cooler itself, So I took off the cooler and the bracket and fetched a stock intel heat sink that I had cooling another computer that I know works fine. Attached it and again booted to bios. But this time my idle temps were at 90c and climbing quickly. So I shut my computer down and started reading numerous forum posts. I read a few which said a batch of the i7 4790k's had odd sensor issues and were doa. However mine has been fine for the past 20 months with frequent monitoring and use.

I have never overclocked or touched the voltage going to the cpu, the cpu voltage was at a steady 1.105v. I am outside of my field of knowledge and am contemplating taking it to a computer store but I thought its worth a shot to ask here seeing as people know what they're talking about.

PC Specs
Motherboard: ASRock Fatal1ty H97
CPU: i7 4790K
Video Card: Asus GTX 970 OC
Memory: 16gb 1866MHz Corsair Vengeance c9
Power Supply: Thermaltake Toughpower G 650W
OS: Windows 7 Ultimate
SSD: Kingston HyperX Fury 240GB
Hard Drive: 2tb WD Blue
CPU Cooler: Corsair H55i

 

[Spring1898]

This is the second time I have heard something like this happen. I believe the last one became a warranty issue and intel wasn't a pain to deal with.
And since you seem to have ruled out cooling issues, it might be worth a call.

You could also temporarily reduce your maximum processor state in the bios and see if that drops your temperatures to make the computer usable.

Have you run CPU-Z to make sure nothing funky is happening with the multiplier in the windows environment?
And have you done any bios updates recently?

Do you have any additional software on your computer that could be set to "maximum" performance? That could be altering the CPU multiplier profile?

Aha found what I remembered:
http://www.tomshardware.com/answers/id-2411226/4790k-overheat.html

 

[iamacow]

How do you know its not the H55i? the 4790k is a hot chip to begin with and those smaller AIO coolers aren't up to the task anyways.

 

[Spring1898]

Even for an H55, that is an oddly high temperature for a stock 4790k, and idle on the stock cooler should not be above 15c over ambient.

I am starting to think (barring other more obvious answers) it could be a motherboard power delivery issue.

 

[zNitto]

How do you know its not the H55i? the 4790k is a hot chip to begin with and those smaller AIO coolers aren't up to the task anyways.

I thought it was the h55i at the start also but after taking a functioning stock intel cooler and got 90c+ I quickly ruled that out.

 

[zNitto]

Even for an H55, that is an oddly high temperature for a stock 4790k, and idle on the stock cooler should not be above 15c over ambient.

I am starting to think (barring other more obvious answers) it could be a motherboard power delivery issue.

I was also thinking that could be an issue. There was a power outage a week or two ago and even though I am running through a surge protector it was fine up until now.

 

[zNitto]

This is the second time I have heard something like this happen. I believe the last one became a warranty issue and intel wasn't a pain to deal with.
And since you seem to have ruled out cooling issues, it might be worth a call.

You could also temporarily reduce your maximum processor state in the bios and see if that drops your temperatures to make the computer usable.

Have you run CPU-Z to make sure nothing funky is happening with the multiplier in the windows environment?
And have you done any bios updates recently?

Do you have any additional software on your computer that could be set to "maximum" performance? That could be altering the CPU multiplier profile?

Aha found what I remembered:
http://www.tomshardware.com/answers/id-2411226/4790k-overheat.html

I still have my old pc and a laptop for my computing needs, but i'll definitely have a word with Intel right now, I didn't think of calling them because I got the i7 through a retailer and the warranty has expired.

I haven't really ran any diagnostics programs for fear of damage what could be repairable right now. And the fact that I'm getting so high temps in my bios and the fact I'm running a H chipset would rule out CPU multiplier profile would it not?

 

[iamacow]

The stock intel cooler will hit 90c+ normally. Its garbage and thats why they stopped including them. You could always Delid the CPU and drop the temps even more. You can also disable thermal throttling in the BIOS, but I would only do that if you 100% sure the sensor is bad.

 

[Spring1898]

I forgot that it was an H series,
That actually makes me lean more toward the motherboard as the culprit, possibly secondary to a PSU infarct during the storm.
Do you know of anyone who could pop your CPU into a different board.

 

[amtseung]

Bios temps are not idle temps. In bios, your motherboard slaps all your components across the face with voltage. Your CPU could be 55C in bios, but after idling at the desktop for 15 minutes, could drop to like 3C over ambient. At least, my 4790k does that.

Over-tightening the cooler onto the CPU can also have negative repercussions, either by warping the cooler, warping the socket, and/or warping the motherboard itself. A hair past finger tight is enough, especially on that particular bracket. It's mostly plastic, unlike the newer AIO's with square or hexagonal blocks, and can only be tightened so tight so many times.

I tried the stock intel cooler off my i5 4460 on my 4790k just now, center copper slug version. It got so hot, it wouldn't boot half the time, and just typing this puts the CPU at like 65C. Just because it works great on my locked i5 doesn't mean it'll work on an i7. Their heat outputs are drastically different. Your results don't surprise me.

Another thing to keep in mind is the number of people out there who have seen temps drop by over 20C by delidding their i7's, especially the 4790k, 6700k, and 7700k. It means Intel's manufacturing process doesn't result in ideal contact between CPU die and the thick copper integrated heat spreader that covers said die. Intel is also known to use crap thermal compound in there as well. In fact, I'm considering delidding my own CPU.

For reference temperature-wise, my H100i (240mm rad) has problems keeping temps down when the CPU is under full load, even with two of the best static pressure fans ever on there: good old Gentle Typhoons. I carved the H100i open, replaced that stiff nasty tubing, and added a reservoir because my H100i gargled like a drowning man. What really did the trick was removing the thick gasket-looking jet plate that directs water from the pump on top to the cold plate underneath, drastically improving water flow, and subsequently, load temps dropped... a lot. Let's just say, it used to thermal throttle within 60 seconds of launching P95 v26.6 small fft's, regardless of thread count. Now, with the modified H100i, my max load temps are around 86-88C peak, and average 80-83C when the loop reaches equilibrium. I am, however running my CPU at 44 multiplier at 1.2v, because it's unstable with any less voltage. I have no idea how you get away with 1.1V. Lucky you.

Spring1898 brings up some good points:

This is the second time I have heard something like this happen. I believe the last one became a warranty issue and intel wasn't a pain to deal with.
Have you run CPU-Z to make sure nothing funky is happening with the multiplier?
Do you have any additional software on your computer that could be set to "maximum performance" that could be altering the CPU multiplier profile?

Intel isn't a pain to deal with. Unlike the time I tried to RMA an FX CPU because my motherboard burst into flames.

Sorry for the wall of text.

 

[Spring1898]

The stock intel cooler will hit 90c+ normally. Its garbage and thats why they stopped including them. You could always Delid the CPU and drop the temps even more. You can also disable thermal throttling in the BIOS, but I would only do that if you 100% sure the sensor is bad.

I think the issue is that he was hitting 90c at idle on the stock cooler. That shouldn't happen

The chips do run hot, but the issue is the sudden change in temperature from what was his norm to what he sees now. So something is not correct in his system, either CPU, Motherboard, or PSU.

 

[zNitto]

So after a long phone call with Intel and doing a few diagnostics on the computer we came to the conclusion that for an unkown reason the cpu's core clock is running at 4.2-4.4 Mhz without the boost enabled. He seems to think that it could be a motherboard issue and recommended I try swap testing. The cpu itself is still under warranty so I don't think I'll try my hand a delidding just yet.

The issue with the cpu running a higher frequency is odd and we reset the bios to factory and it still runs at 4.2-4.4. I was originally getting the voltage reading from bios. But after installing CPU-Z my voltage is sitting between 1.2 and 1.3. This has me extremely puzzled as I have never overclocked or have the ability to on this chipset.

Also my multiplier is between x42.0 and 44.0

 

[Spring1898]

Sound like you may have a solution. Just double check to make sure there isn't something installed like Asus Power4gear, or Samsung performance tuning application etc that could be altering the windows power performance settings. The results you show from CPU-Z sound suspiciously like the case mentioned in the link I posted.

But I would get your PSU tested if you can as well.
Most PC shops have basic equipment to test it, and if they have a haswell chipset they can hook up you might be able to kill 2 birds with one stone.

 

[zNitto]

Bios temps are not idle temps. In bios, your motherboard slaps all your components across the face with voltage. Your CPU could be 55C in bios, but after idling at the desktop for 15 minutes, could drop to like 3C over ambient. At least, my 4790k does that.

Over-tightening the cooler onto the CPU can also have negative repercussions, either by warping the cooler, warping the socket, and/or warping the motherboard itself. A hair past finger tight is enough, especially on that particular bracket. It's mostly plastic, unlike the newer AIO's with square or hexagonal blocks, and can only be tightened so tight so many times.

I tried the stock intel cooler off my i5 4460 on my 4790k just now, center copper slug version. It got so hot, it wouldn't boot half the time, and just typing this puts the CPU at like 65C. Just because it works great on my locked i5 doesn't mean it'll work on an i7. Their heat outputs are drastically different. Your results don't surprise me.

Another thing to keep in mind is the number of people out there who have seen temps drop by over 20C by delidding their i7's, especially the 4790k, 6700k, and 7700k. It means Intel's manufacturing process doesn't result in ideal contact between CPU die and the thick copper integrated heat spreader that covers said die. Intel is also known to use crap thermal compound in there as well. In fact, I'm considering delidding my own CPU.

For reference temperature-wise, my H100i (240mm rad) has problems keeping temps down when the CPU is under full load, even with two of the best static pressure fans ever on there: good old Gentle Typhoons. I carved the H100i open, replaced that stiff nasty tubing, and added a reservoir because my H100i gargled like a drowning man. What really did the trick was removing the thick gasket-looking jet plate that directs water from the pump on top to the cold plate underneath, drastically improving water flow, and subsequently, load temps dropped... a lot. Let's just say, it used to thermal throttle within 60 seconds of launching P95 v26.6 small fft's, regardless of thread count. Now, with the modified H100i, my max load temps are around 86-88C peak, and average 80-83C when the loop reaches equilibrium. I am, however running my CPU at 44 multiplier at 1.2v, because it's unstable with any less voltage. I have no idea how you get away with 1.1V. Lucky you.

Spring1898 brings up some good points:
Intel isn't a pain to deal with. Unlike the time I tried to RMA an FX CPU because my motherboard burst into flames.

Sorry for the wall of text.

You bring up so really good points so after reading and doing some testing my idle temp after leaving my pc on desktop for around 10-15 minutes is in the low 40's. However benchmarking with OpenCL Julia4d at 1280x720 makes the temperature go up to 100c on all 4 cores in a matter of seconds.

The cpu cooler has been fastened properly, I have not tightened it enough to warp or break anything but it is done up tight enough to make good contact.

Like I said earlier whilst my cpu is still under warranty I am not going to think about delidding, as I've never done it before and don't feel like breaking it.

 

[amtseung]

Here are my two guesses for your increased voltages and multiplier:
1.) XMP can do things like this, because EXTREEEEEEME.
2.) Enhanced turbo. Essentially skips Intel's standard turbo function altogether and says, "Hey, your CPU tops out at 44 multi? Let's just run it at 44 multi all the time!" And the voltage follows suit.

Auto voltages running a 4790k at 1.3V+? Sounds like normal behavior to me. My motherboard tried to run at 1.4V Vcore on auto. Blew my mind as much as it blew all of my CPU coolers.

Have you tried fishing through your bios to make sure things, especially the enhanced turbo, aren't on by default? I know that on my MSI Z97 Gaming5, enhanced turbo is on by default, and needs to be manually disabled. I don't know if Asus handles it differently on a different chipset. Obligatory YMMV.

 

[zNitto]

Sound like you may have a solution. Just double check to make sure there isn't something installed like Asus Power4gear, or Samsung performance tuning application etc that could be altering the windows power performance settings. The results you show from CPU-Z sound suspiciously like the case mentioned in the link I posted.

But I would get your PSU tested if you can as well.
Most PC shops have basic equipment to test it, and if they have a haswell chipset they can hook up you might be able to kill 2 birds with one stone.

Yeah, this sounds like the most plausible answer however I don't have either one of those programs installed but there could be a program doing this I'm just not sure how to check which one it would be.

Would you recommend I just take it to a store and tell them what I've found so far?

 

[zNitto]

Here are my two guesses for your increased voltages and multiplier:
1.) XMP can do things like this, because EXTREEEEEEME.
2.) Enhanced turbo. Essentially skips Intel's standard turbo function altogether and says, "Hey, your CPU tops out at 44 multi? Let's just run it at 44 multi all the time!" And the voltage follows suit.

Auto voltages running a 4790k at 1.3V+? Sounds like normal behavior to me. My motherboard tried to run at 1.4V Vcore on auto. Blew my mind as much as it blew all of my CPU coolers.

Have you tried fishing through your bios to make sure things, especially the enhanced turbo, aren't on by default? I know that on my MSI Z97 Gaming5, enhanced turbo is on by default, and needs to be manually disabled. I don't know if Asus handles it differently on a different chipset. Obligatory YMMV.

Looking through the bios I can't really see anything like that. The only culprits are Intel Speedstep Technology and Intel Turbo Boost Technology. However reading through them they don't sound like the culprits. I'm honestly not to sure what I'm looking at however.
Either way I turned them both off, and now the core speed is capped at 3999.09 Mhz at a x40 multiplier. I'm unsure if the turbo boost will kick in under load now. But This looks like a fix for the meantime.

 

[zNitto]

UPDATE: So after sorting everything out and turning off the Intel Turbo Boost Technology in the ASRock Bios, it has solved my issue. The only thing I am a little worried about is as far as I've known that setting has always been on. But I've never seen my temps nearly that high before. Either way my temps are back down to mid 30's at idle and mid 70's under intense load.

I owe you all a lot of gratitude, you have saved me money and headaches. Thankyou.

 

[iamacow]

I think amtseung is on the right track. Your motherboard was overvolting and that little cooler can't handle it. Turning tubro off only kinda fixed the problem because now your CPU isn't running at its full potential. But the voltage problem went away.