i7-4790k overheat at 4.6 Ghz with Corsair H100i

[f688xt6]

So I recently switched my setup back to a cooler master Hyper 212 Evo Turbo X2 (etc...) due to an overheat issue that I had with the H100i. For some reason at sustained 100% load, the CPU jumps to 100 degrees C and thermal throttles after about 5 minutes of prime95. The odd part is that it's only recently. I've run 4.6 Ghz with multiplier only since I bought the CPU and never had a problem. After switching to the Hyper 212 turbo mega super ultra etc... I noticed the same condition, but with lower overall temps at idle or partial load. Also the delay was increased to about 7 minutes.

 

Is it really overheating? Or am I getting some sort of a false reading or what? No noticeable system hangs have happened, and I've had no symptoms of any issues. Never BSOD's and I'm kind of confused.

 

i7-4790k

ASUS Z97 Pro

Hyper 212 Turbo X2 (formerly H100i)

Gskill 2133 DDR3

Samsung 840 Evo 250GB

Cooler Master HAF 932 Advanced with non-stock high performance fans

 

[DannyRyu]

First Don't use Prime95. use something like Aida64 or Intel CPU Burn Test

Whats the Voltage?

[Godlygamer23]

I imagine the temperatures are accurate as the sensors are found on the CPU, and Intel doesn't usually have screwy sensors(note, I said usually). What voltage are you applying to the CPU?

[Godlygamer23]

Just now, DannyRyu said:

First Don't use Prime95. use something like Aida64 or Intel CPU Burn Test

I personally prefer Intel Extreme Tuning Utility. Intel Burn Test I think is an unnecessary stress test, and providing real world workloads is more important than stressing the hell out of the chip IMO.

[f688xt6]

I've never adjusted the voltage. It went to 4.6 out of the box with only a multiplier adjustment. Voltage is set on auto. I'm using the system now, at 4.4 and it's fine. I can load it and it will maybe hit... 76-78? I'd have to bench it again.

 

 

[wyattzx]

Firstly, Haswell is known to run hot, which means a thermal ceiling at 4.6ghz is not unheard of. However, your temperatures are far and beyond "normal." Did you mount the cooler properly? Use adequate (but not too much) thermal paste? How much voltage are you applying to the chip?

 

To give you an idea, I am on an NZXT Kraken x61, and at 4.5ghz @1.3v, I hit 85c+ in AIDA64, and I cannot push the chip any further. So I've backed it off to 4.3ghz @1.25v, and am content with that, though I do wish I could take my voltage own a bit; my idle temperatures are too high.

[f688xt6]

Been using Intel Extreme Tuning Utility lately. Is that acceptable? Why not Prime95? Just curious?

 

[Godlygamer23]

Just now, NonaHexa said:

Firstly, Haswell is known to run hot, which means a thermal ceiling at 4.6ghz is not unheard of. However, your temperatures are far and beyond "normal." Did you mount the cooler properly? Use adequate (but not too much) thermal paste? How much voltage are you applying to the chip?

 

To give you an idea, I am on an NZXT Kraken x61, and at 4.5ghz @1.3v, I hit 85c+ in AIDA64, and I cannot push the chip any further. So I've backed it off to 4.3ghz @1.25v, and am content with that, though I do wish I could take my voltage own a bit; my idle temperatures are too high.

I feel like the OP is confusing some people here. They claimed they had an H100i before, but now they're using a Hyper 212.

Hyper 212 Turbo X2 (formerly H100i)

[wyattzx]

Just now, f688xt6 said:

Been using Intel Extreme Tuning Utility lately. Is that acceptable? Why not Prime95? Just curious?

Prime95 is known to cause long-term damage to Intel CPUs as of the Sandy Bridge series. Due to pushing the chip further than it should, overloading it and ultimately causing thermal-based damage to the chip. That's likely a part of why you saw 100c.

[Godlygamer23]

Just now, f688xt6 said:

Been using Intel Extreme Tuning Utility lately. Is that acceptable? Why not Prime95? Just curious?

 

P95 can be screwy with the VRM that's integrated onto the CPU. Some people have had issues with P95 pushing more voltage than the user wants to the CPU, causing higher temperatures than normal. I think this is mostly for people that are using Adaptive Voltage, but I could be incorrect.

[wyattzx]

1 minute ago, Godlygamer23 said:

I feel like the OP is confusing some people here. They claimed they had an H100i before, but now they're using a Hyper 212.

He might be, but I tend to look past the coolers people are using, and stick to the numbers. Every single variable will affect how good your cooler is, you know? From airflow, to fan positions, amount of thermal paste used, how well it was mounted, et cetera.

[Godlygamer23]

Just now, NonaHexa said:

He might be, but I tend to look past the coolers people are using, and stick to the numbers. Every single variable will affect how good your cooler is, you know? From airflow, to fan positions, amount of thermal paste used, how well it was mounted, et cetera.

I also think the cooler is important, since the H100i is significantly better than the Hyper 212. Context is important because the temperatures wouldn't be unexpected if say the user was using the stock cooler for example or something along those lines, though the CPU might straight up crash with that.

[wyattzx]

Just now, Godlygamer23 said:

I also think the cooler is important, since the H100i is significantly better than the Hyper 212. Context is important because the temperatures wouldn't be unexpected if say the user was using the stock cooler for example or something along those lines, though the CPU might straight up crash with that.

Yes, but a Hyper 212, or an H100i can both drive the 4790k at 4.5ghz~ just fine. Regardless, this seems like less of a cooler issue, and more of a voltage issue, since OP isn't even sure what voltage the auto-tuner used. My past experience with auto-tuning has pushed my chips too far. Usually around 1.35v - 1.45v. (Though this is on a Phenom II x4 975, and FX-6300.)

[f688xt6]

Thanks for the input guys. I appreciate it. My last i7 was an 870 and could run Prime95 all day long at 3.6 without any issues, so thanks for the information. It would make sense that since the problem is limited to only Prime95 (as far as I can tell), that it could be pushing the CPU too hard in the ways explained here. I'm going to run XTU on it again and get an answer on that.

Also, sorry about the cooler confusion. I wasn't intending to mislead anyone.

[wyattzx]

3 minutes ago, f688xt6 said:

Thanks for the input guys. I appreciate it. My last i7 was an 870 and could run Prime95 all day long at 3.6 without any issues, so thanks for the information. It would make sense that since the problem is limited to only Prime95 (as far as I can tell), that it could be pushing the CPU too hard in the ways explained here. I'm going to run XTU on it again and get an answer on that.

Also, sorry about the cooler confusion. I wasn't intending to mislead anyone.

Would you kindly post a screenshot of CPU-Z running?

[f688xt6]

Looks like the motherboard is pushing 1.327v to the thing. Hmm... Also 91 degrees C on this run with a 212 at max speed. 1122 marks though... so yeah...

[APasz]

As far as I can see, not a single person has mentioned that maybe the cooler simply needs to be re-seated. Any slight movement of the mount is enough for the paste to move and become uneven and therefor not conduct heat as well.

[wyattzx]

3 minutes ago, f688xt6 said:

Looks like the motherboard is pushing 1.327v to the thing. Hmm... Also 91 degrees C on this run with a 212 at max speed. 1122 marks though... so yeah...

It looks like it's running at around 83c under load, though your utilization is at 94%. If this is your max temperature, it doesn't BSOD, and you're comfortable with it, then I'd say you're okay. 1.337v is a little bit high for Haswell chips, but not quite in the "danger zone." If you'd want to tweak it a bit, you could try dropping the voltage down to 1.3v~ or so. However as it stands, this is okay. Are you sure the H100i is worse? 

[f688xt6]

As far as re-seating the cooler, I have done that a couple of times. The results are always the same.

 

Yeah, the H100i was approaching 95c on full load. I was suspecting a pump failure, but the results are so close on the 212... The 83c was during benchmark, but not the final peak number which was 91c.

 

I was thinking about building a custom cooling system, but I wanted to make sure that the CPU wasn't defective.

[f688xt6]

Attempting another run at 1.302v... We'll see.

[APasz]

2 minutes ago, f688xt6 said:

As far as re-seating the cooler, I have done that a couple of times. The results are always the same.

 

Yeah, the H100i was approaching 95c on full load. I was suspecting a pump failure, but the results are so close on the 212... The 83c was during benchmark, but not the final peak number which was 91c.

 

I was thinking about building a custom cooling system, but I wanted to make sure that the CPU wasn't defective.

Ok, just wanted to make sure that it's been tried.

[f688xt6]

Interesting result... Higher marks, one degree lower. But under load I was seeing 1.312v

[f688xt6]

2 minutes ago, APasz said:

Ok, just wanted to make sure that it's been tried.

Thanks though. I do appreciate the input a lot!

[f688xt6]

Turned off adaptive voltage and it stayed constant at 1.302v. Starting to see more reasonable stuff here...

[wyattzx]

8 minutes ago, f688xt6 said:

Turned off adaptive voltage and it stayed constant at 1.302v. Starting to see more reasonable stuff here...

I feel like this is you right now.

 

 

Let me know how things go!