[SOLVED] CPU overheating issues - any help much appreciated

[stemeg]

Good morning all and thank you in advance for any light you may be able to shed on this situation. I am experiencing some overheating issues with my PC that I have had for 4 years. The main characteristics of the problem are:

· The CPU temps seem to idle in the 40s
· Low load (<10%) causes temperatures in the 50s or even low 60s
· When I am playing games (even one that are not resource intensive), the temps will often by up in the 80s. Once CPU load goes over around 30-40%, I’m hitting 100c
· Most times I restart the computer I get a CPU overheat warning. If I need to restart, I have to switch it off and leave it for a few minutes

Some specs:

· The CPU is an Intel Core i7 6800K
· The cooling system is a Corsair H100i v2
· The PSU is a Corsair CS Series 650W
· The motherboard is an ASUS X99-A II

It might be notable that I had some issues with cooling about a year and a half ago. A very similar problem (high idle temps, hitting 100c when under even a little load, CPU temp warning on restart), except slightly different in that back then the temp gradually increased to 100c over the course of a few minutes after starting Windows (even when CPU is idle). After that, it would gradually lower back down to the 40s/50s over the course of another few minutes. I don’t have that problem this time; it tends to be around 65c in Corsair Link when I first open it on startup, dropping to the 40s after a couple of minutes on idle. I tried cleaning the heatsink, changing the connector for the cooling block to a different header on the motherboard and switching the cooling pump and fan from quiet to performance with no luck but ultimately I solved the problem almost by accident: I noticed that the issue wasn’t occurring when I had switched the PSU off manually at the back between uses. So, since then, we switch off the PSU power after shutting down every time, and we’ve had nice low temperatures (20s idle, up to 50s when gaming) until very recently, when the latest problem started occurring. Whether this was anything to do with the current problem or not, does anyone have any idea why switching off the PSU at the back would have solved those overheating issues from before?

I am inclined to think I don’t need to reapply the thermal paste; we use the Arctic MX-4 compound which the manufacturers say should last for 8 years before reapplication is needed. Are there any other solutions anyone might suggest? The pump and fans seem fine, pump tends to stay around 3150rpm, the fan around 1800rpm. I’ve noticed the air coming out of the top (where the heatsink is) feels quite cool, although I noticed that also during the last issue mentioned above, so maybe that’s normal.

Is the cooling system simply failing? If so, was the issue a year and a half ago likely just a coincidence?
Thanks so much for any help anyone might be able to provide and sorry for the barrage of information!

 

[Phaaze88]

The cooling system is a Corsair H100i v2

Straight up: It's likely time to replace that thing.
It's just one of the downsides of hybrid coolers.

 

[stemeg]

Straight up: It's likely time to replace that thing.
It's just one of the downsides of hybrid coolers.

Yes you could be right, although that is what most people said when I sought help for the issue a year and a half ago and it turned out to be fine, so I want to eliminate any other possibilities first. It's also worth noting it has a 5 year warranty, so it should be lasting longer than this.

 

[Phaaze88]

I want to eliminate any other possibilities first.

1)Repaste and remount cooler. ✔.
2)Cleaned the radiator and fans. ✔?
3)Physically inspect the cooler under load. The heat should be spread evenly throughout the unit. ✔?
4)Unsecure the radiator and fans from the PC. Keep them elevated above the pump while the PC is running. [Only really applies if you've been running it mounted in the front. Otherwise: ✔!]

It's also worth noting it has a 5 year warranty, so it should be lasting longer than this.

Doesn't matter.
These things(pumps) can last anywhere between less than a year, up to 8 years, depending on several variables.
On average, it's a few years. That you got 4 years out of it is already good.

 

[stemeg]

Thanks for the suggestions I haven't reapplied the paste - I will try if there's no other obvious answer to what's happening but I've only done it before (with a different machine, a long time ago), so I'm a little nervous about messing it up!

I've physically inspected the cooler under load although could you be more specific with how to do this please? I can't reach into the radiator, although I've felt the two tubes; one is hot and one is fairly cool.

I have the radiator and fans above the pump already

With regards the warranty, I get that they can last for less than that time, but surely that's unlikely and something to consider when there are no other clear answers? Just logic would suggest that it must be fairly unusual for these items to fail to last their warranty, otherwise it wouldn't be profitable to manufacture them!

It's worth noting when I tried it this afternoon, it was considerably cooler (idled around 28c). No idea why, the only thing that changed is I had been in the case feeling the tubes. I have no doubt it'll be hotter again next time I switch it on.

Thank you again for the suggestions, much appreciated.

 

[Phaaze88]

Thanks for the suggestions I haven't reapplied the paste - I will try if there's no other obvious answer to what's happening but I've only done it before (with a different machine, a long time ago), so I'm a little nervous about messing it up!

Not hard at all.
Take some alcohol or cleaning wipes, and clean the old paste off.
Dry, or let dry.
Use the little in the middle application for the paste, except if the paste you're using specifies otherwise; EX: Thermal Grizzly Kryonaut requires a light, yet full cover with their provided spreader.

I've physically inspected the cooler under load although could you be more specific with how to do this please? I can't reach into the radiator, although I've felt the two tubes; one is hot and one is fairly cool.

Feel the cpu block, both tubes, and the radiator while you're running one of those '100C loads'.

If it is working normally, there should not be drastic differences between both tubes or the cpu block and radiator. The heat should be spreading throughout most of the unit after all.
If it's 'pooling' in 1 or 2 areas, something's wrong - most likely blockage, or the pump's actually dead; no vibration.

With regards the warranty, I get that they can last for less than that time, but surely that's unlikely and something to consider when there are no other clear answers? Just logic would suggest that it must be fairly unusual for these items to fail to last their warranty, otherwise it wouldn't be profitable to manufacture them!

Not unlikely at all - as I said, there's various factors in place, but the nitty-gritty of 'em all comes down to:
-pump operating speed
-pump exposure to air (Y/N)
-pump frequency: sitting at a static speed, or dynamically changing based on thermals

Over time, the biocides in the fluid disappear, so blockage is inevitable too.

Plus, depending on the scenario, the warranty isn't all that useful.

See, hybrid coolers are just the cheap and convenient alternative to custom liquid: you set it and forget it, and pretty much toss 'em out when they don't work anymore - well, keep the fans that still work, at least.

 

[stemeg]

Thanks for the suggestions I will feel the block and the tubes when it's overheating, but I won't be able to get to the radiator (it's obscured by the fans beneath and the case above). The air coming out of the top of it does feel cool - although that was the case a year and a half ago and I did manage to solve the issue then.

I think it looks like reapplying paste will be my first port of call, then I'll get in touch with Corsair about replacing it.

 

[alceryes]

As @Phaaze88 says, when you are getting into the warmer CPU temps (85+C) you should be able to feel the difference in temp between the inlet and outlet tubing near the waterblock. Although, the difference may be slight depending on whether there is line braiding/insulation. If you don't feel any difference you may have a pump issue.

You should also try an acoustic test. Grab the cardboard piece from the inside of a roll of paper towels. Put it to your ear and have the other end right next to the waterblock without touching it. What does the pump sound like? Any gurgling or hitching sounds? Is it just a constant hum?

Lastly, taking a step back, have you been monitoring your voltage? Did you update your BIOS recently? What does your CPU voltage get up to when gaming/benchmarking?

 

[stemeg]

Hi @alceryes thanks for your suggestions and your time. There is definitely a difference between the two tubes - one is much warmer than the other.

I did the acoustic test you mention and it was hard to get close enough due to the other components, but all I heard was steady hum.

As for voltage, good question, I don't tend to monitor it. There is a voltage listed as VCPU in Corsair Link, is that the right one? It seems always to be at 0.94 or 0.95v. I haven't updated the BIOS recently, no.

I hope those answers are useful!

 

[alceryes]

Is that the voltage at idle or when gaming/stress testing?
Run HWiNFO64 (sensors only, logging on) in the background while you game for at least 20 mins. Check out the log file produced to see what temps your CPU gets up to.

In Corsair LINK try setting your pump to 'performance' as long as the noise increase isn't that great. The difference is barely noticeable with my H80i v2.

 

[stemeg]

Hi @alceryes thanks for the reply and sorry it's taken me so long to get back to this. The voltage doesn't really change from 0.94-0.95v when it's under load. Is that supposed to happen?

 

[alceryes]

Are you sure you're looking at the vcore? Your CPU should go up to around ~1.1-1.2v when it's under full load - maybe even higher.
Is everything at default? Have you performed a BIOS reset? Write down/take pics of your current settings, just in case.

 

[stemeg]

Thanks again @alceryes - yeah the vcore doesn't seem to go any higher at all which I guess is strange. However, I replaced the cooling unit with a h100i pro XT today and, good news, it's back down to normal temperatures! I'll keep an eye on it but it's idling now at ~22c. One difference that I can tell straight away is that the air being pushed out of the top feels warm when before it felt cool. That evidences that it was indeed the pump starting to fail, I think.

Thank you again everyone for the help!