High temps

Associate
Joined
13 Sep 2013
Posts
6
Hi all,

Spec: i7 6700K Skylake, Corsair H100 AIO cooler, 16gb Corsair DDR 4, MSI Nvidia 1070, Asus Maximus VIII Ranger mobo

I've had my PC built for a couple of years now and never bothered to OC it, although when I bought it I did read up on the CPU to see if it had scope for overclocking and had it on my radar, just never bothered to, until today.

I downloaded CPU-Z, SpeedFan, HWMonitor and Prime95 and after much reading on overclocking the 6700k decided to go with a simple multiplier bump from 40 to 42 aiming for 4200mhz with voltage set to 1.25. I ran Prime95 and everything seemed stable, no BSODs, and at this stage I was using SpeedFan looking at the CPU temp and under stress testing the CPU temp was reading around 65/70 max, so I thought all was well. I then aimed for 4.4ghz which at 1.25 caused a BSOD pretty quick, so I bumped the voltage up to 1.3 and tried again and Prime95 seemed stable - but at this point I switched to using HWMonitor and **** myself when I saw the temps.

Under 'Temperatures' in HWMonitor, 'Package' and 'Core #0' (through to 3) were all reading around 98-102c. I hit the stop button on Prime 95.

Intruiged, I reset the BIOS to factory defaults to wipe out the OC, booted up and ran Prime95 again with HWMonitor running - same thing again, core temps reading super high.

The 'CPUTIN' reading was around 55-60c whilst the core temps were ~100c, and SpeedFan's CPU temp reading is in sync with the CPUTIN.

What's also confusing is the core temps are quite happily sitting around 34c (often see this around 28c or less on cooler days) when idle, but within a second of starting the stress test they immediatly spike to ~100c - it's like they don't even need time to reach that temp. As soon as the test stops, they immediately fall back to reasonable temps.

For what it's worth - before all this OC'in lark I've had zero issues with the PC and stability, playing BF4/BF1/BF5 for hours on end in a fairly hot study due to the weather and never had any crashes or freezing up.

Can anybody explain what's going on?

HWMonitor / CPU-Z screenshots

The ~100c temps are during Prime95 execution, Small FFTs test)

bizzare-temps.png


After stopping Prime95:

bizzare-temps-stopped-prime95.png


Thanks for reading!
 
After reading several other posts on people having similar problems I decided to check my PC for any airflow issues / check the CPU cooler block position.

I feel like a bit of an idiot for what I discovered, however I will say I'm no better off / closer to a solution right now.

I'd already recently brushed / vac'd the inside of the PC and removed the case fans, cleaned them and put them back in place - however I'd missed / forgot to take the top plate of the NXZT H440 case.

I was disgusted by what I found - the layer of dust / fluff that had built up was grim and I was confident that once it had been cleaned out it'd make a huge difference to the temps, here it is after cleaning (still not perfect):

20200522_004016.jpg


20200522_005342.jpg


I tightened the CPU block screws as they felt fairly loose (didn't over tighten) and I installed the Corsair iCUE management software.

Upon installing / launching iCUE my PC got significantly louder which, although noisy, gave me hope that maybe the fact I'd never had the Corsair drivers running may mean I've been under utilising the performance of the water cooler.

So now with a clean radiator which should be flowing, a better seated CPU cooling block and the Corsair software running, things should be looking better.

My idle temps went from 28-32c to around 40-45c after having installed the iCUE software - read several forum posts around this and other seem to have experienced it too, some pointing to the Corsair software itself consuming a disproportionate amount of system resource.

corsair-coolant-temp.png


As I write this the coolant temp is sat at 25.4c, cores around 25-30c, CPUTIN at 30c.

I read somewhere about switching the pump between Quiet and Extreme in 10 second blocks to try and remove air bubbles - hasn't seemed to work.

I ran Prime95 and yet again core temps shot up to around 100c.

Read a thread online where somebody is suggesting that Prime95 and Skylake is a bad combination and a bug / incompatibilty exists where it causes the CPU to overheat - not sure if anybody agrees or has heard of this?

I've since ran Realbench and Cinebench and with the Corsair profile on Extreme after a few mins the temps were averaging around 85-90c and hit 95-98c at times.

My next steps are going to be:
  • Order some good quality thermal paste, remove the CPU block and clean / re-apply the paste
  • Buy a new cooler
I won't be overclocking of course until I can get decent stock performance cooling out of my machine.

I'm fairly out of touch with the latest tech, so can anybody please provide some suggestions for decent thermal paste and a decent cooler?

Cheers, Mike
 
Last edited:
After reading several other posts on people having similar problems I decided to check my PC for any airflow issues / check the CPU cooler block position.

I feel like a bit of an idiot for what I discovered, however I will say I'm no better off / closer to a solution right now.

I'd already recently brushed / vac'd the inside of the PC and removed the case fans, cleaned them and put them back in place - however I'd missed / forgot to take the top plate of the NXZT H440 case.

I was disgusted by what I found - the layer of dust / fluff that had built up was grim and I was confident that once it had been cleaned out it'd make a huge difference to the temps, here it is after cleaning (still not perfect):

20200522_004016.jpg


20200522_005342.jpg


I tightened the CPU block screws as they felt fairly loose (didn't over tighten) and I installed the Corsair iCUE management software.

Upon installing / launching iCUE my PC got significantly louder which, although noisy, gave me hope that maybe the fact I'd never had the Corsair drivers running may mean I've been under utilising the performance of the water cooler.

So now with a clean radiator which should be flowing, a better seated CPU cooling block and the Corsair software running, things should be looking better.

My idle temps went from 28-32c to around 40-45c after having installed the iCUE software - read several forum posts around this and other seem to have experienced it too, some pointing to the Corsair software itself consuming a disproportionate amount of system resource.

corsair-coolant-temp.png


As I write this the coolant temp is sat at 25.4c, cores around 25-30c, CPUTIN at 30c.

I read somewhere about switching the pump between Quiet and Extreme in 10 second blocks to try and remove air bubbles - hasn't seemed to work.

I ran Prime95 and yet again core temps shot up to around 100c.

Read a thread online where somebody is suggesting that Prime95 and Skylake is a bad combination and a bug / incompatibilty exists where it causes the CPU to overheat - not sure if anybody agrees or has heard of this?

I've since ran Realbench and Cinebench and with the Corsair profile on Extreme after a few mins the temps were averaging around 85-90c and hit 95-98c at times.

My next steps are going to be:
  • Order some good quality thermal paste, remove the CPU block and clean / re-apply the paste
  • Buy a new cooler
I won't be overclocking of course until I can get decent stock performance cooling out of my machine.

I'm fairly out of touch with the latest tech, so can anybody please provide some suggestions for decent thermal paste and a decent cooler?

Cheers, Mike


right off the bat you need to change the fans on that rad, instead of pulling air in from the top they should be set to exhaust, heat naturally rises so it would be best to change that, have you got any front mounted fans and are they set to intake too? if not install or change the flow so they draw in cool air, next you need to setup fan profiles as i suspect below 1000rpm at full load isn't enough airflow to dissipate the heat from the cpu, i'd say set the fans almost to the max at 65/70 degrees cpu temp and under 45 for the cpu set the fans to what you have in the screen shot (720rpm)
 
@wookiee87 - thanks for that.

There's 3 x front fans which I think are pulling air in to the case, I'll double check tonight:

front-fans.jpg


I thought I'd installed the AIO radiator fans so the direction was pushing the air from inside the case on to the radiator and out, I always struggle to work out airflow direction with these fans (I believe fans flow from front to back, where the front will have a sticker with the main Corsair logo on and the back has a sticker with some minor info).

What would be your suggestion?

  • Move the fans above the radiator to be 'pulling' the air out?
  • If the fans in their current position are pulling air in to the case, flip them around so they're pushing air over the radiator?
  • Something else
Cheers
 
ah right your front fans look good they are indeed pulling in fresh air into the pc, regarding the aio i would go with your second option "If the fans in their current position are pulling air in to the case, flip them around so they're pushing air over the radiator?", that way you aid the natural air movement, heat will always rise, i'd still have a look at custom fan profiles for the aio and set a more aggressive curve if the cpu goes above 65-70 degrees.

if you look closely at any computer fan's there should be a set of arrows which will tell you which way the fan spins and the other arrow will show the direction of airflow :)

change the aio fans so they exhaust air from inside the case out and see if cpu temps improve
 
I shall get on this tonight and flip those AIO fans around if needed, and I'll report back any difference it makes to the temps.

In the meantime I'll get that custom fan profile configured.

Cheers for the advice mate.
 
I shall get on this tonight and flip those AIO fans around if needed, and I'll report back any difference it makes to the temps.

In the meantime I'll get that custom fan profile configured.

Cheers for the advice mate.


also i forgot to mention that when your booting the pc head into the bios and set a manual cpu voltage of 1.2v, since your not overclocking there's no sense keeping it on auto as it will always push more volts than needed, (your first post shows a core voltage of 1.296 - 1.425v which is too much at stock speeds)
 
Getting somewhere!

As you pointed out @wookiee87 , the AIO radiator fans were pulling the hot air coming off the radiator back in and all over the CPU :rolleyes: Absolute noob.

I ordered some Thermal Grizzly thermal paste, cleaned the CPU and cooler block of the old paste, applied the new and reseated the cooler block, taking care not to over-tighten.

I've done an extensive clean of all fans and double checked the direction, so now I have:

  • 3 x front fans (intake)
  • 1 x rear fan (exhaust)
  • 2 x AIO radiator fans (exhaust)
The PC case is now physically cold to touch and I can feel air flowing freely through the NZXT H440's (limited) vents.

I've set up a custom Corsair profile that's now relative to the CPU temp, going to 100% fans @ 60c.

Benchmarks / temps

The first time I ran the Realbench stress test the core temps wouldn't go above 80c, averaging around 70c-75c.

The second time I ran it (after having ran Prime95), the core temps were average between 75c-80c however they very occasionally hit 82/85c and came back down.

So it seems Realbench temps are fairly stable now and definitely improved from before the fixes were implemented where even Realbench pushed temps to 90c-100c.

I ran Prime95 and core temps (again) were hitting 95c-100c. I did a second run and the temps stayed most under between 85c-90c and seemed stable. Third run back to 95c-100c :rolleyes:

Fans the right way around now (exhaust setup)
pc-closeup.jpg


PC configuration
pc-distance.jpg


H115i cooling block before cleaning
cpu-block-dirty.jpg


H115i cooling block after cleaning
cpu-block-clean.jpg


Thermal Grizzly arrived
thermal-grizzly.jpg


Cleaned i7 6700k with Thermal Grizzly pea sized drop applied
cpu-pea-sized-drop.jpg


Realbench results
stress-temps-realbench.png


Prime95 temps
stress-temps-prime95.png


Custom iCUE profile
icue-custom-profile.png


I'm now fairly confident I've done as much as I can to eliminate any obvious problems.

The CPU cooler block has been cleaned, reseated and had decent thermal paste applied, a custom H115 profile established and the PC fans are in the right configuration.

Any thoughts on the Prime95 results? I've heard several complaints about Prime95 and Skylake and how it sends a particular instruction to the CPU that's rarely used in real world applications which causes the high temps?

How are my Realbench temps looking?

Oh, and I also set the AI Overclock profile to 'Manual' in the ASUS BIOS and fixed the Vcore voltage to 1.25 (this was done before all the above stress tests).

I tried to keep my eye on the voltages in CPU-Z and didn't managed to observe anything going over 1.25, though the 'Max' reading in HWMonitor seems to think it's still reaching 1.4+

Is there some LLC lark I can play with that might help with this?

Edit: Just ran Cinebench too, 75c-80c average, didn't go over 80c.

Cheers,
Mike
 
Last edited:
That's really good news to see that temps have dropped significantly, you'll find that synthetic benchmarks like prime will put a very unrealistic work load on any cpu's (AVX instructions) and naturally 90+ degrees is expected, have you tried any games to see real world temperatures?

as above prime is pretty much the worst case for any cpu so if it can pass with no errors, then the 90+ degrees is totally normal, your realbench temps look pretty good tbh around the 70-75 degree mark is spot on.

regarding voltages i'd trust what cpu-z is saying and if you've go in and set a manual voltage via the bios it will be around what you set (1.25v), did you try and set 1.2v or was the pc crashing?

LLC is only needed under overclocking conditions, it will eliminate vdroop from the cpu under heavy load, (eg set 1.3v in the bios and under heavy load with no LLC will drop around the 1.26v mark which can crash the pc, a high LLC setting under overclocking conditions will stop the vcore dropping too far).

have a play in games and any other real world workloads and see how you get on, if temps still climb then i'd say a potential de-lid is on the cards, which is a daunting prospect but if done right can yield huge temp drops, (but de-lid is a last resort)
 
Probably the final update from me!

Although progress has been made I'm not sure I'm willing go much further until I end up with a new AIO.

I've done a fair amount of toying with the OC this morning and here's my results:

(multiplier @ voltage)

⦁ x45 @ 1.35 - temps over 95c
⦁ x44 @ 1.285 - seemingly stable, temps 85-95c
⦁ x44 @ 1.275/1.270 - temps at 85-90c, BSOD
⦁ x43 @ 1.250 - stable, temps at 80-85c

I started out ambitious going for a 4.5ghz OC, runs far too hot.

Aiming for 4.4ghz I managed to get Realbench stable for the 15 mins I ran it at, but didn't see any point running for longer at 85-95c.

I then wondered if I could achieve 4.4ghz with less voltage, so dropped to 1.270v - got a BSOD within 5 mins of Realbench.

Finally I dropped down to 43 and 1.250v - this has stayed stable so far for several hours and seems to sit around 80c, sometimes peaking at 85c.

My main take away from this is either there's still something not right with the install of the H115i, or it's just not a very good cooler.

Anything over 1.25v puts me in the 90c range, so that rules out any decent OC.

I did find an interesting thread on Reddit where a guy with the username 'BackPlateGuy' seems to specifically reply to Corsair owners having temp issues, it appears people can quite easily misfit their backplate and it makes a big difference. I compared my backplate to the image of somebody who had installed theirs wrong and it appears mine is oriented correctly. BackPlateGuy does mention you shouldn't use / need any additional backplate washers though, and I think there's at least 4 per leg on mine.

backplate.jpg


Overall I'm fairly happy to be honest with 4.3ghz, it's not the 4.5-4.8ghz others are managing, but that's life.

I started all this really because I've started playing Battlefield 5, in BF1 I get 115-130fps but BF5 I was only getting 60-75ish. I then did some settings tweaks (disabled DX12, enabled Future Frame Rendering), and it put me back up in the 80-90s. Funny thing is I don't know if this 300mhz OC has made any difference as I didn't go to any effort of recording the differences between the two setups. I'm not even sure if CPU makes much of a difference in BF5, I know upgrading from a 1070 to something like a 2080/2080ti is what I should be doing if better BF5 FPS is my objective - so pricey though!

I might treat myself to an Artic Liquid Freezer II 280 on pay day and see what happens.

Cheers for the all the guidence!
 
I might treat myself to an Artic Liquid Freezer II 280 on pay day and see what happens.

its very likely the cpu is in need of a delid, as the thermal material between the ihs and die has dried up, no way on earth should you hit those temps even at 4.3ghz with only 1.25v.

overclockers sell the delid tool by de8auer last i checked it was £29.99, and there are plenty of videos online on how to do it, plus temps will dramatically drop which will improve temps and allow high overclocks, but only after delidding, grab your self some liquid metal thermal compound too as that will have the very best heat transfer (just be very careful with the application)

if you don't like the idea of delidding then i'd update your bios and have a look at a second hand 7600k, most of those refresh chips did run pretty cool and had good overclocking potential too, your aio you have now is more than enough cooling, you'd waste money on a 280mm as i suspect your chip is the problem
 
Back
Top Bottom