[SOLVED] New build 10700k hitting 100c under load

[OoN41H4NoO]

I know the title suggests I may have done something wrong with this build as 100c is not OK by any stretch especially stock, but I'm not sure I actually believe this reading is true. I completed the build yesterday and begun benchmarking today in prep for some overclocking..

All parts currently stock fresh out of the box
I7 10700k
Arctic liquid freezer II 280
MSI z490 Tomahawk
MSI 2080 Super
16GB Corsair vengeance Pro 3200
500GB M.2
Sahara P35
be Quiet! 700w 80 Gold
4TB HDD for games
BIOS updated to latest firmware, Windows 10 pro up to date, drivers etc. etc. etc.

Thermal paste I used the pea size method, maybe closer to a baked bean
Ambient room temp is quite high, its a hot evening
Idle CPU temp ~45c
Idle CPU frequency 4700MHz (Is this a bit high? I'm sure stock this CPU should idle around 3.8GHz before turbo)
CPU core voltage is fluctuating round 1 and 1.16v
If I run Heaven benchmark @ 1440 Ultra my CPU and GPU temps sit around 65 to 70c which is fine even after half hour of running.

Now comes the weird bit, if I run Cinebench all 8 cores immediately (and I mean immediately, no steady increase in temps) hit 98 to 100c and the CPU throttles down to about 4000MHz.
The same happens if I run CPU-Z's stress test
In dragon centre the core temp shows 100c while the socket temp shows around 50c

I load up PUBG, lobby CPU hovers around 85c then in game hits 100c, again this happens instantly.
I've also tried kingdom come deliverance @ 1440 with the same recorded temps.

I have a few tools to monitor the system
Core temp 1.16
HW Monitor
MSI Dragon centre
GPU-Z

I have set the CPU fan (rad) power curves quite aggressive so as I have 100% by about 70c just for testing, the AIO is connected to the CPU_Fan1 4 pin connector, only 1 connector for both the pump and the rad fans, they are set to PWM and do increase with the temps.

I have opened the case while these temps are being reported and other than the GPU everything is cool to the touch, the AIO rad is cool, the liquid tubes are warm ish, the pump block is cool and the fan is spinning (the pump has a very small fan).

I really don't know if I should believe these temps, they seem to happen too fast and aggressively, and the fact heaven runs within what I would consider acceptable temps (between 65 and 75c for both CPU and GPU) I'm doubting what I see, but at the same time need to acknowledge these temps are alarming....

Any suggestions or ideas would be appreciated

 

[Phaaze88]

Unigine Heaven is a GPU BENCHMARK. Why would it stress the cpu?

Idle CPU frequency 4700MHz (Is this a bit high? I'm sure stock this CPU should idle around 3.8GHz before turbo)

So it's just sitting at 4.7ghz, and not dynamically adjusting frequency? That's what the cpu should be doing if it were truly at stock.
It sounds like you need to disable whatever performance tweaks the Msi implemented...

Sahara P35... you're not choking off the radiator's fans, are you?

I have set the CPU fan (rad) power curves quite aggressive so as I have 100% by about 70c just for testing, the AIO is connected to the CPU_Fan1 4 pin connector, only 1 connector for both the pump and the rad fans, they are set to PWM and do increase with the temps.

Plug the radiator fans into the motherboard - if there's any headers left. If not, get a Y-splitter.
Run the pump at 100% anyway through bios.

the pump has a very small fan

That's not the pump, that's the VRM cooling fan.

 

[OoN41H4NoO]

Unigine Heaven is a GPU BENCHMARK. Why would it stress the cpu?

I know its not meant for the CPU but its one of the tools I'm using so considered it was worth a mention and shows the GPU is working within expected temps

So it's just sitting at 4.7ghz, and not dynamically adjusting frequency? That's what the cpu should be doing if it were truly at stock.
It sounds like you need to disable whatever performance tweaks the Msi implemented...

Sahara P35... you're not choking off the radiator's fans, are you?

Yes just sitting at 4.7ghz, task manager shows 2% CPU load.... Ill remove the tool to see if its automatically applying an OC, it has OC profiles but I wouldn't use them regardless. Sadly its the only way I can control the RGB for the MSI GPU 😪

The rad fans seem fine, I can feel cool air passing though the radiator and out the top of the case, the radiator is also cool to the touch (Its my first AIO so not sure if they do get warm)

Plug the radiator fans into the motherboard - if there's any headers left. If not, get a Y-splitter.
Run the pump at 100% anyway through bios.

I can give this a go, but as mentioned before they seem to be running fine, from the BIOS i am able to manually control fan speed so PWM is working.

That's not the pump, that's the VRM cooling fan.

Ahh ok thanks, I just put this to show the unit has power so I would assume the pump is also working. (Assumptions are the root of all major F up's)

Ill uninstall the MSI software now and see what happens.

 

[Phaaze88]

Yes just sitting at 4.7ghz, task manager shows 2% CPU load.... Ill remove the tool to see if its automatically applying an OC, it has OC profiles but I wouldn't use them regardless. Sadly its the only way I can control the RGB for the MSI GPU 😪

Yeah, that's not normal.
Is there no way to disable Game Mode, or whatever the overclock profile is called, through Dragon Center?

 

[OoN41H4NoO]

Yeah, that's not normal.
Is there no way to disable Game Mode, or whatever the overclock profile is called, through Dragon Center?

I can disable game mode but that really only appears to effect GPU, I've just read up that the dragon software apparently locks CPU frequency with no way stop it...

I've uninstalled it and reset my BIOS but get this, its now applying up to a x50 multiplier bouncing between as low as 800 and as high as 5000MHz... Voltage is fluctuating between as low as 600v and up to 1.26v, and CPU power consumption also darting from 8w up to 20w all while idle.... Just ran PUBG for 10 minutes and temps maxed out between 94c and 97c across all cores yet all parts are cool to the touch...

all readings from Core Temp - Im not sure what to make of this

 

[Phaaze88]

I've uninstalled it and reset my BIOS but get this, its now applying up to a x50 multiplier bouncing between as low as 800 and as high as 5000MHz... Voltage is fluctuating between as low as 600v and up to 1.26v, and CPU power consumption also darting from 8w up to 20w all while idle.... Just ran PUBG for 10 minutes and temps maxed out between 94c and 97c across all cores yet all parts are cool to the touch...

all readings from Core Temp - Im not sure what to make of this

Now THAT behavior, is actually normal:
The temps aren't great though, so we need to get to the bottom of "why so high?"

These coolers are shipped with some plastic strip or cover to protect the copper base plate and pre-applied thermal paste(if it was present). Did you remove this?

Radiator placement in the Sahara P35.
Front intake: Are the radiator's fans positioned behind it to pull air through it, and the tubing entering from the bottom?
The reason: More 'breathing room' for intake as opposed to being almost flush with the glass panel, and the air pockets in the CLC get trapped in the radiator instead of flowing back and forth between the pump(which should stay wet) and rad.

Top exhaust: Are the fans pushing air up and out of the chassis?

 

[OoN41H4NoO]

Now THAT behavior, is actually normal:
The temps aren't great though, so we need to get to the bottom of "why so high?"

These coolers are shipped with some plastic strip or cover to protect the copper base plate and pre-applied thermal paste(if it was present). Did you remove this?

Radiator placement in the Sahara P35.
Front intake: Are the radiator's fans positioned behind it to pull air through it, and the tubing entering from the bottom?
The reason: More 'breathing room' for intake as opposed to being almost flush with the glass panel, and the air pockets in the CLC get trapped in the radiator instead of flowing back and forth between the pump(which should stay wet) and rad.

Top exhaust: Are the fans pushing air up and out of the chassis?

I can understand the multiplier and voltage changing as part of the boost function to adjust for workload but why would it be so erratic under no load at all.

so 100000% removed the plastic strip, that was something I was conscious of at install. I have it in my hand now and it says WARNING!

so the AIO is on top, however they are pulling air in as i figured it would be colder outside and better for CPU temps (maybe a little detrimental to GPU). I can feel cold air passing passing though from the radiator and into the case, as I've said the radiator itself is cold to the touch... I can flip the fans and test....

What gets me is the temps jump directly from ~45/50c to 98/100c instantly, and as soon as I stop the stress test the are instantly back to 55c, there is no heating up or cooling down process.

 

[Phaaze88]

I can understand the multiplier and voltage changing as part of the boost function to adjust for workload but why would it be so erratic under no load at all.

Because there's never truly nothing going on, even just sitting on the desktop. The only true idle state is the powered off state.

so the AIO is on top, however they are pulling air in as i figured it would be colder outside and better for CPU temps (maybe a little detrimental to GPU). I can feel cold air passing passing though from the radiator and into the case, as I've said the radiator itself is cold to the touch... I can flip the fans and test....

Top intake, huh? This would actually be ok if you didn't have a 300-ish watt gpu exhausting most of its waste heat inside the chassis...
In situations where you've got both the cpu and gpu under load, where's the gpu exhaust supposed to go? Sure, some of the heat from the back end should, well, go out the back, but what about towards the front of the card? Running into the top intake air and causing a traffic jam of sorts...
Flip the fans to push out.

What gets me is the temps jump directly from ~45/50c to 98/100c instantly, and as soon as I stop the stress test the are instantly back to 55c, there is no heating up or cooling down process.

-liquid not moving, but you've already checked that, and by hand too. The cpu block would've been hot instead of warm to the touch if there were a flow issue.
-pump not running at 100%. The weakness of AIO/CLC coolers is the base plate and pump speed. The pump speed pretty much needs to be maxed out to deal with high heat loads. You've already confirmed that the pump is running 100%...
-the base plate isn't making good contact with the cpu IHS. This would make sense if the entire liquid cooler is cold to the touch.

 

[Zerk2012]

Sounds like almost dead cooling pump to me or not installed properly.

Your words.

the AIO rad is cool, the liquid tubes are warm ish, the pump block is cool

That to me means it's not doing crap! Pump block is cool if it was making good contact if should heat up since it is transferring heat.

 

[Karadjgne]

View: https://youtu.be/qQ_AETO7Fn4


Dragon Center just makes this even worse, it applies an artificial Full Power Plan, to whatever you have set currently (kinda like software enabled MCE). It works out fine for lower power cpus, shows a decent amount of boost over Intel recommended guidelines for the bios, but in the more powerful and voltage/cooling necessary performance cpus like the 10700k, it can get out of hand in a hurry.

With liquid cooling there's a difference in the way they work from aircoolers. Aircoolers are direct cooling by metal chunk, so wattage output from the cpu is transfered very fast. Cpu temps are closely linked to the ability of the aircooler to maintain transfer rates. As the rate slows or is hindered by lack of dissipative area, cpu temps go up.

Liquids are different. The have a far larger absorbtion capacity than metal. Wattage is transfered to the liquid but that transfer has less affect to the cpu, the cpu is governed more by its workload.

Cpu spike temps are unavoidable, it takes a second for the wattage to travel from the core, through the silicon, through the Tim, through the IHS, through the paste, through the baseplate to be absorbed by the coolant. So instantaneous spikes (the temp is registered every 256milliseconds) are going to happen. That's where you see the jumps from 40's to 90's in your pc. That's workload, voltage, power limit settings at work. Those spikes are happening before any cooler would even have a chance at controlling.

Longer term temps are different. Your cpu should (by Intel recommendations) have 50ish seconds (Tau) at max boost (PL2) then drop to TDP (PL1) wattage for the duration, happening (I believe) no more than once every 5 seconds. Motherboard 'stock' bios settings are changing that, you'll get an indeterminate amount of PL2, max power. Which isn't fun for a cooler. PL2 for a 10700k is supposed to be 229w, but by changing the power limits for 'stock' settings, it's very possible to far exceed that amount, over 250w at stock settings.

Add high room ambient, a 280mm AIO forcing gpu temps and cpu temps back into the case, poor airflow management basically creating a hotbox and stifulling the aio airflow, and yes, I can see exaggerated temps from the cpu. It's highly likely that you'll need to actually OC and take manual control of the bios settings to drop the temps.

But as suggested, I'd also check the mounting of the cooler, it's not unheard of for the pump block to not have enough pressure on the cpu to facilitate good thermal transfer.

 

[OoN41H4NoO]

OK sorry to bump this back up but it looks like I've fixed the issue, I literally went back to square one, removed AIO, CPU, rad fans, and re assembled with fresh paste, flashed the BIOS then reinstalled windows from USB just to make sure the MSI software hadn't left anything behind when I uninstalled.

I now have idle temps in the high 30c's after being on all day, and after an hour of PUBG at 1440 ultra I was seeing ~140fps with both CPU and GPU temps in the high 60c low 70c range so imp happy. Prime 95 pushed CPU temps into the mid 80s but I doubt gaming will ever put it under that much load.

I haven't OCd for now as it seems happy to boost to 4.8 stock with around 1.3v without throttling. This weekend I may go for the 5GHz mark and see if I can drop the voltage down for hopefully even less heat..

Thanks everyone one chipped in, wish I could pinpoint the cause for you but I'm just happy its running as I would expect.

 

[vMax]

I have the same..ish setup and the temps are now great on your side...on the overclocking side, the 10700K is very good and with a bit of work you should be able to get 5GHz all core at or below 1.3v....I eventually managed 5.1GHz at 1.290v and if you go over to overclocker.net you will find a great post on settings for the 10700K....