Intel i9-10850 hot

Petrolhead · 26 Oct 2020 at 21:00

I recently built a pc using an intel i9-10850. All was fine, idle about 32deg, room temp 20deg. On load ie Cinebench it was 88deg

The cpu is cooled with a Fractal Design Celcius 360

today i decided to Replace the fans and do some general tidying so removed the block from the cpu.

once done fired the pc back up, idle 50deg, hmm thats not right so i removed the block a reseated it.

refired the pc, idle 40deg but as soon as i ran cinebench it went straight to 100deg! and crashed

so i removed the block again and took a couple of photos for you to look at. One thing idid notice, both times i removed the hold down screws there seemed to be not much resistance/spring.

its as though the cooler block is not seating properly

any thoughts

edit: i should ad the the cpu is running 5ghz all cores

Petrolhead · 26 Oct 2020 at 22:39

Update

i have a spare Corsair H110 Gt so rigged that up, very crude, just clagged it onto cpu using what paster was on the gpu and the rad and one fan sitting next to the case.

idle 28deg and 88deg when running Cinebench

so is the Fractal buggered?

qweezy · 26 Oct 2020 at 23:44

Those temperatures are still high if you are stock.

Have you checked what your voltage is? It seems the majority of motherboards overvolt the 10 series chips when left on auto. Also is multicore enhancement enabled?

Petrolhead · 27 Oct 2020 at 00:14

qweezy said:
Those temperatures are still high if you are stock.

Have you checked what your voltage is? It seems the majority of motherboards overvolt the 10 series chips when left on auto. Also is multicore enhancement enabled?

Should have said its running all cores 5ghz.

re your other question would need to check but 1.29v comes to mind. Re multicore enhancement, cant remember, i assume i would need to check the bios?

turbot1984 · 27 Oct 2020 at 00:43

I have got 5ghz at 1.240v, not checked temps with Cinebench, but around 70c with cpu-z stress test using Corsair 100X. Temps do ramp up quickly when you get near 1.3v so I backed off going 5.1 all core.

marms · 27 Oct 2020 at 06:17

Are you using new thermal paste? You need to clean off the old stuff and use new paste every time you remove the cooler.

Petrolhead · 27 Oct 2020 at 07:38

marms said:
Are you using new thermal paste? You need to clean off the old stuff and use new paste every time you remove the cooler.

Yes except when testing my old cooler, see post 2

qweezy · 27 Oct 2020 at 08:07

Petrolhead said:
Should have said its running all cores 5ghz.

re your other question would need to check but 1.29v comes to mind. Re multicore enhancement, cant remember, i assume i would need to check the bios?

It would definitely be worth double checking the voltage. Also your LLC as any of these settings set to auto tend to push way too much voltage than required.

Multicore enhancement is also worth disabling. It will allow all cores to use the max turbo profile, but in reality this just results in significantly more heat.

Apart from that to isolate if there is a cooler problem I would revert the overclock back to stock settings and see how the coolers perform. It could be that after checking the above, neither can quite handle a 5ghz all core. Really your max load temperatures with realbench or similar need to be below 85c.

Petrolhead · 27 Oct 2020 at 08:27

qweezy said:
It would definitely be worth double checking the voltage. Also your LLC as any of these settings set to auto tend to push way too much voltage than required.

Multicore enhancement is also worth disabling. It will allow all cores to use the max turbo profile, but in reality this just results in significantly more heat.

Apart from that to isolate if there is a cooler problem I would revert the overclock back to stock settings and see how the coolers perform. It could be that after checking the above, neither can quite handle a 5ghz all core. Really your max load temperatures with realbench or similar need to be below 85c.

Thanks

i would agree apart from the fact that its been running the OC for a month with not problems plus the other cooler runs fine

Slinky69 · 27 Oct 2020 at 09:07

If all you did was take it off and put it back, it does sound like the mounting hardware isn't getting it close enough to the CPU? I presume the cooler's fan(s) are actually spinning?

Petrolhead · 27 Oct 2020 at 09:17

Yes

i seem to have sorted it, at the 5ghz oc its idling at 33deg and when running Cinebench 88deg. I cleaned everything and reseated the cooler but turned it 90deg

to be honest i am new to overclocking and got my setting from a youtube video by Boosted Media and another site, which i cant remember.

maybe i should start a new thread and ask there about overclocking as according the HWinfo the vcore is running at 1.32v

Edit: i have started a new thread https://www.overclockers.co.uk/forums/threads/overclocking-intel-i9-10900k.18904008/

Acolyte · 27 Oct 2020 at 13:15

What orientation is the AIO in your case?

Petrolhead · 27 Oct 2020 at 14:23

Top, exhausting out

Petrolhead · 30 Oct 2020 at 07:24

Well i thought it was sorted but it started playing up again, so i put the bios back to default however 40-45deg idle and when running Cinebenck 100deg!

I remembered that on the Fractal AIO you can switch between Auto and PWM. It was on Auto. So I switched to PWM. Idle 30 and when running Cinebench 60.

so thereis obviously somit up with Auto mode.

going to run Cinebench on a loop for a while and see how it goes.

i also think will just leave the CPU std

Davidus29 · 1 Nov 2020 at 09:37

Petrolhead said:
Well i thought it was sorted but it started playing up again, so i put the bios back to default however 40-45deg idle and when running Cinebenck 100deg!

I remembered that on the Fractal AIO you can switch between Auto and PWM. It was on Auto. So I switched to PWM. Idle 30 and when running Cinebench 60.

so thereis obviously somit up with Auto mode.

going to run Cinebench on a loop for a while and see how it goes.

i also think will just leave the CPU std

Maybe something bad happened to the AIO Pump? I had similiar problem and it turned up as broken pump.

Destination · 1 Nov 2020 at 09:42

Depending on angulation it is possible some air bubbles trapped in the block, and with the rotation it has allowed them to move onwards, thus freely up more cooling capacity.
It might have cleared itself over time and improved temps.

Petrolhead · 1 Nov 2020 at 10:00

Davidus29 said:
Maybe something bad happened to the AIO Pump? I had similiar problem and it turned up as broken pump.

Possible, only bought it 1st sept

Davidus29 · 1 Nov 2020 at 10:15

Petrolhead said:
Possible, only bought it 1st sept

You can check that by touching one of the tubes that takes the hot liquid from CPU to radiator. If it's very warm close to the pump and then suddenly becomes cold that would suggest the pump died.

EsaT · 1 Nov 2020 at 10:50

Petrolhead said:
i seem to have sorted it, at the 5ghz oc its idling at 33deg and when running Cinebench 88deg. I cleaned everything and reseated the cooler but turned it 90deg

Despite of marketing numbers high core count Intels are in reality 200+W heaters under full load:
https://cdn.mos.cms.futurecdn.net/ayKpWuEAsSLrXdwwXXxZjW-1814-80.png
And that's simply too much heat to keep CPU cooler than that for those slim radiators.

Despite of BS hype water isn't some thermal black hole making energy disappear: In steady state operation all that heat must be dissipated into air just like in heatpipe coolers.
And that needs big surface area.

Originally DIY people used car radiators, or cabin heater cores with genuinely extreme amount of heat dissipating power.
And when designed for PC cooling radiators originally appeared those used to be like 5cm thick.
But when waterpipe coolers become fashion pushed for everyone and their pet dog, people didn't want to that big things into their case.
So marketers shrinked radiators to slim ones.
At the same time shrinking also heat dissipating power.

Davidus29 said:
You can check that by touching one of the tubes that takes the hot liquid from CPU to radiator. If it's very warm close to the pump and then suddenly becomes cold that would suggest the pump died.

When pump dies you know it instantly: CPU is cooking up from alive at idle!
Because without active coolant circulation there's zero cooling...
Just like what happened in Chernobyl.

Petrolhead · 1 Nov 2020 at 13:11

EsaT said:
Originally DIY people used car radiators, or cabin heater cores with genuinely extreme amount of heat dissipating power.
And when designed for PC cooling radiators originally appeared those used to be like 5cm thick.
But when waterpipe coolers become fashion pushed for everyone and their pet dog, people didn't want to that big things into their case.
So marketers shrinked radiators to slim ones.
At the same time shrinking also heat dissipating power.

interesting and valid

i have just finished a water loop for the GPU (RTX 3090) using a EK PE 360 rad which is keeping the gpu nice and cool.

i have been wondering about, what if i added the cpu to the loop. Another 360 rad maybe thicker?