CPU idle at 40c-60c watercooled.

blackcell1

blackcell1

blackcell1

Associate
Joined
30 Jan 2012
Posts
39
just finished building my watercooled loop for my pc and I'm getting idle loads on my CPU around 40 to 50 at 1% load, after stressing my cpu out it hits around a solid 60c under 100% on furmark's cpu stress test.
it wasn't what i expected when i went into building a loop as i was expecting something lower under zero load. my gpu sits at a good 23c idle and around 37c to 40c under load.

and advice on what I've done wrong if I've installed something incorrectly ? I've even ramped up all my fans to 100% without any improvements.

specs:
cpu 5900x
gpu msi 3080 trio
ram 32gb corsair 3600
res ek 360 ftl
rad ek 360 pe / 480 xe
gpu block ek 3080 trio
cpu block ek velocity am4
case phanteks enthoo pro 2

ive setup the fans to pull from the top and push to the front of the pc, with 3 fans on the back panel pulling in.

any advice would be great.
 
from information on the forum idle temps are ok its not intel so don't expect the same temps. 60c at 100% is a crap out use cinebench 20 to get some higher temps give your water cooling a proper test
 
For a 105W TDP cpu sub 60 under load is pretty great which suggests it's not a mounting/block problem but just to be sure, the velocity has an in/out port. Are they the right way round?

Otherwise, check your voltages when the cpu is idle. You don't mention if it's OC'd at all but if your voltage is locked at something quite high that would be pushing temps unreasonably high when idling.

How long are you stressing for? In a loop with that much rad you might not see the water temps reaching equilibrium for 30+ minutes depending on fan speeds etc.
 
Im no expert im hoping to build my 1st water cooled build when i can get all the parts but would u not be better pulling the air through the front and out/push through the top as this is where warm air is naturally gonna want to go? Also the warm air going out through the front might be pulled into the top intake?
 
FrostFox said:
Im no expert im hoping to build my 1st water cooled build when i can get all the parts but would u not be better pulling the air through the front and out/push through the top as this is where warm air is naturally gonna want to go? Also the warm air going out through the front might be pulled into the top intake?
Click to expand...
the numbers for this sort of thing are greatly exaggerated. you could have a fan running at less than 100 rpm and it would easily over write the natural heat rise.
 
My all AMD all water-cooled X570/5900X/6800XT - MB/CPU/GPU/Chipset idle temps are 32/33c once the waters up to ambient room temp of around 26c.

Full load Cinebench is 60c and gaming on CodCW can see the CPU get up to 70-75c - GPU 50c Hot spot 60-65c. Thats the highest it will go with the water temp sitting at about 35/36c all fans silent at around 600rpm.

1 x 420 1 x 280 monster rad. :)
 
I'll say first off that 60°C is a remarkably low load temperature for any typical CPU, especially a high-end one like a 5900X. On air coolers you'd expect anything from 70 to 90 as a temperature ceiling. And 90 is still safe, although ideally not 24/7. It should be perfectly fine to run a modern CPU at 70°C continuously though.

Onto the temperatures under water. CPUs are generally slightly warmer than GPUs for a given water temperature. This comes down to a few factors:

- The CPU die (silicon surface) is usually further away from the water and separated by layers of solder, heatspreader, and sometimes more thermal paste. This distance and extra transfer layers means heat cannot escape as fast to the water.

- The CPU silicon internally tends to have a larger temperature delta due to voltage. So whatever temperature the surface of the silicon is, at the point where the Vcore is applied, there will be a temperature increase. This is not the same as the power/heat output - it's purely a "step" difference in temperature. Test it: change the CPU core voltage a little bit up or down but keep the clock speed the same. You should see the idle voltage step up or down a degree or two.

Typically CPUs at stock have a bigger temperature delta from the water than GPUs. This difference gets bigger as overclocking is applied - for instance my i5-4690K at stock was around 5° warmer than my GPU. When overclocked, my GPU caps out at 53°C but my CPU can spike to 65 or 67 under heavy load.

Then main point is - it doesn't sound like a CPU block mounting problem or a water flow/cooling problem. CPUs get hotter than GPUs despite using less power, it's the nature of their silicon layout. A GPU uses twice the power in 4x the surface area so its power density is smaller. 60°C maximum is a fantastic result. Most computers sit at 60 all day long and go to 70-80 when idle, remember some people are using the same CPU as you in a small box with a generic air cooler, or even in a laptop.
 
Personally i think the temps for both CPU and GPU are fine. I run a 5800x and 6800 xt on a full loop, and my temps are slightly higher on the load (but i guess it can't be a fair comparison as i heard the 5800 gets a lot toaster than the rest of the AMD range).

Shouldn't matter much in terms of loop configuration but after a bit of testing in my setup, noticed CPU temps creep higher (sub 5c delta) if running pump > RAD > GPU > CPU > RAD > PUMP than it did when it was originally pump > RAD > CPU > GPU > RAD > PUMP, and CPU can reach to 80c load when running OCCT small data set. Normally it idles sub 30-40c with random spikes from 40-50c
 
60C won't kill your CPU, don't worry.
Under load, under normal use, the CPU won't get as hot as when running synthetic tests.
But if you're running tasks/apps that do push the CPU, then the temperature should be similar to the ones you're getting.
One more thing I could say about CPU and GPU is that the GPU's temperature increases linearly, if that makes sense. Not spikes here and there, like CPU.
During tests, a CPU could go from 40 to 60, 70C, and then stay at 70C, for example. The water would take much longer to saturate the cooling capacity than an air cooler would, but the air cooler reacts faster to spikes. More noticeable on AIO, thought, as a custom loop, proper big boy pump, the fluid is moving much faster than an AIO ever will.
A slightly offset voltage helps for the 3000 series, but for the 5000 series seems to be a different approach to get better results
 
You must log in or register to reply here.
Back
Top Bottom