Jump to content
Corsair Community

NEW Build high temp and low


Pandas213
 Share

Recommended Posts

So I built a new pc and I also got Corsair H100i v2 Platinum SE.

Is this a normal temp for this Ryzen 3800x @ 4.5GHz 1.35v?

I set the fans as intake and I've tried several thermal paste, including getting another 3800x to see if it solved. It did not.

 

Once I pressed start in aida64 with idle temps of 48C. It would instantly shoot up to 90C for 1-2 seconds and stabilises at 86C for 30 mins but the cooling temp seems to be 32C. any possibilities on what's wrong? should I ask for replacement? I can still retun the item.

 

57.PNG

 

 

Also I tested in stock clock and it seems to be around 40C idle without anything running or open.

light browsing gives me around 50C temp. and AIDA64 caps it at around 80C

 

 

My computer specs are:

ASUS Strix B550-F Motherboard

ASUS Dual Evo RTX 2070S GPU

Corsair RM750x PSU

G.Skilll Trident Z Neo 32GB CL16 3600Mhz RAM

Corsair 220T Case

 

Room temp is 23C

 

This is also the picture of how I set up my H100i

Snapchat-2069360930.jpg

Edited by Pandas213
Link to comment
Share on other sites

You are looking at the typical result of voltage level limitations. Large watt power devices like GPUs or high core count CPUs that use 250W+ benefit from having a large surface area cooling system. You have a definite level wattage you need to get rid of in order to prevent additional heat rise for hardware components.

 

This is not the issue with Ryzen processors. You have a fairly low TDP and the cooler's settings won't make a lot of difference. It also magnifies where the limitation occurs - at the CPU. All of use are going to be limited by voltage, but this is very obvious on the 7nm size where overall watts are low (thus little coolant change), but CPU temps will still be high because you are running 1.35v+ through it. The instant load temperature of the CPU is going to be directly related to physical construction and voltage levels. The cooler end of it never comes in when you are 90C in the first second. The cooler keeps you from getting warmer than that over the course of time, but the CPU temp is created where the pins meet the metal. It has to go through the CPU.

 

Your higher idle temps are a common lament on Ryzen. It seems to be reactive to most everything and keeps the voltage levels up. If yous see 45C idle with a 0.65v Vcore, tht is more concerning. It is more likely the Vcore is at elevated levels even during idle. However, that doesn't automatically mean it is using more power and the control of that has moved down to the CPU level below observation.

 

Can't tell you what the going CPU temp is for AIDA for your CPU and 1.35v. Ryzen don't have a lot of overclocking room and this is the reason why. If the result of one test is uncomfortable, try a different one to get more data. I usually suggest the CPU-Z stress test in the Bench Tab. It is linear and relatively mild. It makes it easier to assess thermal capacity and function vs a pure stability test like AIDA. It is linear so the CPU temps should sit on their initial rise and then you can observe coolant temp changes and how they correspond to the CPU temp (+1 additive).

Link to comment
Share on other sites

I get that but I ran AIDA64 for 30 Minutes with constant 80C+ and cooland still hovers around 32-33C. I felt like it isnt transferring heat. I tried reinstalling with different thermal paste and stuff. Even at stock settings clock of ryzen, it still hovers around 80C. No better than the stock cooler it comes with. So I was assuming that there's some form of blockage or air inside liquid cooling. But I'm not sure entirely.

 

My brother does have a 10600K and H115i while running AIDA64 on his computer for 5 min, the coolant temperature would rise to 38C within 5 mins. My H100i has been running for 30 mins at constant 80C+ and the coolant temperature is still around 33C.

 

 

This one is taken from cold boot just now (10 Minutes after turning on my PC). No stress test done or any games opened yet. This is also stock settings so no overclock done. my coolant temperature runs around 32C+ without any stress test prior. So I'm assuming there's a problem with heat transfer to the H100i?

I used the correct socket for the H100i so there's no problem there.

Room temp is around 23C

 

WRhP2Hu.png

The only program was opened is HWMonitor, Discord, Browser, (so I can check this forum) and the iCUE. Nothing else in the taskbar.

 

 

So yes basically what I was saying, the coolant temperature is too stable and it isn't even going up even for 30 min 100% cpu usage and processor running hot 80C+.

 

For overclocking, I've seen people keep theirs under 80C without using the expensive liquid cooling or water loops with the same clock and nearly the same voltage. Mine just runs hot directly.

Edited by Pandas213
Link to comment
Share on other sites

Remember coolant change is watts in vs watts out. If you put 150W in, but take 150W out, the net coolant change is zero. That is the criticism of the 10K series Intel CPUs. They use a lot of power and that 10600K may use another +80W above what your CPU does for AIDA, especially if the AMD is at stock settings. There a lot of advantages to the smaller die architecture, like a reduction in wattage and less signal distance. A reduction in CPU temp is not one of them.

 

There is usually going to be some rise in coolant temp because the sensor is post CPU heat pick up and before the radiator release point. However, if you are at 33C in a 23C room, you are showing some coolant temp rise. +10C over ambient is significant for a top intake installation. I don't know if you have a dust filter up there or there are environmental factors, or the air above the case is not really 23C.

 

When you have a contact issue between CPU and cold plate, most everything is untenable. You can't browse the web, run programs, and your AIDA would hit the 100C shutdown point instantly. It would be unusable. Since you have already re-done the paste job and there isn't any evidence of something wrong there, I wouldn't over-analyze that aspect. There is little chance you got the one piece of copper in the world that does not conduct energy.

 

Total system behavior can take a while to shake out. There are a lot of factors involved. However, in the short term the clear point of interest is the CPU temperature. Cooling system changes, whether its fans, fan speed, pump speed, direction, all are small factor changes -- 1 or 2C here or there. What I do see from the above is your VID request from the CPU is between 1.43-1.47V. I cannot see the actual Vcore being applied, but it seems highly likely you are seeing higher than expected CPU temps because you have a higher than expected Vcore. That +60C difference between coolant and CPU temp is high. If it's not the contact, it is going to be the voltage.

Link to comment
Share on other sites

the 10600k will evacuate heat a bit better because of die/IHS differences, so your brother will probably see lower core temps and higher water temps under load.

 

It's almost impossible to compare between the two. Besides CPU differences, you may not have the same room temp (unless you live in the same room :p) fans may be diffrent, fan speeds, case air flow etc..

 

32° coolant with 23° ambient sounds about right.

 

Regarding the coolant temp that is constant, maybe there's an explanation here (people who kno AIOs better may correct me) :

 

At idle, your CPU pumps like 45W constantly, and the fans run at their minimum around 800 rpm, so, water temp will naturally rise.

 

Under load, you are at 120W with fans running at 1900rpm, so you push more heat into the water but it gets dissipated a lot faster too.

 

I feel, what is surprising is that your CPU gets loaded this much in idle.

Just as a comparison (apples and oranges..) my 10900k hovers around 5W in idle. Your brother may see that value drop even lower.

If i load a single core cinebench (~30W) i see core temps like yours in the 40 - 50°C.

 

Maybe chase programs running in the background, and you'll see your idle temps and water temps actually go to a more realistic base line? just an idea :)

Link to comment
Share on other sites

Ohh yeah I tried to close everything it seems like this 3800x never went below 20W at start. But yeah it was weird that the temperature of the coolant was stable compared to Intel 10600K. The fans for H115i is ML 140 Pro RGB while H110i Platinum SE is LL120.

 

Just I never felt any heat at all coming out of my radiator with that 30 min stress test at over 80C.

Link to comment
Share on other sites

One of the more difficult things to assess is what happens when you have the radiator the rest of the cooling system in a warm environment (inside the case) and there is a clear difference between the interior and exterior temperatures. Even if you are blowing cooler outside air in, an object in a 33C environment will eventually become 33C as well. It's possible you are not seeing more coolant change because the environment is keeping the entire cooling system warmer than it otherwise would be. It is the limiting factor and not the watts off the CPU. If you stick you hand into a 33C box, the 33C air temp coming off the radiator isn't going to feel warmer or colder. Just the same.

 

However, we are talking about a couple of degrees here or there and not the -15C or so you are looking for. That can only come from voltage control. I can't see your actual Vcore in the HWMon shots. It is in the motherboard section at the top. Swapping the cooler to a 10700K seems like a hard way to test, especially since environment plays such a large part in the final numbers. Checking your Vcore is lot easier and most motherboards require tweaking to be able to successfully run synthetic stress tests at expected temperature levels.

Link to comment
Share on other sites

 Share

×
×
  • Create New...