Jump to content
Corsair Community

Corsair CPU Water Block XC8 RGB PRO JTC High temp.


Recommended Posts

Hello. I get very high temp in CB23 If my Fans and Pomp set in icue Hydro x fans and Hydro x dx5 pump i can hit up to 100*, I change by my self Fan speed and pomp up to 2800 rmp i get max 95*

I have impose termopaste with spatule and i mount cpu water block.

"Secure the XC7 RGB PRO onto the motherboard by tightening
the screws by hand until you reach the end of the thread.
Tighten the screws evenly, preferably in a cross-pattern
(Figure 2)."

 

My setup it's:

- Asus ROG Maximus Z690 Hero
- 12900K
- Corsair Dominator DDR5, 6200 MHz, CL36-39-39-76 2x16gb
- Strix 3090 OC
- Corsair HX 1000i

- Corsair 7000x RGB

- Thermal Grizzly Kryonaut 1g

- Custom loop
2x Corsair Hydro X Series XR5 420mm Radiator
Corsair Hydro X XD5 RGB Pump/Reservoir Combo Black
Corsair Hydro X Series XG7 RGB 30-SERIES STRIX
Corsair CPU Water Block XC8 RGB PRO JTC
11x Corsair SP 120 og 140mm RGB Elite Fans

Coolant XL5

Like everybody see i have connected DX5 -> GX7->XC8-> Topp Radiator-> Front Radiator-> DX5

Should i change that to:

DX5 -> GX7-> Topp Radiator->XC8-> Front Radiator-> DX5

Can i keep coolant from my system and fil one more time? i have use maybe 3 weeks.

single multi -055v 2800rmp pomp.jpg

flow direction.jpg

Link to comment
Share on other sites

Posted (edited)

I suggest trying a different cpu stress test to compare before making any physical changes. Cinebench and other programs using complex instructions may cause a dramatically higher temp than some other programs. Try the stress test in the “bench tab” of cpu-z. This is linear and mild. 
 

If you have a contact issue between cpu and cold plate, almost every activity will cause alarmingly high temps. Opening a browser might cause spikes into the 70s. Any actual sustained load will be near the safety shutdown and temps will rise and fall dramatically with any voltage change. 
 

On the other hand if you run cpu-z stress test and it rides at 75C flat, then you know your cpu is packing on the voltage during R23. Not uncommon and this must be fine tuned at the bios level. It is not something you can address with loop order or fan speed. Loop issues will show continually increasing coolant temp, but I suspect your coolant temp is not increasing much while the cpu instantly goes to 100C. 
 

Be aware it is possible to wreck your cpu bench mark temps by starting in a low flow state. At idle desktop with little cpu and gpu wattage, a slow flow rate doesn’t matter. But if you initiate a max cpu or gpu load test at a minimal flow rate, the coolant may linger in the block too long and cause a marked rise in cpu temp. Some of the Hydro X presets are too slow for this purpose and it may take too long for the pump to speed up, like with more gradual building loads like games. I would set the pump to 100% prior to starting the run. You can then back off the speed at 20 second intervals. You might go up 1C (not a meaningful change) but eventually you will drop to a pump speed where the cpu temp immediately goes up 5-9C. That is floor or minimum effective flow rate and you should keep the pump at that speed or higher. 

Edited by c-attack
Link to comment
Share on other sites

Posted (edited)

Hello thx for tips. 

 

I have test in cpu z and aida64 stability. It's very nice temp but in 3000 rpm pump. I have temp sensor in pump XD5 and show around 30* and 1 temp sensor in case, where best to mount that sensor ?

How change rmp DX5 pump ? Coolant temp ? temp sensor from pump ? Can adjust sensor ?

test-pump-300rmp.jpg

test pump 300rmp.jpg

Edited by kampers
Link to comment
Share on other sites

Posted (edited)

Ok, since AIDA did not trigger the same extreme temperature, then the issue is voltage loading during Cinebench. I don’t know if that is important to you or not, but look for a motherboard specific overclocking guide to help you find appropriate settings. 
 

It doesn’t particularly matter where you put the temp sensor in the loop. The pump/reservoir is an easy spot since the holes are there, but it could be slightly warmer or cooler depending on the loop order. Coolant temperature is a comparative value. It is the right value to use for pump and fan speed control but you have to learn your normal temperature range first. I suggest most users try various fixed pump speeds for a few weeks so they can make a determination of performance vs noise. What you want to avoid is running the pump too slow. There is a specific rpm point in every system where the fluid is too slow leaving the block and cpu temps jump dramatically. It’s unique to every loop, but more components, more radiators, and more vertical rise increase the need for higher pump speed. 

Edited by c-attack
Link to comment
Share on other sites

OK OK I will test 🙂

Link to comment
Share on other sites

Cinebench 23 is using AVX512 instructions that will really heat up that CPU.

Link to comment
Share on other sites

Someone use that witch Corsair ? it's better temp in cpu ?

 

 

Thermalright_LGA1700_2.jpg

Link to comment
Share on other sites

What are we looking at?  Is that a 3rd party backplate?  Don’t do that. The block tension is specific an extra millimeter here or there can cause real problems. Too tight can damage pins and that is a costly mistake. 
 

The AIDA vs R23 test shows you don’t have a contact problem. You need to look at the Vcore level when running R23 and compare that to when running 100% in AIDA, CPU-Z, etc. I suspect your motherboard is pouring on extra voltage in response to the AVX instructions to ensure stability. This why most newer boards from Z490 on have a separate AVX multiplier in the bios that will drop you down in frequency (and thus voltage) when AVX is detected. 

Link to comment
Share on other sites

Ok will check in asus forum, meybe they can help with bios 😀

Link to comment
Share on other sites

Look at what CUE reports as the Vcore when running both tests to confirm. I am expecting a very high value for R23. 
 

Asus is pretty easy to set this. In the AI Tweaker/Tuning column of the Advanced BIOS it will be right at the top under the normal frequency/multiplier. It’s likely set to auto, but -1 drops the frequency by 100MHz and thus it moves down the voltage table. 

Link to comment
Share on other sites

I have look in youtube OC from asus. And they use AI OVERCLOCKING. Work nice for me 🙂And it's not big difrend whe i have pump on 4000rmp or 2800.

But in bios we have cooler rank and i have 169 pts and the guy from youtube in AIO have 172pts.
 

Maybe i should change pipe for lover CPU temp.

 

AiOC Coling 169 nowy.jpg

Bios setings Cooler.jpg

Aio info bios.jpg

Link to comment
Share on other sites

You can ignore the cooler rating from the Asus BIOS.  My rating is 169 with dual 480mm external radiators and coolant that returns to the block at ambient temperature.  It can't possibly perform any better.  There is no way for the BIOS or any other program to assess the cooler performance without some very specific data not available to the BIOS.  All it can do is measure CPU temp peak and averages.  So if I drop my OC from 5.2x10 to 5.1 individual core clocking, I make my cooler score go up because I am applying less voltage to the CPU.  Ultimately it is the CPU temp that matters, but the CPU itself and voltage have a much larger say.  In fact, my score has been 169 with these settings no matter how many radiators and pumps are connected and in a wide array of designs and different CPU blocks.  It's the CPU that has not changed the last 3 years.  

 

I can't quite read all the small print on the screenshots, but if your R23 temps are have dropped, then I would say what you did was a benefit.  Generally speaking, you can't pop in a modern CPU and run a lot of synthetic tests without some tweaking.  MB makers are getting better about this, but the 'auto-overclock' to score higher on the benchmarks in reviews is still a thing.  

 

What did you mean about "changing the pipe"?  A different flow direction?  Like pump->GPU->Radiator->CPU->radiator->pump?  If you run a lot combination max GPU + CPU loads, like folding or special types of rendering, then it might save you 1-3C on the component that is 2nd in line (I think your CPU).  However, for things like running R23 it makes no difference.  The GPU is not working during it and adds no measurable heat.  Typically you pick up 1-3C as you pass through a block depending on wattage, flow rate, and few other things, but so at worst it would be a 3C penalty to the CPU when the GPU is maximally loaded.  However, for gaming and the like it won't matter.  The CPU is never maximally loaded and isn't going to downclock.  Your GPU on the other hand will start to downclock when it hits specific temp markers depending on the VBIOS.  Those can start as early on as 39-40C, so if your priority is gaming, then returning the coolest liquid to the GPU makes the most sense.  This is how I have set up my last several builds and also going from GPU <-> CPU directly usually makes for a more elegant tube run.  You did a nice job with your tube work.  I wouldn't be too eager to change it.  

 

Link to comment
Share on other sites

CB23 it's only benchmark 🙂 i only play games 🙂 in GPU after OC GPU Clock 2080MHz and mem clock 20500 i get 58* in Cpu not all core have between 70-75*, coolant 45*. So lets test more my setup. Thak you for help and a lot of info 🙂

Link to comment
Share on other sites

45° for the coolant seems pretty high. Not concerning, but just high. Maybe you should speed up those fans ^^'

What's the temperature in the room ?

Link to comment
Share on other sites

I'm running 3 x 360 RADs (2 x 30mm & 1 x 45mm). My idle coolant is under 28°C and goes up to around 38~39°C under very heavy load with an ambient of 21°C . Probe is inline between the 1st 360 rad (45mm) and the second with both the CPU and GPU (3090) plumbed in the loop, XMP1 (6200MT/s) and ASUS enhancements enabled (limits removed) in the BIOS. Case is a Lian Li O11D XL and fans (9) used on the radiators are Corsair ML-120 Elites.

My 12900k reports the following characteristics:

  • P-Core SP = 92
  • E-Core SP = 69
  • Overall SP = 84

The BIOS also reports:

  • Cooler = 172 pts

Der8auer did some testing (overclocking) with a number of 12900K's and found that the higher the CPU's SP, the less voltage he could get away with for a given frequency. With the voltage being lower on CPUs with a higher SP; it goes without saying that the recorded temperatures were also lower. Have a look at your CPU's SP values and see how it is compared to mine (higher is better).

 

Link to comment
Share on other sites

I have test now with open window 🙂 And it much better with temp. my fans warm up room very much. Usually i have 22/23* when i play games i have 25 or more 😞 so i have to keep open window And it's better when i take off my glass from top case. 

temp test.jpg

Link to comment
Share on other sites

I have:

P - core sp 97

E - core sp 75

Sp 89 

Cooler 164.

But I think sp for cooler take from idle temp. If vill be low will be high SP.

 

20220521_151427.jpg

20220521_151447.jpg

Link to comment
Share on other sites

Posted (edited)

Strictly in terms of R23 and other CPU benchmarks, the applied "actual" voltage (which will include some Vdroop) is going to determine your immediate CPU temperature.  Radiators and fan speed won't matter in those first 1-2 minutes.  So if the 5.4 auto voltage is 1.504, it's the 5.2 voltage setting on the curve that will be active when running a stress test.  It's hard to make predictions about what it will be as the more recent Asus boards are now apply more individualized curves.  It's likely to be between 1.36 and 1.42v and then on actual load may be down near 1.30v with LLC4 level Vdroop.  The exact value is critical and a few hundredths this way or that when you are close the limit will shift your CPU temp dramatically.

 

However, I am more interested in the 28 -> 45C change in coolant temp.  You should not be able to move your coolant temp by 17C with a 2x420 mm set up, except maybe if you leave it on those Hydro X presets for pump and fans.  That's like trying to put the brakes on a runaway train after you heat up the box.  Back at the first post, I thought you were indicating you were using both radiators (top and front) as exhaust and then the side wall fans are clearly intake.  The rear fan is exhaust.  Can you confirm that?  With that much coolant temp change it's more like when you dump the heat from one radiator into the other, eliminating the cooling effect up top, and effectively making it a 1x420 system.

Edited by c-attack
Link to comment
Share on other sites

Yes I have 2x 420mm radiator top and front - exhaust, side fans are intake and rear fan is exhaust. Yesterday I play and I get very high coolant temp. But I have closed windows and was very hot in room about 26* pc hitting up 😞 coolant he's over 40* gpu 60 cpu 80. Today I try open window and I se temp in room was around 23* so I play again and was surprised cpu max 70* gpu 55* coolant around 30 * but whe I take off my glass from topp was 3/4* lower temp 🙂 I think it's to hot in the room sometimes;)

Link to comment
Share on other sites

Posted (edited)

Also, since your rads are set as exhaust, make sure you removed the dust filters in those positions. they would hinder air flow and serve no purpose, filering dust on the way out

Edited by LeDoyen
Link to comment
Share on other sites

Ok I have 2 😀

20220521_183546.jpg

Link to comment
Share on other sites

that's a few more °C you'll gain then 😛 and they won't hinder lighting on top of that

  • Thanks 1
Link to comment
Share on other sites

A combination of increased resistance from dust filters or glass panels and lower fan speed could account for a 17C shift.  Coolant temperature is the baseline component temp and +-1C to coolant is also +-1C to GPU/CPU.  No need for the front dust filter at all, but then you can assess other changes.  

Link to comment
Share on other sites

Again i play 1h and it's very good temp for GPU 55* and CPU 70/75* coolant show in ique max 40* but was between 35-39. Radiator fans 850rmp, pump 3500rmp, another's fans 700rmp. 

Link to comment
Share on other sites

  • 2 weeks later...

Will test new build witch 140mm radiator:) should I clean, flush all pipe and radiator witch distilled wate? It's not old my loop 1mnd. I have clean gpu block. 

20220602_233827.jpg

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share

×
×
  • Create New...