Question 12900-k immediatly jumps to 100°C in Cinebench whilst being cooled in a custom waterloop (420mm Radiator + 360mm Radiator)

[Twily]

My 12900-K is instantly jumping to 100°C in Cinebench.

In some games Temps jumped to 80°C for a short time (just starting e.g.).
However I observed this now during summer time, when my Room is heating up quite a bit. On the other side I have a mobile air conditioning unit cooling my room when I deem it to be to hot.

This is my first PC build btw.
Have the system running since Let's say late February/ early March.

Another scenarios:
Playing Cyberpunk 2077 in a rather hot room. -> No problems. Atleast it stays under 80°C (else I get a warning popup).
Playing VR games -> Also no problems. Even on hot days with my room heating up.

But when I (just for fun) ran Cinebench 23.2 it immediatly jumped to 99°C (I guess that's highest amount the "Thermal Console" from Gigabyte goes). This time my room was rather cold.

Do I have to worry? Because I would expect with 2 rads and 10 Fans that even a 12900-K does not hit critical Temps in an instant.

I did not do anything yet to resolve that problem.

At one point I had a Kink in my Loop. Also my pump was set to run at linearly before I changed it. When I fixxed the kink, I also installed a flowmeter (spinning wheel thingy that shows water is moving).
I found out, that water was NOT moving. This was during winter, So I didn't had problems with CPU - temp.
When I fixxed the kink, I also made the pump run at 20% when under 20°C, at 80% if under 30°C, 90% if under 46°C and 100% if above.
The water is flowing.
Neither the CPU-block nor the GPU-block have Air bubbles. Can't see any in the tubes. Again system running for quite a while now.

So... do I have to worry? Can I run a 10min. Cinebench Test with my CPU at 99°C without causing harm to it?


My Specs:

• Case Corsair 7000D AIRFLOW
• Motherboard: Gigabyte Z690 AORUS Ultra
• 12900-K (obvsl.)
• MSI 3080 Ventus 3X 10G OC
• RAM: CORSAIR Vengeance DDR5 32GB (2x16GB) DDR5 4800 (PC5-38400) C40 1.1V - Black
• Pump: Hydro X Series XD5 (Corsair)

-CPU Block: Hydro X Series XC7 (Corsair) -> Which had preapplied thermalpaste (and yes I removed the plastic cover xd)

• One 420mm Radiator with 3x140mm Fans as Exhaust (Top) (Hydro X Seires XR5 420mm-Radiator)
• One 360mm Radiator with 3x 120mm Fans as Intake (Side) (Hydro X Seires XR5 360mm-Radiator)
• Another 4x140mm Fans (1 is Exhaust on the back and the other 3 are Intake Front)

I use softtubing and my Loop is as follows:
Pump -> GPU -> 420mm Rad with Exhaust Fans located on top (top mounted) -> CPU -> Side Radiator 360mm with 3x120mm Fans Intake -> Pump.

Pump is set to run at 100% as soon as temps hit 50°C

Pictures:
https://prnt.sc/fNwqATSjM2Xt

Screenshot

Captured with Lightshot

prnt.sc

https://prnt.sc/LJzmbQk8HhWi

Screenshot

Captured with Lightshot

prnt.sc

Video:
https://drive.google.com/file/d/1Uj8XynYVJvBMrFl08xNX_9nU0A78fff_/view?usp=sharing<

https://drive.google.com/file/d/1Ggi6SUxMEk_-O8APoLmlyBHD8ys_ltih/view?usp=sharing

 

[Deleted member 362816]

Are you overclocked to oblivion at a high voltage dear god with your setup you should never hit 80c. Even if your room was 90F those temps don't seem right on a custom loop like that.

 

[JWNoctis]

What's the powerdraw? HWiNFO should show that.

Weren't there some sort of bending problem and consequent poor thermal contact with desktop Alder Lake processors?

As to the heat: Modern processors - that is, mostly anything after the first few years of this century - has thermal protection, and they'll throttle down to nearly nothing, and then shut down the whole machine if that still wouldn't be enough, before they'd fry themselves. Not that they'd perform well or last long, if they are hot enough to be throttling.

I wouldn't do that to mine. Chances are it won't show its true, unthrottled, performance anyway.

 

[Karadjgne]

Reservoir?
Reservoirs are important. They have 2 functions. First is they vastly increase the volume of coolant in the loop, higher the volume, the easier it is to regulate temps. Second, they provide an air passage. Any micro bubbles will eventually pass into the reservoir, and get trapped, flow coming out from well below the coolant level.

What coolant?
Flow direction is fine, it literally does not matter how or where the components are placed or order. Some say it does, but they are wrong, it only counts when using the pc for a few minutes, after that the coolant acclimates and temps settle down.

Check 2 things, the input/output on cpu and gpu blocks and the pump. Because while direction and placement isn't really important, input and output IS, on the blocks/pump. If flow is going in the out, the coolant doesn't hit the jetplates, doesn't get forced through the microfins in the blocks, doesn't absorb much of any heat.
If the pump is pushing flow backwards, you'll get the same affect, even if the blocks are supposedly correct, but actually aren't.

The second thing to check is mounting pressure, if the cpu block isn't good-n-snug, you'll get higher temps.

With a full loop, case fans are pretty much irrelevant, the rad fans should be doing all the work moving air. Case fans are only really important to air cooled components, which you don't have.

What rads? What fans? This isn't an AIO, rads/fans make a huge difference. Some rads are purpose built for lower rpm, lower SP airflow, some are built for higher SP, higher rpm fans. Get the wrong rad combined with the wrong fans and you get next to useless power dissipation. My Hardware Labs/XSPC rads are low rpm, 750-1000 rpm gets best efficiency, any higher than that is basically just noise. There are other rads that are almost linear in efficiency, at 2300 rpm a 240mm has same wattage dissipating capacity as a 360mm rad from another company.

Fancy rgb fans that look pretty can definitely be an issue, severe lack of SP and cfm can make a 400w possible rad only good for 200w. Even at 1500rpm.

A good loop requires lots of planning and forethought, brand or price is meaningless, the model and purpose is everything.

You have just over 500w worth of heat output and a 420mm rad should be good for 400-450w, a 360mm rad ranges from 350-400w. So somewhere around potentially 750w-850w capacity. Supposedly. So something is wrong somewhere. I'd suggest you get an inline temp sensor, to see exactly what the coolant temps are, or at a minimum an IR temp reader to spot check pipe temps. All piping should be within 1-3°C anywhere in the loop.

At 100% on a D5, that flowmeter should be spinning so fast it will look like it's about to explode out of its case. Anything less and you have some severe flow restriction somewhere.

 

[Twily]

What's the powerdraw? HWiNFO should show that.

Weren't there some sort of bending problem and consequent poor thermal contact with desktop Alder Lake processors?

As to the heat: Modern processors - that is, mostly anything after the first few years of this century - has thermal protection, and they'll throttle down to nearly nothing, and then shut down the whole machine if that still wouldn't be enough, before they'd fry themselves. Not that they'd perform well or last long, if they are hot enough to be throttling.

I wouldn't do that to mine. Chances are it won't show its true, unthrottled, performance anyway.

Are you overclocked to oblivion at a high voltage dear god with your setup you should never hit 80c. Even if your room was 90F those temps don't seem right on a custom loop like that.

Added some pics to my post. As far as I checked, I should not be D:

 

[Twily]

What's the powerdraw? HWiNFO should show that.

Weren't there some sort of bending problem and consequent poor thermal contact with desktop Alder Lake processors?

As to the heat: Modern processors - that is, mostly anything after the first few years of this century - has thermal protection, and they'll throttle down to nearly nothing, and then shut down the whole machine if that still wouldn't be enough, before they'd fry themselves. Not that they'd perform well or last long, if they are hot enough to be throttling.

I wouldn't do that to mine. Chances are it won't show its true, unthrottled, performance anyway.

So, added a picture of HWinfo. I guess/hope I took a picture of the right things D:

As to the heat. I get a warning if the CPU gets over 80°C and as said, it did that only a few times while starting a game. Did so on VERY few occasions. Like in the span of it running maybe like 4 times.

When it comes to continuesly being over 80°C, it does so only by being benchmarked with Cinebench

 

[Twily]

Reservoir?
Reservoirs are important. They have 2 functions. First is they vastly increase the volume of coolant in the loop, higher the volume, the easier it is to regulate temps. Second, they provide an air passage. Any micro bubbles will eventually pass into the reservoir, and get trapped, flow coming out from well below the coolant level.

What coolant?
Flow direction is fine, it literally does not matter how or where the components are placed or order. Some say it does, but they are wrong, it only counts when using the pc for a few minutes, after that the coolant acclimates and temps settle down.

Check 2 things, the input/output on cpu and gpu blocks and the pump. Because while direction and placement isn't really important, input and output IS, on the blocks/pump. If flow is going in the out, the coolant doesn't hit the jetplates, doesn't get forced through the microfins in the blocks, doesn't absorb much of any heat.
If the pump is pushing flow backwards, you'll get the same affect, even if the blocks are supposedly correct, but actually aren't.

The second thing to check is mounting pressure, if the cpu block isn't good-n-snug, you'll get higher temps.

With a full loop, case fans are pretty much irrelevant, the rad fans should be doing all the work moving air. Case fans are only really important to air cooled components, which you don't have.

What rads? What fans? This isn't an AIO, rads/fans make a huge difference. Some rads are purpose built for lower rpm, lower SP airflow, some are built for higher SP, higher rpm fans. Get the wrong rad combined with the wrong fans and you get next to useless power dissipation. My Hardware Labs/XSPC rads are low rpm, 750-1000 rpm gets best efficiency, any higher than that is basically just noise. There are other rads that are almost linear in efficiency, at 2300 rpm a 240mm has same wattage dissipating capacity as a 360mm rad from another company.

Fancy rgb fans that look pretty can definitely be an issue, severe lack of SP and cfm can make a 400w possible rad only good for 200w. Even at 1500rpm.

A good loop requires lots of planning and forethought, brand or price is meaningless, the model and purpose is everything.

You have just over 500w worth of heat output and a 420mm rad should be good for 400-450w, a 360mm rad ranges from 350-400w. So somewhere around potentially 750w-850w capacity. Supposedly. So something is wrong somewhere. I'd suggest you get an inline temp sensor, to see exactly what the coolant temps are, or at a minimum an IR temp reader to spot check pipe temps. All piping should be within 1-3°C anywhere in the loop.

At 100% on a D5, that flowmeter should be spinning so fast it will look like it's about to explode out of its case. Anything less and you have some severe flow restriction somewhere.

Coolant is XL5 (Green) from the Hydro Series (Corsair)

Fans: https://www.bequiet.com/en/casefans/3452
https://www.bequiet.com/de/casefans/3453

The 120mm ones are mounted on the 360mm Radiator.
The other 140mm ones are mounted in the Front. Both as Intake.

The top Radiator also has 3x140mm Fans. Can't really find them. Those were the preinstalled Fans of the Corsait 7000D Case.

The one rear exhaust 140mm Fan is also from beQuiet. It's a "Silent Wings 3".

Checked for mounting pressure.
Pressed against it lightly. Then with more force. In both cases couldn't really feel it moving or anything.
I also tried screwing harder. These thumbscrews wouldn't move an inch. So I dunno. Is good?

https://drive.google.com/file/d/1Ggi6SUxMEk_-O8APoLmlyBHD8ys_ltih/view?usp=sharing

Wouldn't call that spinning so fast it's about to explode.

 

[maikutech]

-CPU Block: Hydro X Series XC7 (Corsair) -> Which had preapplied thermalpaste (and yes I removed the plastic cover xd)

I am going to take a guess and blame the pre-applied thermal grease. :/
When I went with water cooling by corsair, I always clean off their thermal grease.

After that I would apply thermal paste artic mx-4 or better artic mx-5.
If you do that, put 5 dots on the cpu like a cracker, then put your water block back on it turn it clockwise "genttly" before you lock it down, then let it settle for 15 minutes before you boot it up.


If you do that, way your cpu tempature should start off at 38c and stay stable even if you use ultimate settings in each one of your games.
Just for jokes if my core i5 9400F ever neared 95c it would power off immediatly and make a huge siren noise.

 

[TerryLaze]

Option 1: go into your bios and check all the settings that are AUTO and put them to what intel suggests or at least to something more sensible than the "everything she'll take" setting that is pre-applied. Especially Vcore and MCE should be tuned down but also Pl2 should be adjusted to the max you are comfortable with.

Option 2: Download IXTU and make a profile for everything that hits 100 degrees and reduce settings only for those.


Having Adaptive Boost Technology enabled will always boost as long as you are below 100 degrees meaning it will always reach 100 degrees unless you have ridiculously overpowered cooling.
https://www.intel.la/content/www/xl...intel-technologies-boost-cpu-performance.html

 

[Twily]

I am going to take a guess and blame the pre-applied thermal grease. :/
When I went with water cooling by corsair, I always clean off their thermal grease.

After that I would apply thermal paste artic mx-4 or better artic mx-5.
If you do that, put 5 dots on the cpu like a cracker, then put your water block back on it turn it clockwise "genttly" before you lock it down, then let it settle for 15 minutes before you boot it up.


If you do that, way your cpu tempature should start off at 38c and stay stable even if you use ultimate settings in each one of your games.
Just for jokes if my core i5 9400F ever neared 95c it would power off immediatly and make a huge siren noise.

Just for clarification.

It never happened DURING actually playing a game. As I said in 1 or 2 games, it'll jump to 90° and then normaly be around 60.

The big 100°C Problem ONLY occured on Cinebench.

In this case I would replace the Thermal paste as you suggested.
Although I would have to drain my loop. Is it save to use the computer, until I drain my loop for the annual loop draining and cleaning procedure, to THEN swap the thermal paste?

 

[Twily]

Option 1: go into your bios and check all the settings that are AUTO and put them to what intel suggests or at least to something more sensible than the "everything she'll take" setting that is pre-applied. Especially Vcore and MCE should be tuned down but also Pl2 should be adjusted to the max you are comfortable with.

Option 2: Download IXTU and make a profile for everything that hits 100 degrees and reduce settings only for those.


Having Adaptive Boost Technology enabled will always boost as long as you are below 100 degrees meaning it will always reach 100 degrees unless you have ridiculously overpowered cooling.
https://www.intel.la/content/www/xl...intel-technologies-boost-cpu-performance.html

Ok, I'll try that.

What can I be confortable with? Without having to worry to shorten the lifespan of my CPU?

Also, I feel like I should be able to atleast overclock a bit, but I have the feeling that if I should try, I would seriously get Problems regarding temperature.

 

[TerryLaze]

Ok, I'll try that.

What can I be confortable with? Without having to worry to shorten the lifespan of my CPU?

Also, I feel like I should be able to atleast overclock a bit, but I have the feeling that if I should try, I would seriously get Problems regarding temperature.

You probably already are heavily overclocked, that's the issue.
Just start with putting MCE and ABT to off and report back.
Also Vcore should be below 1.4 ,at most 1.5 is the safe limit but a lot of mobos have a difference between what they show and what thy actually apply.
Lowering the Vcore will have a big effect on temps but going too low can cause software to crash.

 

[Karadjgne]

BeQuiet lite wings. Cfm 41.5 at 1700rpm. That's really low amount of air moved for that rpm. At anything less than full speed, the cfm when faced with a restriction such as a rad will drop off drastically, not in a straight line. At 50% rpm, you'd be lucky to see 15-20cfm, which is abysmal for a rad. SP 1.66mmH²0. Ouch. That's seriously too low to be effective with a rad. By comparison my Noctua A12x25 120mm will move just over 60cfm at 2.36mmH²0, and the curve drop is far more linear, so at 50% I'm still around 30cfm and 1.1mmH²0. Basically double what your rgb fans can do.

The Corsair fans are the new SP140 Elite Air guide, 1200rpm, 68cfm, 1.66mmH²0. Old Noctua A14 at 1200rpm moves 68cfm at 1.51mmH²0 but has a 1500rpm limit, 82.5cfm at 2.08mmH²0.

So basically you'll always need to keep fans at max rpm, or just below or you'll get zip for airflow through the rads, which highly affects coolant temps and heat dissipation.

Corsair uses rebranded Mayhems coolants. Keep an eye on clarity, if it clouds up I'd stop using the Corsair coolant and switch to distilled water, biocide and green dye. (Primochill has some decent colorings)

Did you check input/output on the blocks/pump are correct for flow direction?

Did you fully flush out the rads correctly and completely before assembly? There's often leftover weld material in the rads that can mess up and restrict coolant flow through the blocks.

 

[Twily]

BeQuiet lite wings. Cfm 41.5 at 1700rpm. That's really low amount of air moved for that rpm. At anything less than full speed, the cfm when faced with a restriction such as a rad will drop off drastically, not in a straight line. At 50% rpm, you'd be lucky to see 15-20cfm, which is abysmal for a rad. SP 1.66mmH²0. Ouch. That's seriously too low to be effective with a rad. By comparison my Noctua A12x25 120mm will move just over 60cfm at 2.36mmH²0, and the curve drop is far more linear, so at 50% I'm still around 30cfm and 1.1mmH²0. Basically double what your rgb fans can do.

The Corsair fans are the new SP140 Elite Air guide, 1200rpm, 68cfm, 1.66mmH²0. Old Noctua A14 at 1200rpm moves 68cfm at 1.51mmH²0 but has a 1500rpm limit, 82.5cfm at 2.08mmH²0.

So basically you'll always need to keep fans at max rpm, or just below or you'll get zip for airflow through the rads, which highly affects coolant temps and heat dissipation.

Corsair uses rebranded Mayhems coolants. Keep an eye on clarity, if it clouds up I'd stop using the Corsair coolant and switch to distilled water, biocide and green dye. (Primochill has some decent colorings)

Did you check input/output on the blocks/pump are correct for flow direction?

Did you fully flush out the rads correctly and completely before assembly? There's often leftover weld material in the rads that can mess up and restrict coolant flow through the blocks.

Hey, no. I actually did not... D:

Yes, the input Output is correct.

So stronger Fans?

 

[Karadjgne]

I Really hate telling ppl to spend money, unless there's no other choice, every avenue has been covered etc.

Especially when it involves a somewhat major aesthetics change and the equipment is otherwise decent. You have decent fans, they just aren't quite optimized for rads.

Are your blocks clear enough to get a good view of the fins? Is it possible to see if there's any obstruction? A D5 should have plenty of flow for a 420+360 rad. Possible to get a good picture view of the loop?

 

[MavenAce]

Your CPU temps should never exceed 80 degrees celsius. I'd ensure that your fans are working properly and that the liquid inside of the cooling system is flowing correctly/reflush the rads. Make sure that the fans have proper flow outside of the case as well, and that thermal paste is properly applied, or reapply it.

 

[Mr.Spock]

I suspect it's how you've set the pump speed curve - what is it when <50'c? just for giggles I'd set it @100% period and retry Cinebench. I suspect the lag between pump ramp up and maybe badly set fan curves may be the cause.