i9-10850k cooling and some other things

[Alec19]

Hello to everyone that will read this.

 

I've got a lot of information here so please bear with me.

 

For background, here are my system specs:

CPU: i9-10850k

CPU cooler: MSI MAG Coreliquid 240R

RAM: G.Skill Ripjaws V 2x16 gb kit 3200 MHz

Motherboard: Gigabyte Aorus Master z490

GPU: AMD Radeon RX 5600xt

Storage: WD Black SN750 500gb, Samsung 860 Evo 1tb

PSU: Cougar VTE600 (600 watt)

Case: Thermaltake V200 Tempered glass RGB edition

 

CPU thermals:

When I first built this computer, I had a Hyper 212 Evo on my CPU. It was very evident that it was not capable of handling my 10850k so I quickly upgraded to an MSI 240mm aio. I was doing some thermal testing earlier and I found some interesting numbers. All of this is documented in the included pictures which are labeled based on the testing conditions. I tested my CPU temperatures under full load with the case panels on and off as well as with the GPU running under full load and not. For putting the CPU under load I was doing a test render on Autodesk Inventor and I used Heaven for loading the GPU. I also made sure to give ample time between each test for the liquid in the aio to get to room temperature (roughly 21 °C).

 

I have a couple of concerns about this data. The CPU appears to get above 90 °C within 10 minutes of running the render without the GPU on with the side panel on and off. The side panel made about a five degree change at the 15 minute mark of testing in all scenarios but temperatures still climbed to 100 °C after more time. When I put the GPU under load, I was able to hit 100 °C well before 10 minutes compared to roughly 20 minutes without the GPU under load. I was assuming that the GPU could be introducing more hot air into the system which decreases the amount of fresh air that the aio gets, but testing with an arduino and a thermistor suggest that this is not the only cause. I will go into more detail in the CPU power usage section, but it seems that there is something going on there.

 

AIO noise levels:

Another concern of mine is the noise of the AIO fans. MSI's website advertises 42 dB but as my data suggests, this is very wrong. Under full load, I recorded 75.30 dB with side panels on. At idle, 54.73 dB again with the side panels on. Granted, I am in a dorm room with a sound floor of roughly 35 dB so I expect that these numbers are not perfectly accurate, but that stills seems load to me. Especially since my computer sits less than 15 inches from my head. Does this seem too loud or is the manufacturer usually blatantly wrong with their numbers? I expect them to give best case numbers, but I am not far off of doubling the noise output under load. Seems wrong to me.

 

CPU power usage:

I found even more interesting number while checking the power output of my components. When under load with no GPU, the CPU was pulling between 215 and 235 watts of power. I was not able to replicate it for this, but I have seen the CPU pull 250 watts with the GPU under load as well. In the image "10 min no graphics side panels" you can see that the CPU hit a max load of 489.2 watts. I figure that it was a spike likely cause by me opening task manager a little bit after I started the test. Especially considering that I am pretty sure that an EPS connector cannot handle that much power. Either way, when I was running the GPU at the same time as I was testing the CPU I was noticing more power being delivered to the CPU. Is that normal and is it a problem?

 

I am going to be upgrading my GPU to a rtx 2080 soon so I was going to get a new PSU so that I don't have to daisy chain my PCIE power connector. My motherboard supports a second EPS connector so I was wondering if I should look for a PSU with another EPS cable. I know that it is not necessary unless doing extreme overclocking, but 230+ watts is a lot of power. Would it not be more stable with the extra EPS power? I have also attempted to overclock recently and it is never stable no matter how little of an overclock that I do. I need a lot of CPU power so I am trying to squeeze out what I can. I guess that my overclocking woes could be caused by my clearly not winning any silicone lottery given that my all core boost is 100 MHz short of spec.

 

Other cooling details:

I am in a very dusty area so leaving side panels off is not really a good idea. Also, for cooling, I have three Thermaltake RGB fans for intake running at a constant speed that I cannot measure (front of case), one Thermaltake non RGB fan acting as an output (back of case) operating between 400 and 1500 rpm depending on CPU thermals, and two MSI RGB fans on the aio acting as outputs operating between 650 and 2077 rpm again dependent on CPU thermals (top of case).

 

 

I know that there are a lot of questions in there. If I missed something, it may be in the pictures that I have included. Otherwise, I am happy to answer more questions. Any information helps.

 

Thanks for sticking though that,

Alec

[StarsMars]

5 hours ago, Alec19 said:

Hello to everyone that will read this.

 

I've got a lot of information here so please bear with me.

 

For background, here are my system specs:

CPU: i9-10850k

CPU cooler: MSI MAG Coreliquid 240R

RAM: G.Skill Ripjaws V 2x16 gb kit 3200 MHz

Motherboard: Gigabyte Aorus Master z490

GPU: AMD Radeon RX 5600xt

Storage: WD Black SN750 500gb, Samsung 860 Evo 1tb

PSU: Cougar VTE600 (600 watt)

Case: Thermaltake V200 Tempered glass RGB edition

 

CPU thermals:

When I first built this computer, I had a Hyper 212 Evo on my CPU. It was very evident that it was not capable of handling my 10850k so I quickly upgraded to an MSI 240mm aio. I was doing some thermal testing earlier and I found some interesting numbers. All of this is documented in the included pictures which are labeled based on the testing conditions. I tested my CPU temperatures under full load with the case panels on and off as well as with the GPU running under full load and not. For putting the CPU under load I was doing a test render on Autodesk Inventor and I used Heaven for loading the GPU. I also made sure to give ample time between each test for the liquid in the aio to get to room temperature (roughly 21 °C).

 

I have a couple of concerns about this data. The CPU appears to get above 90 °C within 10 minutes of running the render without the GPU on with the side panel on and off. The side panel made about a five degree change at the 15 minute mark of testing in all scenarios but temperatures still climbed to 100 °C after more time. When I put the GPU under load, I was able to hit 100 °C well before 10 minutes compared to roughly 20 minutes without the GPU under load. I was assuming that the GPU could be introducing more hot air into the system which decreases the amount of fresh air that the aio gets, but testing with an arduino and a thermistor suggest that this is not the only cause. I will go into more detail in the CPU power usage section, but it seems that there is something going on there.

 

AIO noise levels:

Another concern of mine is the noise of the AIO fans. MSI's website advertises 42 dB but as my data suggests, this is very wrong. Under full load, I recorded 75.30 dB with side panels on. At idle, 54.73 dB again with the side panels on. Granted, I am in a dorm room with a sound floor of roughly 35 dB so I expect that these numbers are not perfectly accurate, but that stills seems load to me. Especially since my computer sits less than 15 inches from my head. Does this seem too loud or is the manufacturer usually blatantly wrong with their numbers? I expect them to give best case numbers, but I am not far off of doubling the noise output under load. Seems wrong to me.

 

CPU power usage:

I found even more interesting number while checking the power output of my components. When under load with no GPU, the CPU was pulling between 215 and 235 watts of power. I was not able to replicate it for this, but I have seen the CPU pull 250 watts with the GPU under load as well. In the image "10 min no graphics side panels" you can see that the CPU hit a max load of 489.2 watts. I figure that it was a spike likely cause by me opening task manager a little bit after I started the test. Especially considering that I am pretty sure that an EPS connector cannot handle that much power. Either way, when I was running the GPU at the same time as I was testing the CPU I was noticing more power being delivered to the CPU. Is that normal and is it a problem?

 

I am going to be upgrading my GPU to a rtx 2080 soon so I was going to get a new PSU so that I don't have to daisy chain my PCIE power connector. My motherboard supports a second EPS connector so I was wondering if I should look for a PSU with another EPS cable. I know that it is not necessary unless doing extreme overclocking, but 230+ watts is a lot of power. Would it not be more stable with the extra EPS power? I have also attempted to overclock recently and it is never stable no matter how little of an overclock that I do. I need a lot of CPU power so I am trying to squeeze out what I can. I guess that my overclocking woes could be caused by my clearly not winning any silicone lottery given that my all core boost is 100 MHz short of spec.

 

Other cooling details:

I am in a very dusty area so leaving side panels off is not really a good idea. Also, for cooling, I have three Thermaltake RGB fans for intake running at a constant speed that I cannot measure (front of case), one Thermaltake non RGB fan acting as an output (back of case) operating between 400 and 1500 rpm depending on CPU thermals, and two MSI RGB fans on the aio acting as outputs operating between 650 and 2077 rpm again dependent on CPU thermals (top of case).

 

 

I know that there are a lot of questions in there. If I missed something, it may be in the pictures that I have included. Otherwise, I am happy to answer more questions. Any information helps.

 

Thanks for sticking though that,

Alec

 

Next time please list your questions first, then the data. 
 

As far as your cpu temps you have to issues. That 240 radiator is less than ideal for that cpu. Would be better off with a 280 or 360. 
 

Second, you have a closed off front panel. So  I’m not sure if the front tempered glass opens up or not. If so try it open. 
 

What you can do is either open the front. Switch the radiator fans from exhaust to intake. Or swap the cooler for a larger one. 
 

 

The noise readings don’t seem correct. Not sure how you are measuring that. The fans are probably ramping up because the case is restrictive and your temps are high. Same options as before. 
 

 

Your power readings I couldn’t really say. It might have something to do with the quality of the psu.  Not sure

[Alec19]

1 hour ago, StarsMars said:

Next time please list your questions first, then the data. 
 

As far as your cpu temps you have to issues. That 240 radiator is less than ideal for that cpu. Would be better off with a 280 or 360. 
 

Second, you have a closed off front panel. So  I’m not sure if the front tempered glass opens up or not. If so try it open. 
 

What you can do is either open the front. Switch the radiator fans from exhaust to intake. Or swap the cooler for a larger one. 
 

 

The noise readings don’t seem correct. Not sure how you are measuring that. The fans are probably ramping up because the case is restrictive and your temps are high. Same options as before. 
 

 

Your power readings I couldn’t really say. It might have something to do with the quality of the psu.  Not sure

Hey StarMars. Thank you for responding.

 

Sorry about the confusing order of my questions. I think I was trying to explain what I was seeing to give context to the questions but I can see how I may have made that weird to read.

 

The problem with larger radiators is that the top of my case has mounts for a 280 but the vrm heatsinks get in the way of anything bigger than a 240. The front panel appears to be riveted into place so removing it is not really ideal. If I was to switch the radiator fans to intake should I switch the front to outtake so that I have more than a single 120mm fan pulling air out. Also, would introducing all of that hot air into the case not make a lot of the other components overheat or at least make them hotter than necessary? The front has small holes on the left side and on the bottom so they do not really provide much air into the case. Kind of a waste of three fans if you ask me. I could potentially put a 360 on the front with some modification to the fan bracket but because of the front panel it would not have much access to fresh air and it would be restricted as an output. 

 

For the sound measurements I used my phone and then verified using a sound sensor and an arduino which got similar results. Even if those measurement are inaccurate, I can still confidently say that it is significantly louder than advertised.

 

I also forgot to mention something about the CPU temps that I found odd. When starting those tests on the CPU, the CPU temp almost immediately jumped from 30 °C to 70 °C. By almost immediately I mean within 15 seconds on average. Given the thermal mass of an aio I would expect that it would take longer to jump in temperatures like that. Is there a chance that I got a partially filled aio or that it is drying up somehow?

 

Thanks,

Alec

[StarsMars]

 

9 hours ago, Alec19 said:

I need a lot of CPU power so I am trying to squeeze out what I can.

What do you need the cpu power for? What is your use case? Are these workloads similar to what you will be doing?

Are you running an overclock? Is your motherboard bios running an auto overclock of some kind?

 

2 hours ago, Alec19 said:

If I was to switch the radiator fans to intake should I switch the front to outtake so that I have more than a single 120mm fan pulling air out. Also, would introducing all of that hot air into the case not make a lot of the other components overheat or at least make them hotter than necessary?

If anything, it's going to help. The problem right now is getting more air in.

Your case is restrictive and limiting what the AIO can work with. If you switch them around you are going to be giving the AIO direct access to fresh air. Non restrictive source = more air moving through the fins = cooler.

 

The other components like the VRMs and GPU? The VRM's are going to be fine with the heatsinks they have. The gpu is under control so I wouldn't worry.

2 hours ago, Alec19 said:

The front has small holes on the left side and on the bottom so they do not really provide much air into the case. Kind of a waste of three fans if you ask me.

It is a waste of the whole front fan section. They sacrificed function for aesthetics. That would be fine with a lower power cpu. 

 

 

 

 

So

If you have an overclock going. Maybe dial that back

Switch the AIO fans to intake.

You could try repasting the cpu.

 

If you have the money to throw at it.

Maybe look into getting a case with better airflow. If you need suggestions, I've got them.

If you swap the case and really want to throw money around. You could get a better AIO. Then sell the old case and AIO. Shouldn't come to that.

 

And for noise. You could change to Arctic p12 or some noctua nf a12. If you have a big ass cooler in a case with good airflow. It's not going to be running the fans like crazy which is where all the noise comes from.

 

You have a hot ass chip and not best case scenario for cooling.

If it were me, I'd get a case with great airflow and sell that thermaltake.

 

 

[Alec19]

10 hours ago, StarsMars said:

 

What do you need the cpu power for? What is your use case? Are these workloads similar to what you will be doing?

I do a lot of CAD, 3D modeling, simulations, some rendering (CPU based), and large Matlab calculations. Those applications really use a lot of CPU power and even if they needed more GPU I have found that there is a surprising amount of software that does not support AMD drivers. Hence my upgrade to a 2080 which is way more power than I need but it actually has support and I have a good connection for it.

 

Quote

Are you running an overclock? Is your motherboard bios running an auto overclock of some kind?

I have not overclocked my CPU at all. It's just that hot. I attempted to about a month ago but every attempt made the whole system jittery and slow. I am by no means an expert so that may be the cause but either way I could not get it to work. I have also checked the bios and there is no automatic overclock applied. I also tried to use the auto overclock feature earlier and it had the same results as my manual attempt.

 

Quote

If you have the money to throw at it.

Maybe look into getting a case with better airflow. If you need suggestions, I've got them.

If you swap the case and really want to throw money around. You could get a better AIO. Then sell the old case and AIO. Shouldn't come to that.

 

And for noise. You could change to Arctic p12 or some noctua nf a12. If you have a big ass cooler in a case with good airflow. It's not going to be running the fans like crazy which is where all the noise comes from.

 

You have a hot ass chip and not best case scenario for cooling.

If it were me, I'd get a case with great airflow and sell that thermaltake.

Sadly I do not really have money to throw at it right now. I am pretty much stuck with modifying what I have right now. That being said I might want to change cases somewhat soon. I travel across the country at least four times a year and having a a smaller case would make that process a lot easier. Given my full size atx board, an eventual 2080, and my thermal constraints, I am sure that small cases are limited. That being said, do you know of any good airflow small cases?

 

Quote

So

If you have an overclock going. Maybe dial that back

Switch the AIO fans to intake.

You could try repasting the cpu.

In the meantime, I will switch my aio fans to intake. I do not think that repasting my CPU will do much as I put the aio and new thermal paste on about a month and a half ago, but I will try it anyways. I will post my results when I get that done.

 

Thanks,

Alec

[StarsMars]

2 hours ago, Alec19 said:

That being said I might want to change cases somewhat soon. I travel across the country at least four times a year and having a a smaller case would make that process a lot easier. Given my full size atx board, an eventual 2080, and my thermal constraints, I am sure that small cases are limited. That being said, do you know of any good airflow small cases?

Fractal Meshify 2 Compact comes to mind

Not a ton of small cases for atx boards

 

Phanteks Eclipse P360A

Lian Li LANCOOL 215

Corsair 400D Airflow

Be Quiet! Pure Base 500DX

Phanteks Eclipse P400A

 

If you decide to look at other coolers. The Arctic Liquid Freezer ii line has a VRM fan and an thicker than average radiator. More mass for your $$.

Go 280 or 360 if you decide to.

 

Side note

Not sure what it would be

[Alec19]

Quote

Fractal Meshify 2 Compact comes to mind

Not a ton of small cases for atx boards

 

Phanteks Eclipse P360A

Lian Li LANCOOL 215

Corsair 400D Airflow

Be Quiet! Pure Base 500DX

Phanteks Eclipse P400A

Thanks for the case suggestions. That is not something that I can get to right now but I will eventually.

 

Quote

If you decide to look at other coolers. The Arctic Liquid Freezer ii line has a VRM fan and an thicker than average radiator. More mass for your $$.

Go 280 or 360 if you decide to.

I will also have to wait on a new cooler especially if I can get my current one working as it should. Thank you for the suggestions though.

 

Quote

Side note

Not sure what it would be

The only thing that I can guess about this warning is that it might be flagging the second EPS connector on the motherboard which the PSU does not have. Otherwise, the PSU is single rail. Not sure if that would cause errors, but maybe.

 

Other than that, here's the update.

 

I have switched the aio fans to intake and redid my tests with the exclusion of the tests without a side panel. There is too much dust where I am to make the omission of side panels an option.

 

Almost immediately I could tell that aio fans were quieter. They came it at roughly 70 dB under load which is at least an improvement although still well north of advertised. This benefit also impacted the no load sound of the fans although not as much as it did under load.

 

The CPU temperatures technically improved but not really. Even after 35 minutes I never hit 100 °C but I got to 99 °C so not really much of an improvement. That being said, its performed almost identically through the first 15 minutes. The test when the GPU was under load at the same time did increase the temperature of the GPU but it did not make any substantial change to the CPU temperature. That makes sense to me although the CPU temperatures still seem way higher than they should be.

 

Anymore ideas? The front intake fans run at one constant speed that really does not do much. The holes on the left are roughly 1 cm^2 and there are 81 of them. There is not enough air being pulled through to hold up a piece of paper cut to an eighth of the size. They are attached to a cheap case fan controller because the headers are not compatible with my motherboard. I have included a picture of the controller. Do you know if there is a compatible controller that allows me to control the speed?

 

(images of tests included and labeled accordingly)

[StarsMars]

1 hour ago, Alec19 said:

Anymore ideas?

Yes. I've been reading around trying to figure this out. I haven't dealt with the intel 10series.

 

I read through this thread.

 

The problem for him ended up being

@Mateyyy

"there's something clearly very wrong here. Your board's pushing 1.5V into the CPU.

 

Clear the CMOS and try running a stress test again. If nothing changes, update your BIOS."

 

 

ElementHD

"YOU HERO.

 

Updated bios (which resulted in it turning off auto boot management )

 

But after sorting that out, she is much much cooler and only drawing 1.31 V @ full load.

 

TEMPS hit 74c at full load on all Cores. Thank you everyone for your help."

[StarsMars]

Can you run a quick test again and show the voltages?

[Guest]

38 minutes ago, StarsMars said:

Yes. I've been reading around trying to figure this out. I haven't dealt with the intel 10series.

 

I read through this thread.

 

The problem for him ended up being

@ShrimpBrime

"there's something clearly very wrong here. Your board's pushing 1.5V into the CPU.

 

Clear the CMOS and try running a stress test again. If nothing changes, update your BIOS."

 

 

ElementHD

"YOU HERO.

 

Updated bios (which resulted in it turning off auto boot management )

 

But after sorting that out, she is much much cooler and only drawing 1.31 V @ full load.

 

TEMPS hit 74c at full load on all Cores. Thank you everyone for your help."

Actually @Mateyyy pointed out the 1.5v v-core. I actually had missed it in the screen shot. He's got a good eye!

[StarsMars]

25 minutes ago, ShrimpBrime said:

Actually @Mateyyy pointed out the 1.5v v-core. I actually had missed it in the screen shot. He's got a good eye!

You're right. So did this solve your issue?

[Guest]

10 minutes ago, StarsMars said:

You're right. So did this solve your issue?

I don't have this issue lol. Huh. We are mixed up somewhere. Thread reads bios update did solve the issue from the OP of that thread.

[StarsMars]

1 hour ago, ShrimpBrime said:

I don't have this issue lol. Huh. We are mixed up somewhere. Thread reads bios update did solve the issue from the OP of that thread.

0.0 My bad. I think I'm a little messed up from studying.

 

[Guest]

19 minutes ago, StarsMars said:

0.0 My bad. I think I'm a little messed up from studying.

 

It's ok, that thread is a good lead. Did a good job with it.  

[Imannudein]

Check to see if you have MCE enabled. I have the same CPU but on ROG Maximus board. with MCE on, it's pushing 1.32V at load and 1.28V with AVX. Something's wrong there as your vCore is high. 

[Alec19]

Turns out that I completely forgot to update my bios when I put the computer together. I was on version F3 and the current is F20b. Looking through the supported CPU list on Gigabyte's website I found that the i9-10850k was not supported until version F5.

 

After a quick update, thanks Q-flash, I ran the test again and the max temperature that I saw over 10.5 minutes was just over 90 °V. While this still seems hot to me, it is a massive improvement. The only thing to note is that my all core boost is now limited to 4.7 GHz instead of the 4.8 GHz that I could get earlier. That being said, if you compare the render iterations to the last test of similar length they were close, but there was still a performance hit. I should also note that the CPU only drew a maximum of 205 watts compared to the 220-250 watts that it was drawing before.

 

Any idea why I am getting a reduction in performance? Also, should I be getting lower temperatures than I am now? ElementHD said that they got down to 74 °C and I get that they have a 280mm radiator compared to my 240mm aio, but it still seems like it should be a little cooler. Am I wrong in that assumption?

 

Quote

with MCE on, it's pushing 1.32V at load and 1.28V with AVX. Something's wrong there as your vCore is high. 

I also used CPUz to check the voltage and it now tops out at 1.26 volts. Is that low? If so, could that be the cause of my lower boost clock?

 

Thanks,

Alec

[StarsMars]

2 hours ago, Alec19 said:

ElementHD said that they got down to 74 °C and I get that they have a 280mm radiator compared to my 240mm aio, but it still seems like it should be a little cooler. Am I wrong in that assumption?

Different tests. Different case. Different ambient temps. Different cooler. 

[Alec19]

15 hours ago, StarsMars said:

Different tests. Different case. Different ambient temps. Different cooler. 

Good point. Any clues about that reduced performance?

 

Thanks,

Alec

[StarsMars]

3 hours ago, Alec19 said:

Good point. Any clues about that reduced performance?

 

Thanks,

Alec

I’m guessing the bios update made the boosting algorithm a little more conservative. It sees you are near a thermal limit and dialed back a bit. For the better

[Alec19]

42 minutes ago, StarsMars said:

I’m guessing the bios update made the boosting algorithm a little more conservative. It sees you are near a thermal limit and dialed back a bit. For the better

I suppose it is. Too bad though.

 

Thanks,

Alec