Temperature weirdness - Let's try to fix it

[Varusal]

Hey folks,

 

[Update below]

 

I have a little problem with my current setup. It's just too hot.

 

 

First of all I will show you / list my Setup and airflow of the case.

• CPU: Ryzen 5 1600 (1st Gen)
• Mainboard: ASUS PRIME B350M-K
• GPU: MSI GTX 1080 Ti Armor OC
• RAM: CMK16GX4M2A2400C14
• SSD 1: CT500MX500SSD1
• SSD 2: CT256MX100SSD1
• M.2: ADATA SX8200PNP
• CPU Cooler: Alpenföhn Ben Nevis Advanced 130mm (1500 rpm)
• Case: Sharkoon TG5 (Front-Glass removed)
• Case Fans:
  • 5 x ARCTIC F12 120mm (1350 rpm)
  • 1 x be quiet! Pure Wings 2 120mm (1500 rpm)

 

Case Airflow / Fan setup

Spoiler

 

All case fans are running at 100%, same goes for the CPU Fan. They are all cleaned!

 

Airflow / Fan Setup (blue: intake || red: exhaust):

 

 

GPU Temperatures / Usage

Spoiler

 

My GPU (MSI GTX 1080 Ti Armor OC) is constantly at about 40°C-45°C while only being used about 2%-5%.

The GPU runs with factory settings so no additional OC is being used.

 

This is with 100% fan speed of the GPU.

 

One thing I know is that the Core Clock is basically stuck. Which could be the cause of the GPU temperatures.

 

As soon as the GPU reaches maybe 45%-50% usage the temperature already hits the 70°C mark even with 100% fan speed.

 

 

CPU Temperatures / Usage

Spoiler

 

I have a Ryzen 5 1600 (1st Gen not AF), no OC.

 

 

So in short: I want to fix that.

 

Would be great if we can find a solution / pointers to a possible solution.

 

Thanks

 

EDIT:

 

So some might think that those temperature are fine but as soon as I use 0db you will understand why it is not okay.

 

Spoiler

 

The GPU jumps to over 73°c in idle mode with 0% fan usage in like 5 minutes and still climbing. That is not how a GPU should heat up.

 

 

 

 

Edited June 4, 2021 by Varusal
Additional information

[Xaring]

Wym? 
54º on the hot spot is chill AF. Theres no issue with what you're showing lol.

High 70s are perfectly fine. GPUs boost themselves as hard as they can to squeeze performance.
Your GPU is fine.

[Tech87]

I agree.

You aren't having any thermal issues. Some people would love to have your temps!

[Pixelfie]

That's normal. I would start worrying if the temps are above 85c under load or it's thermal throttling. It probably uses just as much power on 50% usage as 100%, power = heat, so nothing wrong here. You can try undervolting using MSI afterburner and replace thermal paste if you really need lower temps. The GPU is 6 years old so the thermal paste is probably dried.

 

Edit: Just saw it's hot with 100% fan speed, follow the steps I said.

[Latvian Video]

6 minutes ago, Varusal said:

Hey folks,

 

I have a little problem with my current setup. It's just too hot.

 

 

First of all I will show you / list my Setup and airflow of the case.

• CPU: Ryzen 5 1600 (1st Gen)
• Mainboard: ASUS PRIME B350M-K
• GPU: MSI GTX 1080 Ti Armor OC
• RAM: CMK16GX4M2A2400C14
• SSD 1: CT500MX500SSD1
• SSD 2: CT256MX100SSD1
• M.2: ADATA SX8200PNP
• CPU Cooler: Alpenföhn Ben Nevis Advanced 130mm (1500 rpm)
• Case: Sharkoon TG5 (Front-Glass removed)
• Case Fans:
  • 5 x ARCTIC F12 120mm (1350 rpm)
  • 1 x be quiet! Pure Wings 2 120mm (1500 rpm)

 

Case Airflow / Fan setup

  Reveal hidden contents

 

All case fans are running at 100%, same goes for the CPU Fan. They are all cleaned!

 

Airflow / Fan Setup (blue: intake || red: exhaust):

 

 

GPU Temperatures / Usage

  Reveal hidden contents

 

My GPU (MSI GTX 1080 Ti Armor OC) is constantly at about 40°C-45°C while only being used about 2%-5%.

The GPU runs with factory settings so no additional OC is being used.

 

This is with 100% fan speed of the GPU.

 

One thing I know is that the Core Clock is basically stuck. Which could be the cause of the GPU temperatures.

 

As soon as the GPU reaches maybe 45%-50% usage the temperature already hits the 70°C mark even with 100% fan speed.

 

 

CPU Temperatures / Usage

  Reveal hidden contents

 

I have a Ryzen 5 1600 (1st Gen not AF), no OC.

 

 

So in short: I want to fix that.

 

Would be great if we can find a solution / pointers to a possible solution.

 

Thanks

The CPU temps seem fine to me, but for the GPU, maybe try replacing the thermal paste. 

[Varusal]

On 6/4/2021 at 11:56 PM, Xaring said:

Wym? 
54º on the hot spot is chill AF. Theres no issue with what you're showing lol.

High 70s are perfectly fine. GPUs boost themselves as hard as they can to squeeze performance.
Your GPU is fine.

It's not fine if the GPU fans blow hard in idle mode. And there is still the issue that the Core Clock is stuck and doesn't go down which increased power draw and with that the temerature.

 

On 6/5/2021 at 12:00 AM, Pixelfie said:

That's normal. I would start worrying if the temps are above 85c under load or it's thermal throttling. It probably uses just as much power on 50% usage as 100%, power = heat, so nothing wrong here. You can try undervolting using MSI afterburner and replace thermal paste if you really need lower temps. 

That is the other problem it goes higher than 85°C under load.

 

On 6/5/2021 at 12:00 AM, Latvian Video said:

The CPU temps seem fine to me, but for the GPU, maybe try replacing the thermal paste. 

Any recommendations for a good thermal paste?

[Pixelfie]

9 minutes ago, Varusal said:

Any recommendations for a good thermal paste?

Arctic MX5 or Thermal Grizzly Kryonout should be good.

[Varusal]

6 minutes ago, Pixelfie said:

Arctic MX5 or Thermal Grizzly Kryonout should be good.

Thanks that is something I will try out.

 

Updated the Post:

 

27 minutes ago, Varusal said:

So some might think that those temperature are fine but as soon as I use 0db you will understand why it is not okay.

 

  Hide contents

 

The GPU jumps to over 73°c in idle mode with 0% fan usage in like 5 minutes and still climbing. That is not how a GPU should heat up.

 

[Xaring]

Do a clean driver install - use DDU and see if it fixes anything.

Also, check what is using the GPU - the 0db mode should still ramp up the fans after, lets say, 50 or 60ºC. If you tell the fans to never turn on, then of course its gonna heat up XD.

I can also recommend Noctua NT-H2 as thermal paste on top of the others mentioned-

[Varusal]

9 minutes ago, Xaring said:

If you tell the fans to never turn on, then of course its gonna heat up XD.

Yeah I would agree if the GPU was not in IDLE. It get's constantly hotter and hotter despite not being used (like literally reached 90°C in idle). Sorry but that does not make sense at all. Something that isn't in use should not get increasingly hotter.

 

Even if you look here they have a constant range for example idle at 0% fan speed, temps are between 30°C and 40°C.

 

[Varusal]

10 hours ago, Xaring said:

If you tell the fans to never turn on, then of course its gonna heat up XD.

To proof you wrong I did download Nvidia Inspector and activated the multi-display power saver.

As you can clearly see the GPU Clock and Memory Clock are not stuck anymore which means it runs cooler. And to top things off, 0% fan speed and higher GPU usage (16%) and it is still under 37°C.

 

[Xaring]

Dude, you need to show us those "HOT" numbers, we keep seeing high 30s low 40s on these screenshots... 
16% is basically "no load" - I'd expect my GPU in the high 50s if fan-less.
Im not saying you're lying or that what you're claiming is wrong, but bring more proof and we might be able to help - with the info you've provided... It just seems like you're overly cautious or you might not be looking at the right numbers or whatever.

[LogicalDrm]

Could you post your GPU fan profile? If I'm understanding this correctly, your issue is with how high temps go with certain specific fans speeds? With 0rpm meaning something like 70C? And 100% speed meaning under 40C?

[Varusal]

On 6/6/2021 at 5:59 PM, Xaring said:

Dude, you need to show us those "HOT" numbers, we keep seeing high 30s low 40s on these screenshots... 
16% is basically "no load" - I'd expect my GPU in the high 50s if fan-less.
Im not saying you're lying or that what you're claiming is wrong, but bring more proof and we might be able to help - with the info you've provided... It just seems like you're overly cautious or you might not be looking at the right numbers or whatever.

Have you looked at the first post. I updated it. Not that hard to see that the GPU temp reaches 70°C+ while doing nothing.
I currently use the Nvidia inspector and I replied to you with the same 0% fan usage. GPU stays under 40°C+ easily now. Not sure what you want from me but here to clarify again:
 

Without Nvidia inspector (Multi-Monitor Power Saver):

• No Overclock
• GPU Fan Speed -> 0%
• GPU utilization -> under 5%
• GPU Core Clock (stuck): 1531MHz
• GPU Memory Clock (stuck): 1377MHz
• GPU Temperature after 5 mins: 74°C
• GPU Hot Spot Temperature after 5 mins: 88°C

Screenshot:

Spoiler

 

 

With Nvidia Inspector (Multi-Monitor Power Saver):

• No Overclock
• GPU Fan Speed -> 0%
• GPU utilization -> under 20%
• GPU Core Clock (stuck): 607MHz
• GPU Memory Clock (stuck): 101MHz
• GPU Temperature after 5 mins: 36°C
• GPU Hot Spot Temperature after 5 mins: 50°C

Screenshot:

Spoiler

 

I don't know what you want from me. Both test have the same environment and same setting but one does not have high Core / Memory Clocks that are stuck. Which is a bug and can be fixed by using Nvidia Inspector (Multi-Monitor Power Saver). I really don't understand what you want from me. There is literally proof (Screenshots) for what I did say.

[Varusal]

On 6/6/2021 at 9:44 PM, LogicalDrm said:

Could you post your GPU fan profile? If I'm understanding this correctly, your issue is with how high temps go with certain specific fans speeds? With 0rpm meaning something like 70C? And 100% speed meaning under 40C?

Kinda, that was actually only a test to see if there is something fishy going on and yes there is.

The Core and Memory Clocks get stuck which forces the GPU to draw more power which also means it gets hotter that also includes idle loads like under 5% for example.

With the Nvidia Inspector (Multi-Monitor Power Saver) the Core and Memory Clocks are lower and don't get stuck anymore which means less power usage and less heat.

 

I did the same exact 0% RPM test without the Nvidia Inspector and with the Nvidia Inspector. Just postet this reply to @Xaring

 

So basically my problem is solved if I use the Nvidia Inspector. Which is a temporary thing. So far I have not found a solution to fix the Clock issue.

[Jeppes]

Some cards just are garbage when it comes to multimonitor idle. Try to use one high-refresh rate monitor and set rest to max 60.

[freeagent]

I haven't read the whole thread, but you mentioned you updated the first post so..

 

First thing I would do is flip your top fans and use them as intake. You aren't moving a ton of air, so its fine.. My logic is since you cant get the warm air out of the case fast enough, might as well bring as much cool in as you can. Any air your fan cant move will make its way out on its own just fine. You can try repasting too. You can buy stronger fans.. but not much you can do right now with what you have other than what I mentioned trying. 

[Varusal]

13 minutes ago, Jeppes said:

Some cards just are garbage when it comes to multimonitor idle. Try to use one high-refresh rate monitor and set rest to max 60.

That is exactly how I have it (3 Monitors).

If I only have two, I need to set the High-Refresh one to max. 120Hz. If it is 144Hz the clocks get stuck again.

[Jeppes]

1 minute ago, Varusal said:

That is exactly how I have it (3 Monitors).

If I only have two, I need to set the High-Refresh one to max. 120Hz. If it is 144Hz the clocks get stuck again.

And if all three are 60Hz? 1x 100/75HZ + 2x 60Hz? At least on my system win10 will remember the refresh rates separately from extended and main-monitor only when using win+p. So you can quickly swap from high-refresh rate gaming to extended desktop with low idle power. Not ideal but a workaround.

[Xaring]

At this point i'd try reflashing your VBIOS,maybe there is a flaw with yours 
https://www.techpowerup.com/vgabios/191228/msi-gtx1080ti-11264-170317

 

[Jeppes]

4 hours ago, Xaring said:

At this point i'd try reflashing your VBIOS,maybe there is a flaw with yours 
https://www.techpowerup.com/vgabios/191228/msi-gtx1080ti-11264-170317

 

Its a feature they never fixed and some AMD cards have it too. I guess too few people using more than two monitors to make them care.

[Varusal]

21 hours ago, Jeppes said:

And if all three are 60Hz? 1x 100/75HZ + 2x 60Hz? At least on my system win10 will remember the refresh rates separately from extended and main-monitor only when using win+p. So you can quickly swap from high-refresh rate gaming to extended desktop with low idle power. Not ideal but a workaround.

I tried that too but that also did not work.

[Xaring]

i've had 3 monitors, 1080p, 144hz and 2x 60hz for literal years now. No issues like you're mentioning what so ever.
On a GTX 970 first and recently on my RX 6800. 

I believe theres a chance that theres some BIOS shenanigans going on

[Varusal]

2 hours ago, Xaring said:

i've had 3 monitors, 1080p, 144hz and 2x 60hz for literal years now. No issues like you're mentioning what so ever.
On a GTX 970 first and recently on my RX 6800. 

I believe theres a chance that theres some BIOS shenanigans going on

It's a common Nvidia bug.