Defiance 15" 3070 Model - High GPU temps

[djsubtronic]

So after a few weeks with the laptop, I've concluded the temperatures for the GPU are a little high for my liking. When being thrashed at 99% (reaching around the max 105W dynamic boost), it easily surpasses 85 C (and over 93 C on the "Hot Spot" temperature in GPU-Z). Left alone at these temps for more than 10-15 mins, the GPU then throttles down to 210 MHz and the only way to remove the throttle is to reboot, or disable/reenable the 3070 in device manager. Simply letting it cool off and return to a normal temperature does not stop the throttle.

Note these temperatures are reached even with max fan profile, the CPU was meanwhile peaking around around 98 C (in short bursts) but generally being around 90 C. I've even raised the back of the laptop to get extra airflow.

A repaste with Thermal Grizzly Kryonaut dropped the the CPU temperatures considerably, now it peaks around 85 C during those earlier 98 C bursts, and hovers in the high 70s and low 80s under load otherwise. But the GPU temperature did not see as noticeable a drop, still hitting mid 80s which is still a little high for my liking.

Is this just something I have to live with due to the nature of this laptop (i.e. tiny size + 105 W GPU). At the moment I have resorted to framecapping most AAA titles to 60 fps to artificially drop the GPU load (or reduce some of the graphics settings), but it would be nice to be able to play at higher framerates without the GPU frying itself and throttling.

Any ideas?

Edit: Just realised I posted this in the wrong section, can a mod pls move to the technical forum? Thanks.

 

[Steveyg]

Have you tried a cooling pad? I'm honestly pretty clueless when it comes to laptops but I've seen a few discussions about those

@Macco26 seems like a genius when it comes to laptops, hopefully he can offer some help

 

[Macco26]

I don't have much grasp on the Clevo's, but repasting was my first suggestion and it was done. Good.
Mind you try Throttlestop and see if the CPU can be undervolted (FIVR)? I don't know if Clevo does allow that.
I know your problem is the GPU, but since most of the heatpipes might be shared, the less overall thermals, the better.
Plus with MSI Afterburner you can try something like undervolting the GPU, but sorry I've never done that, so can't guide you further.

 

[barlew]

If the GPU is hovering around 85C I wouldn't be concerned that's acceptable.
I wouldn't want it any hotter than that though.

 

[djsubtronic]

Have you tried a cooling pad? I'm honestly pretty clueless when it comes to laptops but I've seen a few discussions about those

@Macco26 seems like a genius when it comes to laptops, hopefully he can offer some help

Not tried a cooling pad yet, but lifted up the back of the laptop, didn't make much difference over 1-2 C.

I don't have much grasp on the Clevo's, but repasting was my first suggestion and it was done. Good.
Mind you try Throttlestop and see if the CPU can be undervolted (FIVR)? I don't know if Clevo does allow that.
I know your problem is the GPU, but since most of the heatpipes might be shared, the less overall thermals, the better.
Plus with MSI Afterburner you can try something like undervolting the GPU, but sorry I've never done that, so can't guide you further.

Already undervolted CPU and cache by -100 mV, been running stable for over a week now under various loads. Had a bit of instability at 110 and 105 mW so I've settled on 100 for now. As I said, after the repaste my CPU's avg temperature dropped by nearly 10 C - however the GPU didn't really change much. Next step would be to start messing with the GPU voltage curve in Afterburner, but I was hoping to leave it on stock.

If the GPU is hovering around 85C I wouldn't be concerned that's acceptable.
I wouldn't want it any hotter than that though.

I personally would be fine with it hovering at 85 C if only it wouldn't throttle to the absolute minimum frequency soon after, without much temperature increase. That's the part which I find most frustrating, especially since the GPU is basically crippled at such a "low" temperature, and stays that way until I reboot or restart the device.

 

[barlew]

Not tried a cooling pad yet, but lifted up the back of the laptop, didn't make much difference over 1-2 C.


Already undervolted CPU and cache by -100 mV, been running stable for over a week now under various loads. Had a bit of instability at 110 and 105 mW so I've settled on 100 for now. As I said, after the repaste my CPU's avg temperature dropped by nearly 10 C - however the GPU didn't really change much. Next step would be to start messing with the GPU voltage curve in Afterburner, but I was hoping to leave it on stock.


I personally would be fine with it hovering at 85 C if only it wouldn't throttle to the absolute minimum frequency soon after, without much temperature increase. That's the part which I find most frustrating, especially since the GPU is basically crippled at such a "low" temperature, and stays that way until I reboot or restart the device.

Just so I fully understand this. Even when the GPU is sitting around 85C it is thermal throttling?

 

[Macco26]

Just a reference (it's not from Clevo but from Tongfang), 87 is the max allowed tempearture Tongfang allows before throttling:

you can lower it (down to 75 °C) but not above.

Strangest thing is Clevo's GPU reaches this limit, thottle to minimums and then stops there. Until Reboot? What the heck! This is starting to seem to me an RMA candidate.

 

[djsubtronic]

Just so I fully understand this. Even when the GPU is sitting around 85C it is thermal throttling?

Just a reference (it's not from Clevo but from Tongfang), 87 is the max allowed tempearture Tongfang allows before throttling:
View attachment 23759
you can lower it (down to 75 °C) but not above.

Strangest thing is Clevo's GPU reaches this limit, thottle to minimums and then stops there. Until Reboot? What the heck! This is starting to seem to me an RMA candidate.

I'm trying to get some accurate numbers and do a bit more testing on this, as the throttle issue has only happened about 3 times so far and all those times I had no monitoring active, just ingame temperature readings, and 85 was the last I saw (it could have gone to 87 or 88 and then quickly dropped back once throttled, before I got a chance to see the true temperature when the throttle started).

I'll post back here when I can reproduce another throttling event, but wanted to start the conversation in case anyone knew this to be normal behaviour for the new RTX cards or something.

 

[Macco26]

My Ionico has never reached the Max temperature so far, so I can't say what happens, but it's very weird if it doesn't recover from that.
I'd suggest you to launch a session of Furmark so to monitor very quickly what happens at those temperatures.

 

[djsubtronic]

Well this is weird asf it just throttled during not even intense use (first time under such conditions):

The core temp was 75 C and even the hot spot was only 79.2 peak.

 

[djsubtronic]

I ran Furmark for around 5 mins immediately after that last throttle...

GPU max temp reached 81 C, hotspot max 89 C as shown... didn't throttle. I'm hoping it's a software/driver/setting related issue.

 

[Macco26]

Very weird... Of course you are using latest Nvidia drivers, needless to say..

 

[djsubtronic]

Very weird... Of course you are using latest Nvidia drivers, needless to say..

Yeah I just updated them yesterday to 461.92.

 

[barlew]

Again I don't think those temps are anything to worry about.
It the only thing is concerning is the thermal throttle.
Have you tried completely uninstalling your Nvidia drivers with DDU and reinstalling them?

 

[djsubtronic]

Again I don't think those temps are anything to worry about.
It the only thing is concerning is the thermal throttle.
Have you tried completely uninstalling your Nvidia drivers with DDU and reinstalling them?

Done DDU etc a couple days ago.

Today I cold booted the laptop in the morning, didn't run any games or GPU related activities (my browsers are explicity set to use the iGPU) other than Discord (which is explicity set to use the Nvidia). Around an hour ago, I fired up Warzone and noticed that before I even entered the game, the GPU was in this max throttled state. So something during the day had randomly caused this to happen.

I disabled/reenabled the GPU and then played Warzone for an hour and it was completely fine.

Strange one this.

 

[barlew]

Done DDU etc a couple days ago.

Today I cold booted the laptop in the morning, didn't run any games or GPU related activities (my browsers are explicity set to use the iGPU) other than Discord (which is explicity set to use the Nvidia). Around an hour ago, I fired up Warzone and noticed that before I even entered the game, the GPU was in this max throttled state. So something during the day had randomly caused this to happen.

I disabled/reenabled the GPU and then played Warzone for an hour and it was completely fine.

Strange one this.

Have you carried out a clean install of Windows?

 

[djsubtronic]

Have you carried out a clean install of Windows?

Not yet, that would be a last resort when I have a full day on my hands and exhausted all troubleshooting other options.

 

[Martinr36]

Not yet, that would be a last resort when I have a full day on my hands and exhausted all troubleshooting other options.

No point in trying to troubleshoot if you aren't 100% certain that you have a good clean os install

 

[barlew]

If the DDU renstall hasn't worked your next step is a full clean reinstall of Windows.

 

[djsubtronic]

So, I think the problem was a missing ACPI Bridge driver that comes bundled with CCC. I installed this standalone, and so far the GPU has behaved OK for a couple of weeks.