• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Who is messing with my GPU temperature limit

rimix2

New Member
Joined
Sep 25, 2021
Messages
9 (0.01/day)
Hi all, I recently bought an old ASUS UX333N (i7 8565U, MX150) and I would like to use it for light gaming such as CIV6 or TPH.
I noticed that my GPU was getting throttled at 75°C (using Heaven benchmark), despite GPU-Z states a Current Temp Limit as 91°C. And this is just the first weird thing.

I read some forums and I downloaded GPU Tweak 3, set the limit to 85°C. I see in GPU-Z that the new limit is kept only for a few seconds, and then re-set at 91°C. So this is apparently not working.
And nevertheless, once I pass 75°C, then the clock is reduced at 139Mhz until the temp goes below 60°C.

Can someone explain what's happening?
Did I buy an haunted laptop?
 
Thermal throttling

,That is normal for notebooks, they lack capacity to keep anything cool. If you want to game you must keep the guts cool. So you're stuck.

Even cell phones thermal throttle

Its limited by asus bios parameters and sensors on the pcb/cpu/gpu, glue logic
 
Last edited:
Hi, thanks for the reply.
I know that termal throttling is normal, but I would like to be able to set the limits as I want.
I just run a test, you can see the images below.

I start with the limit at 85, but it gets throttled at 70 (the new limit is set at 60).
As soon as it goes under 60, then there is a new limit at 91 (but the throttling starts at 70 nonetheless).

I just would like to have the "control" of my laptop... that's it...
70°C is really too low.
Is it smoething related to BIOS as you say? I looked into the BIOS options and there is nothing related to this.

1.PNG


2.PNG


3.PNG
 
The bios is locked down and limits you.

You wont "control" unless if you learn how to write custom firmware that removes said limits, then you will just hit physical barriers due to thermistors/thermostats.
 
Do you think a BIOS downgrade would help? I read in this forum that some users were using GPU tweak to successfully increase the thermal limit.
I just would like to do the same thing.
 
You can't really limit temps unless you have really really good cooling in there! With enough load/insufficient cooling it will easily cross that threshold. So unless you're looking to limit the clocks these settings are mostly useless.
 
That units gpu is mounted as a bga on the same pcb as the cpu so the gpu bios is intergrated with the mainboard bios.

Good luck finding a solution

Do you think a BIOS downgrade would help? I read in this forum that some users were using GPU tweak to successfully increase the thermal limit.
I just would like to do the same thing.
You may not be able to "downgrade" the bios.

It is a crapshoot
 
Are you sure that it throttles due to temperature and not due to power constraints? While benchmarking, take a look at the "PerfCap Reason" graph in gpuz to figure that out. Because under same load, but at a higher temperature the gpu would draw more W.
 
Its why if i game on a mobile system it has to be a laptop.
 
You are saying that bios always wins right?
Perhaphs the other users had more "tolerant" BIOSes?
 
Are you sure that it throttles due to temperature and not due to power constraints? While benchmarking, take a look at the "PerfCap Reason" graph in gpuz to figure that out. Because under same load, but at a higher temperature the gpu would draw more W.
Yup and knowing how those notebooks sip power the brick wont be big enough.

You are saying that bios always wins right?
Perhaphs the other users had more "tolerant" BIOSes?
Im saying physical sensors to prevent you from damaging your equipment always wins.
 
Would GPU undervolt be a possibility?
I'm undervolting the CPU and it got far better benchmarks with the same temperatures
 
You wont know unless if you try
 
If you can keep the card ~63C or lower I believe you'll hold max freq. indefinitely. Much like my 1050Ti (Pascal architecture) once you hit a certain temperature the turbo bin steps down, and it will only step back up if power and temp limit is not reached. Given your power limit is locked and unchangeable, your only control over throttling is the temperature.

Definitely normal for that generation of laptop. Even brand new laptops act much in the same way, albeit with better boosting tech/algorithms these days.

The settings switching back from what you input to the defaults is a sign the firmware is locked.

If you're on Windows 10 and the MX150 uses PCI-E connection you can look in Power Settings > Link State Power Management > Change from "Max Power Savings" to "Off" and you may see better graphics performance if it doesn't immediately power limit throttle.
 
Anyhow, @eidairaman1 I tried ;) and I actually managed to have decent FPS and almost no throttling at all by undervolting / underclocking the MX150 at 810Mhz/712V.
This is keeping the temp around 62-65°C. And the clock is good enough for the light gaming I am aiming to.

@rethcirE you were right!
Throttling around 65°C, and firmware locked... this is what is happening.
I'm wondering what were thinking at ASUS :|
Luckily I only spent 150 for this.
 
Anyhow, @eidairaman1 I tried ;) and I actually managed to have decent FPS and almost no throttling at all by undervolting / underclocking the MX150 at 810Mhz/712V.
This is keeping the temp around 62-65°C. And the clock is good enough for the light gaming I am aiming to.

@rethcirE you were right!
Throttling around 65°C, and firmware locked... this is what is happening.
I'm wondering what were thinking at ASUS :|
Luckily I only spent 150 for this.
Most laptops and for that matter most desktop GPU are bios limited, even the max oc limits are hard set in bios.

Your on the right path ,undervolting.
 
Anyhow, @eidairaman1 I tried ;) and I actually managed to have decent FPS and almost no throttling at all by undervolting / underclocking the MX150 at 810Mhz/712V.
This is keeping the temp around 62-65°C. And the clock is good enough for the light gaming I am aiming to.

@rethcirE you were right!
Throttling around 65°C, and firmware locked... this is what is happening.
I'm wondering what were thinking at ASUS :|
Luckily I only spent 150 for this.

A bit more detail;
There are two kinds of throttling behaviour and the former is in fact just boost clocks getting max performance within the temp/voltage limits - this is when you see small jumps in frequency, 13-26-52 mhz at at a time. The second kind is hard throttle where you lose boost clock entirely and lose over a hundred mhz. That one you want to avoid with lower power targets.
 
That's right. I was just not expecting such low temps to start the "hard throttle". I mean, 65°C is way to low!
 
In Nvidia systems, there is a way to issue Pstates in gpus, so you can delist P0 if you issue P1(which is a lower state) and I think it was done using MSI Afterburner's shortcut command list. Better check it out and find which state the cooling can tolerate on a prolonged basis and can get you over with the profile shuffling.
 
In Nvidia systems, there is a way to issue Pstates in gpus, so you can delist P0 if you issue P1(which is a lower state) and I think it was done using MSI Afterburner's shortcut command list. Better check it out and find which state the cooling can tolerate on a prolonged basis and can get you over with the profile shuffling.
Wow, that was a great hint but I guess I would need more information about that. My goal would be to raise the throttling limit to 70-75°C.
What would be the benefit of modifying Pstates?
Lower temps (like unvervolt) or increase of the throttling threshold?
Thanks!
 
Wow, that was a great hint but I guess I would need more information about that. My goal would be to raise the throttling limit to 70-75°C.
What would be the benefit of modifying Pstates?
Lower temps (like unvervolt) or increase of the throttling threshold?
Thanks!
You need to find the guide on that. I suggest you search gpuboost profile states. GN or some other website made an article on that. Just type in the state you want and you may play around its voltage setting too, perhaps? I don't know, haven't tried it myself, but that is what I would do...

PS: this is the best I can do, https://www.overclock.net/threads/g...boost-fixed-clock-speed-undervolting.1267918/
 
Hey! Check to see if you have a "hotspot" sensor in GPU-z they can be quite a bit higher than your reported temps depending on a few factors, they can throttle your gpu the way your used to seeing (for good reason).
 
Wow, that was a great hint but I guess I would need more information about that. My goal would be to raise the throttling limit to 70-75°C.
What would be the benefit of modifying Pstates?
Lower temps (like unvervolt) or increase of the throttling threshold?
Thanks!
that is not possible since pascal. kepler refresh (770/780ti) introduced temp limits aside from power limits. maxwell started the down clocking as low as 35c though most cards were around ~65c before it was noticeable. it could be ignored by modifying the bios but since pascal bios modding is a no go w/nvidia locking them down.
TempComp_575px.png

as mentioned, you're best to go undervolting and maybe keep it a s cool as possible.

E: added image from here
 
The only real thing you can do is improve cooling (thermal paste, pads, cleaning fan and exhaust), and lower the power consumption (underclocking)

Anything that keeps you away from that power intake limit and heat limit, will help
 
Back
Top