Question How to fix high GPU temps ?

anonsky

Commendable
Nov 9, 2021
52
4
1,535
My 1650 (I think it's LPU, it doesn't have a 6 pin connector but the specs are the same as a normal 1650 according to GPU-Z) is three years old now and it has never once had normal temps. Under load it's always hitting 80 and if I'm unlucky it hits 83C (which I've heard that it' where nvidia cards starts lowering the boost clock). Three years ago it was somewhat understandable because the GPU was in an acer pre built and the case had no airflow and fans at all, I was able to circumvent the issue by taking off the side panel and literally hanging an external fan beside the GPU, but it has been 3 years since and I just recently built an entirely new system but reusing the GPU. I now have an actual case (not the best airflow, but hey my CPU only maxes out at 70c AT MOST so it's definitely something, granted it is undervolted). I'm really annoyed, it's such a low power card so I'm very surprised it's acting like this. Some solutions I've thought of are:
  1. Deshrouding the GPU
  2. Adding an aftermarket static pressure fan (something like the arctic p8) and connecting it to the motherboard so I can control the RPM
I've repasted the GPU as well and it did absolutely nothing even though the paste itself was kinda crusty.
Also HWINFO64 says my 1650 is Sapphire? I did not know they made nvidia cards.
 
Manufacturer? Can you upload a pic of it?

image.png
gpuz and task manager says nothing. it came with an acer pre built so i guess acer is the actual manufacturer? idk man
 
is three years old now and it has never once had normal temps. Under load it's always hitting 80 and if I'm unlucky it hits 83C (which I've heard that it' where nvidia cards starts lowering the boost clock).
80C on the core is high, yes, but not critical(that's 90C).
The boost curve increases/decreases in temperature intervals(don't know the exact value), the temperature limit isn't the only factor affecting boost.

I've repasted the GPU as well and it did absolutely nothing even though the paste itself was kinda crusty.
That likely didn't help. Use some fresher paste?

Is the gpu cooler the type with the horizontal(front to back) finstack, or the flower(all directions)? For reference:
Flower
i-bought-this-graphics-card-on-offerup-labeled-nvidia-v0-1pokb4gwdbe81.jpg

Horizontal(front to back)
Screenshot_153.png.09163fbbe8329c3ed8dfbaf422262214.png
 
I doubt anything is wrong.

Unless you have amazing cooling, and GTX 1650 don't usually rate expensive coolers, you should expect the GPU to be at its throttle point. That means it is giving you the best possible performance it can under the load you are asking of it.

Only when your cooling capacity is over specced will you see temperatures below the throttle point.

You could invest in a replacement cooler or do the fan mod you are talking about. That may help. But at a certain point you might just want to consider an upgrade.

$175 for a low profile RTX 3050 6GB, so supplemental power needed. And if you have room for a larger card. Something like the RX6600 or 6650XT ($190, and $230) are good options.

--I believe that was a reference to the stock thermal paste being crusty, which is normal. Long life thermal compounds are used from the factory and they often get quite hard over time.
 
Hey there,

Please list your full pc specs, including the model of the case. Then we can better advise.
CPU: Ryzen 5 5500
Motherboard: A520m k v2
Case: 1st player B5-M it has 3 intake molex and 1 exhaust molex
I doubt the rest matters in terms of temps? But just in case.
PSU: MSI MAG A650N
RAM: mixed 3200mhz ram unknown cas latency
 
I doubt anything is wrong.

Unless you have amazing cooling, and GTX 1650 don't usually rate expensive coolers, you should expect the GPU to be at its throttle point. That means it is giving you the best possible performance it can under the load you are asking of it.

Only when your cooling capacity is over specced will you see temperatures below the throttle point.

You could invest in a replacement cooler or do the fan mod you are talking about. That may help. But at a certain point you might just want to consider an upgrade.

$175 for a low profile RTX 3050 6GB, so supplemental power needed. And if you have room for a larger card. Something like the RX6600 or 6650XT ($190, and $230) are good options.

--I believe that was a reference to the stock thermal paste being crusty, which is normal. Long life thermal compounds are used from the factory and they often get quite hard over time.
Third world country lol. Prices here don't really reflect the west and vice versa. I did find a rx6600 for msrp tho and that's what I want to buy in the future, but I'd have to wait a while.

Also are you just saying it's normal for my GPU to reach those temps unless I have really good cooling? sound weird if you ask me.
 
Last edited:
Yes, it's common for GPUs to hit 80C or higher while playing games.

What did you expect the temperatures to be? And how did you come to that conclusion?
Probably disillusioned by YouTube benchmarks I guess. A lot of people there have safer temps floating around 70 and those were my expectations too.
 
Last edited:
I'm going to assume that 90C+ is a typo, 'cause the vbios of the GTX 1650 has a temperature limit of 90C: LINK

Look at the pictures again. The finstacks are different. Which one matches yours, the flower or the horizontal? The orientation of the fins has an impact on cooling.
The flower ones have always been lacking though; the 'just enough' solution.
The horizontal(front to back) types intake from below with half the exhaust exiting the front and the other half the rear, but the front exhaust conflicts with front case intake, so the design is sub-optimal in many cases.
 
Last edited by a moderator:
I'm going to assume that 90C+ is a typo, 'cause the vbios of the GTX 1650 has a temperature limit of 90C: LINK

Look at the pictures again. The finstacks are different. Which one matches yours, the flower or the horizontal? The orientation of the fins has an impact on cooling.
The flower ones have always been lacking though; the 'just enough' solution.
The horizontal(front to back) types intake from below with half the exhaust exiting the front and the other half the rear, but the front exhaust conflicts with front case intake, so the design is sub-optimal in many cases.
HWi-NFO64-Yv-U8g-Mlv-Dn.png

THIS WAS ON THE HOME SCREEN OF BALDURS GATE 3. HOME SCREEN!!!!

20240614-180910.jpg

Seems like the latter
 
Last edited:
If repasting and better airflow doesn't help then the problem is that you're hitting thermal capacity for the cooler itself. I had the same situation with my 1660 Ti (though the temps were a fair bit lower than you're seeing) where nothing I did would lower the temps beyond a certain point. On a card that low end chances are any type of cooler replacement you might be able to do wouldn't be worth the money.
 
If repasting and better airflow doesn't help then the problem is that you're hitting thermal capacity for the cooler itself. I had the same situation with my 1660 Ti (though the temps were a fair bit lower than you're seeing) where nothing I did would lower the temps beyond a certain point. On a card that low end chances are any type of cooler replacement you might be able to do wouldn't be worth the money.
But you're still implying that my COOLER is the weak point no? I don't think running the card with these temps is worth it either
 
My 1650 (I think it's LPU, it doesn't have a 6 pin connector but the specs are the same as a normal 1650 according to GPU-Z) is three years old now and it has never once had normal temps. Under load it's always hitting 80 and if I'm unlucky it hits 83C (which I've heard that it' where nvidia cards starts lowering the boost clock). Three years ago it was somewhat understandable because the GPU was in an acer pre built and the case had no airflow and fans at all, I was able to circumvent the issue by taking off the side panel and literally hanging an external fan beside the GPU, but it has been 3 years since and I just recently built an entirely new system but reusing the GPU. I now have an actual case (not the best airflow, but hey my CPU only maxes out at 70c AT MOST so it's definitely something, granted it is undervolted). I'm really annoyed, it's such a low power card so I'm very surprised it's acting like this. Some solutions I've thought of are:
  1. Deshrouding the GPU
  2. Adding an aftermarket static pressure fan (something like the arctic p8) and connecting it to the motherboard so I can control the RPM
I've repasted the GPU as well and it did absolutely nothing even though the paste itself was kinda crusty.
Also HWINFO64 says my 1650 is Sapphire? I did not know they made nvidia cards.
Use GPU z to get the model number of card

Pictures of the cooler and pcb would help and does the card have a backplate of metal or plastic do the memory have thermal pads

Higher quality paste and thermal pads should help.

You can also adjust the fan curve of the card Nvidia aims for a target of 85 before fans go full power.

If you use MSI afterburner you can manually adjust that .

View: https://youtu.be/ZShtbkA8e3M?feature=shared

https://www.msi.com/Landing/afterburner/graphics-cards

 
Last edited:
Use GPU z to get the model number of card

Pictures of the cooler and pcb would help and does the card have a backplate of metal or plastic do the memory have thermal pads

Higher quality paste and thermal pads should help.

You can also adjust the fan curve of the card Nvidia aims for a target of 85 before fans go full power.

If you use MSI afterburner you can manually adjust that .

View: https://youtu.be/ZShtbkA8e3M?feature=shared

https://www.msi.com/Landing/afterburner/graphics-cards

upon looking up it seems my card is

"Acer GTX 1650 OEM"​

not surprising

also the video you sent reminded me under volting my gpu was an option and it lowered my temps by like, 5c
 
94 is the hot spot not to terrible but a little toasty. Being a Acer I would probably put it on just bad airflow in the case.
i specifically mentioned i built an entirely new PC with significantly better airflow and it has proper fans
 
My 1650 (I think it's LPU, it doesn't have a 6 pin connector but the specs are the same as a normal 1650 according to GPU-Z) is three years old now and it has never once had normal temps. Under load it's always hitting 80 and if I'm unlucky it hits 83C (which I've heard that it' where nvidia cards starts lowering the boost clock). Three years ago it was somewhat understandable because the GPU was in an acer pre built and the case had no airflow and fans at all, I was able to circumvent the issue by taking off the side panel and literally hanging an external fan beside the GPU, but it has been 3 years since and I just recently built an entirely new system but reusing the GPU. I now have an actual case (not the best airflow, but hey my CPU only maxes out at 70c AT MOST so it's definitely something, granted it is undervolted). I'm really annoyed, it's such a low power card so I'm very surprised it's acting like this. Some solutions I've thought of are:
  1. Deshrouding the GPU
  2. Adding an aftermarket static pressure fan (something like the arctic p8) and connecting it to the motherboard so I can control the RPM
I've repasted the GPU as well and it did absolutely nothing even though the paste itself was kinda crusty.
Also HWINFO64 says my 1650 is Sapphire? I did not know they made nvidia cards.
upon looking up it seems my card is

"Acer GTX 1650 OEM"​

not surprising

also the video you sent reminded me under volting my gpu was an option and it lowered my temps by like, 5c


Strange normally gives a reference number still I would see if there's uneven pressure on the die see if your getting proper GPU coverage of paste.
 
Strange normally gives a reference number still I would see if there's uneven pressure on the die see if your getting proper GPU coverage of paste.
image.png

also i seriously doubt mounting pressure is the issue. for starters it's unlikely that it came out of the factory with uneven pressure, and when i repasted i made sure all the screws were tight enough, they literally could not move anymore
 
also i seriously doubt mounting pressure is the issue. for starters it's unlikely that it came out of the factory with uneven pressure, and when i repasted i made sure all the screws were tight enough, they literally could not move anymore
Most GPUs have spring loaded screws, the springs should not be fully compressed, they are there to provide the needed pressure. Check that the board isn't bending or anything. That could make the die and heatsink lose contact at the center.
 
  • Like
Reactions: beyondlogic
My 1650 (I think it's LPU, it doesn't have a 6 pin connector but the specs are the same as a normal 1650 according to GPU-Z) is three years old now and it has never once had normal temps. Under load it's always hitting 80 and if I'm unlucky it hits 83C (which I've heard that it' where nvidia cards starts lowering the boost clock). Three years ago it was somewhat understandable because the GPU was in an acer pre built and the case had no airflow and fans at all, I was able to circumvent the issue by taking off the side panel and literally hanging an external fan beside the GPU, but it has been 3 years since and I just recently built an entirely new system but reusing the GPU. I now have an actual case (not the best airflow, but hey my CPU only maxes out at 70c AT MOST so it's definitely something, granted it is undervolted). I'm really annoyed, it's such a low power card so I'm very surprised it's acting like this. Some solutions I've thought of are:
  1. Deshrouding the GPU
  2. Adding an aftermarket static pressure fan (something like the arctic p8) and connecting it to the motherboard so I can control the RPM
I've repasted the GPU as well and it did absolutely nothing even though the paste itself was kinda crusty.
Also HWINFO64 says my 1650 is Sapphire? I did not know they made nvidia cards.
upon looking up it seems my card is

"Acer GTX 1650 OEM"​

not surprising

also the video you sent reminded me under volting my gpu was an option and it lowered my temps by like, 5c


Strange normally gives a reference number still I would see if there's uneven pressure on the die see if your getting proper GPU coverage of paste.

Is the fan axial or blower ? Style
image.png

also i seriously doubt mounting pressure is the issue. for starters it's unlikely that it came out of the factory with uneven pressure, and when i repasted i made sure all the screws were tight enough, they literally could not move anymore

There where no thermal pads around it ?
 
Most GPUs have spring loaded screws, the springs should not be fully compressed, they are there to provide the needed pressure. Check that the board isn't bending or anything. That could make the die and heatsink lose contact at the center.
wait, then how does it work? do i just loosen the screws to alleviate some pressures? also, for now at least; it's not bent
 
Strange normally gives a reference number still I would see if there's uneven pressure on the die see if your getting proper GPU coverage of paste.

Is the fan axial or blower ? Style


There where no thermal pads around it ?
i have a picture above, i'm not sure which it is.
i think so? they were black as opposed to white/grey